article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
it has been proposed that the damped ly@xmath0 ( dla ) systems seen as broad absorption lines with neutral hydrogen column density @xmath1 2 10@xmath2 @xmath3 in quasar spectra are large and massive ( milky way sized ) , galactic disks with maximum circular speed @xmath4 200 km / s ( e.g. , prochaska & wolfe 1997 , 1998 ) . if true , this is of profound interest in relation to theories of the formation and evolution of disk galaxies , but it is obviously important to test this hypothesis in as many ways as possible . in this contribution
i present a test alternative to the ones used by prochasca & wolfe .
my test is based on fairly elaborate and realistic models of the formation and evolution ( including chemical evolution ) of large and massive disk galaxies , in particular on comparing the observed distribution of zinc abundance with neutral hydrogen column density found for dlas with model predictions . in section 2
i briefly describe the models , in section 3 the data , in section 4 the model predictions are confronted with the data and finally section 5 constitutes a brief conclusion .
the star - formation rate is assumed to depend on gas surface density as @xmath5 where n=1 - 2 , and the weight factor @xmath6 for models without a star - formation threshold and @xmath7 for models with a star - formation threshold given by ( e.g. , kennicutt 1989 ) @xmath8 where @xmath9 is the epicyclic frequency @xmath10 @xmath11 is the velocity dispersion of the gas , @xmath0 is a dimensionless constant of the order unity and @xmath12 is the epicyclic frequency at the solar distance from the center of the galaxy .
i have adopted @xmath0 = 0.7 and @xmath11 = 6 km / s ( kennicutt 1989 ) .
infall of primordial gas , but no gas out - flows are included in these models of large and massive disk galaxies .
chemical evolution is calculated assuming a scalo ( 1986 ) initial mass function ( imf ) .
oxygen is assumed to be produced in massive stars only ( @xmath13 8 m@xmath14 ) and instantaneously recycled by type ii supernova ( sn ) explosions .
iron is assumed to be produced partly by massive stars ( @xmath13 8 m@xmath14 ) and instantaneously recycled by type ii sn explosions and partly by less massive stars and recycled by type ia sn explosions with a time - delay of @xmath15 = 1 gyr .
unprocessed gas , locked up in stars , is assumed to be non - instantaneously recycled .
the proportionality constants in the star - formation laws ( eq . [ 1 ] ) are determined by applying eq . [ 1 ] to the local , solar cylinder and fitting data on the local gas fraction , g - dwarf distribution etc .
the global properties of the large disks as a function of time ( or redshift ) are determined by assuming that an average , spin - parameter @xmath16 = 0.05 , galactic disk of either early or late type ( see below ) has an exponentially decreasing stellar profile with radius @xmath17 in the disk ( e.g. , freeman 1970 ) truncated at 4 stellar scale - lengths ( e.g. , van der kruit 1987 ) and an exponentially decreasing oxygen gas abundance profile with radius in the disk ( e.g. , zaritsky et al .
1994 , garnett et al .
early type disk galaxies ( sab / sbc ) are assumed to have an average star - formation history parameter @xmath18 , defined as the ratio of the current to the past average star - formation rate , of @xmath19 0.33 and late type disks ( scd / sdm ) to have an average @xmath19 1.0 ( kennicutt et al .
an ellis ( 1983 ) present day ( @xmath20=0 ) morphological mix of e / s0:sab / sbc : scd / sdm = 0.28:0.47:0.25 is assumed and the models are averaged over spin - parameter @xmath16 .
the models thereby cover the full range of morphology and global surface brightness of large , galactic disks and are fully constrained : there are no free parameters left , except for the star - formation law index @xmath21 and the absence or presence of a star - formation threshold and the conclusions reached in this work turn out to be quite insensitive to this for reasonable values of @xmath21 ( @xmath21=1 - 2 ) - see also below .
observational data on 34 dla systems ( including zinc abundances or limits ) have been taken from pettini et al .
( 1994 , 1997 ) .
zinc is used because it is not or only very slightly affected by dust depletion .
figure 1 shows @xmath22 $ ] versus neutral hydrogen gas column densities @xmath23 for the 18 dlas with zinc detected .
they have a mean redshift of @xmath24=1.91 .
interestingly , higher zinc abundance corresponds to _ lower _
@xmath23 , which is _ not _ to be expected if the dlas are large , galactic disks , even when effects of dust obscuration of quasars are taken into account - see below .
figure 2 shows @xmath22 $ ] versus @xmath23 for the 16 dlas with upper limits on the zinc abundance .
the mean redshift of these is @xmath24=2.56 .
figure 3 shows the model predicted @xmath25 $ ] distributions for @xmath23 = 2 - 10 10@xmath2 @xmath3 and @xmath23 = 10 - 50 10@xmath2 @xmath3 for the no star - formation threshold models with star - formation law index @xmath21=1.5 .
the distributions have been weighted over morphological mix , spin - parameter ( and hence effectively global surface brightness ) and projected area and corrected for the effects of dust obscuration of quasars using the clever methodology developed by fall & pei ( 1993 ) .
figure 4 shows the same thing , but for the models _ with _ a star - formation threshold ( and still @xmath21=1.5 ) .
neither of the two sets of model distributions appear to match the data at all ( assuming that @xmath22 \simeq [ fe / h]$ ] as is the case for galactic stars of all abundances ) .
more quantitatively , komolgorov - smirnov ( ks ) tests show that both sets of model distributions can be ruled out as fits to the data with more than 99% confidence ( this result is more generally obtained for all values @xmath21=1 - 2 and also for wyse & silk ( 1989 ) type star - formation laws ) .
the mismatch between the data and the models is further illustrated by figure 5 .
the figure shows the number of predicted dla systems per redshift interval with zinc detected , @xmath26 , for @xmath23 = 2 - 10 10@xmath2 @xmath3 and @xmath27 10@xmath28 @xmath3 versus redshift @xmath20 .
the models have been normalized using observational information about the luminosity function and sizes of present day ( @xmath20=0 ) , large and massive disk galaxies ( with @xmath4 200 km / s ) .
the model predictions have been weighted and corrected as described above and furthermore corrected for the effect of a zinc detection threshold as a function of @xmath23 .
figure 5 is for the models with star - formation law index @xmath21=1.5 and no star - formation threshold .
also shown are apm survey based observational results of storrie - lombardi et al .
( 1996 ) corrected by an observationally deduced zinc detection probability as a function of redshift @xmath20 and @xmath23 derived from the pettini et al .
( 1994 , 1997 ) data .
as can be seen from the figures the two sets of model predictions lie more than two orders of magnitude below the observations - the situation is even worse for the models _ with _ a star - formation threshold ( not shown ) .
the analysis presented above suggests that the dla systems may not in general be accounted for by large and massive disk galaxies with @xmath4 200 km / s , not even when account is taken for the existence of some very large and low surface brightness disk galaxies ( physically : high @xmath16 galactic disks ) .
the true situation is most likely quite complicated , but it is tempting to associate the dlas at low redshift with the large population of faint , blue dwarfs and at higher redshift with the large number of smaller galaxies expected at such redshifts ( @xmath29 2 - 4 ) in hierarchical galaxy formation scenarios ( e.g. , kauffmann 1996 , mo et al .
dlas with indications of large internal velocities ( @xmath30 200 km / s ) may , at least partly , be dynamically unrelaxed systems ( e.g. , haehnelt et al .
i have benefited from discussions with mike edmunds , don garnett , bernard jones , bernard pagel and max pettini . | the hypothesis that the damped ly@xmath0 systems ( dlas ) are large , galactic disks ( milky way sized ) is tested by confronting predictions of models of the formation and evolution of ( large ) disk galaxies with observations , in particular the zinc abundance distribution with neutral hydrogen column density found for dlas .
a pronounced mismatch is found strongly hinting that the majority of dlas may not be large , galactic disks . |
a contour plot of the in - plane resistivity @xmath26 of cecoin@xmath0 in the @xmath126 plane .
three different regions are found in the contour plot , which are distinguished by different values of the exponent @xmath27 of the power - law @xmath3 dependence of the resistivity ; i.e. , @xmath127 .
the @xmath128-axis of this plot is along the @xmath74axis of cecoin@xmath0 and @xmath4 is the angle between the diagonal of the rectangle in region i and the @xmath128-axis ( @xmath74axis ) . ,
scaledwidth=50.0% ] figure 2(a ) scaling of resistivity @xmath26 vs the component of the applied magnetic field @xmath2 perpendicular to the @xmath16-planes , hcos@xmath24 , measured at a temperature @xmath3 of 6 k and for 1 t @xmath129 5 t. the arrows mark the points at which the curves start to deviate from each other .
insets : @xmath26 vs the angle @xmath24 measured at in an applied mahnetic field of 2 t and 9 t. 2(b ) resistivity @xmath26 vs the component of the applied magnetic field @xmath2 parallel to the @xmath16-planes , hsin@xmath24 , measured at a temperature @xmath3 of 6 k in scanning @xmath2 at fixed orientation @xmath130 . the @xmath26 vs @xmath41 data scale for @xmath131 . ,
scaledwidth=50.0% ] figure 3 . in region
i , scaling of resistivities @xmath47 and @xmath132 when plotted vs the component of the magnetic field perpendicular to the @xmath16-planes , @xmath48 , and parallel to the @xmath16-planes , @xmath55 , times the inverse anisotropy @xmath133 , respectively .
inset : plots of @xmath47 and @xmath54 curves measured in magnetic fields up to 14 t. , scaledwidth=50.0% ] figure 5 .
angular ( @xmath24 ) dependence of the resistivity @xmath26 measured at 6 k in a magnetic field h = 2 t ( open symbols ) .
the solid line is the fit of eq .
( 6 ) to the data .
the dashed lines are the @xmath38 and @xmath41 scalings laws .
, scaledwidth=50.0% ] | in - plane angular dependent resistivity adr was measured in the non - fermi liquid regime of cecoin@xmath0 single crystals at temperatures @xmath1 k and in magnetic fields @xmath2 up to 14 t. two scaling behaviors were identified in the low - field region where resistivity shows a linear @xmath3 dependence , separated by a critical angle @xmath4 which is determined by the anisotropy of cecoin@xmath0 ; i.e. , adr depends only on the perpendicular ( parallel ) field component below ( above ) @xmath5 .
these scaling behaviors and other salient features of adr are consistent with d - wave density waves .
the recently discovered heavy fermion compound cecoin@xmath0 @xcite has generated a lot of interest , partly due to the many analogies present between this compound and the high transition temperature @xmath6 superconductors .
like the cuprates , cecoin@xmath0 has a layered crystal structure @xcite , a quasi - two - dimensional electronic spectrum @xcite , and a superconducting phase appearing at the border of the antiferromagnetic phase @xcite .
nmr knight shift @xcite and heat conductivity @xcite measurements revealed even parity pairing consistent with @xmath7-wave pairing symmetry , while specific heat @xcite and electrical transport @xcite experiments have shown typical non - fermi liquid behavior and have evoked the possibility of a pseudogap .
in fact , infrared spectroscopy studies have revealed the development of a gap in the density of electronic states of cecoin@xmath0 below @xmath8 100 k @xcite . as in the case of cuprates ,
all these features suggest the existence of quantum critical phenomena @xcite . however , the fermi surface topology is much more complex than in cuprates , with multiple sheets , which implies the participation of several bands in the pairing process @xcite . in this complex picture ,
the nature of the normal state in both systems is not trivial . until recently ,
the nature of the pseudogap ( or the non - fermi liquid ) phase present in cecoin@xmath0 has not been addressed . the giant nernst effect found in this material above @xmath6 @xcite , however , reopened the problem of the origin of the pseudogap state , not only in this compound , but also in the cuprates .
it has recently been shown that the giant nernst effect in cecoin@xmath0 is consistent with unconventional density waves udw or d - wave density waves d - dw @xcite .
we recall that in underdoped cuprates , the pseudogap phase has been attributed to the existence of phase fluctuations of the superconducting order parameter @xcite as well as d - dw @xcite .
in particular , the giant nernst effect observed in la@xmath9sr@xmath10cuo@xmath11 , yba@xmath12cu@xmath13o@xmath14 , and bi@xmath12sr@xmath12cacu@xmath12o@xmath15 has recently been interpreted satisfactorily in terms of d - dw @xcite . here ,
we present new information concerning the nature of the pseudogap of cecoin@xmath0 through angular dependent resistivity adr .
two different scaling behaviors are revealed in this compound in the non - fermi liquid regime for applied magnetic fields less than 5 t. specifically , for angles below ( above ) a certain critical angle @xmath5 , the field component perpendicular ( parallel ) to the @xmath16-planes determines the angular - dependent dissipation .
the critical angle @xmath5 is determined by the anisotropy of the material .
these results and the other salient features of adr measured in high applied magnetic fields are consistent with d - wave density waves with landau quantization of the quasiparticle spectrum .
our findings bring further understanding of the underlying physics in the non - fermi liquid regime of this exotic compound and could also advance the understanding of the pseudogap state of the cuprates .
high quality single crystals of cecoin@xmath0 were grown using the flux method .
typical sizes of the crystals are 0.5x0.5x0.1mm@xmath17 with the @xmath18-axis oriented along the smallest dimension .
we determined the out - of - plane @xmath19 and in - plane @xmath20 resistivities using the electrical contact configuration of the flux transformer geometry , as described in ref .
@xcite .
the angular dependent resistivity @xmath21 was measured by rotating the single crystal from @xmath22axis to @xmath23-axis , with @xmath24 the angle between @xmath2 and the @xmath18-axis . here
we report only in - plane resistivity results .
the angular dependent in - plane resistivity of cecoin@xmath0 single crystals was measured in the non - fermi liquid regime , for @xmath25 k and in magnetic fields @xmath2 up to 14 t. figure 1 is a typical contour plot of resistivity obtained from @xmath26 vs @xmath24 data measured at different fixed applied magnetic field values and at a temperature of 6 k. a similar topologic contour is obtained over the whole @xmath3 range investigated . the squares , circles , and triangles correspond to @xmath27 = 0.45 , 1 , and 2/3 , respectively , which consistent with previously report @xcite , where @xmath27 is the exponent of the power - law @xmath3 dependence of the resistivity ; i.e. , @xmath28 .
note that the contour plot can be divided into three different regions based on the shape of the topologic plots and the values of @xmath27 .
region i is the rectangular part , for which @xmath27 = 1 .
this region is the so - called low - field region ( @xmath29 t ) for reasons given later on .
region ii is the elliptical part , for which @xmath27 = 0.45 and which corresponds to the region around the maximum value of the @xmath30 curves for different angles .
region iii is a large area with @xmath27 = 2/3 , which is outside of regions i and ii at even higher fields .
region i is more interesting since it represents the typical non - fermi liquid behavior ( for this region @xmath27 = 1 ) and displays the two scaling behaviors .
therefore , we will discuss the anomalous magnetoresistivity behavior in this region in more detail .
the bottom inset to fig .
2(a ) shows @xmath31 measured in a low magnetic field of 2 t. [ data were measured in region i. ] the resistivity decreases as @xmath24 increases from @xmath32 , reaches a minimum around @xmath33 , and then displays a peak at @xmath34 .
the data of the top inset to fig .
2(a ) were measured in a high field of 9 t. [ data were measured in region iii . ]
the resistivity increases with increasing angle and displays a shoulder , followed by a peak at @xmath34 .
in the low field region ( region i ) , a spectacular scaling of these data is achieved by plotting the resistivity as a function of the component of the applied magnetic field perpendicular to the @xmath16-planes , i.e. , @xmath35 , as shown in the main panel of fig .
2(a ) .
note that all the curves measured in @xmath36 t overlap for all @xmath24 values between @xmath32 and a critical angle @xmath37 , marked by the arrows , and deviate from this @xmath38 scaling at higher angles .
this value of the critical angle is independent of temperature and applied magnetic field for the range investigated .
we employed a second protocol to measure resistivity , in which we kept the angle between the magnetic field and the @xmath18-axis fixed and scanned the magnetic field .
figure 2(b ) shows the resistivity as a function of the field component parallel to the @xmath39axis ( @xmath40 ) , measured at 6 k for different orientations of @xmath2 .
note that all the resistivity curves scale this time with @xmath41 for @xmath42 .
therefore , the resistivity data follow two scaling laws : @xmath43 figure 1 also clearly shows the presence of the two scaling laws in region i as evidenced by the rectangular region of the @xmath26 contours .
the presence of these two scaling behaviors indicates that below ( above ) the critical angle , the field component perpendicular ( parallel ) to the @xmath16-planes determines the in - plane dissipation .
a similar analysis of the angular dependent resistivity data measured in high magnetic fields ( region iii ) has shown that the data are still dominated by the perpendicular ( parallel ) field component below ( above ) the critical angle , although the two scaling laws are no longer present .
next we try to understand the above shape of @xmath31 measured in both low and high fields and the relationship between the two scaling behaviors .
we plot in the inset to fig .
3 the resistivity measured at 6 k in scanning @xmath2 up to 14 t at the two fixed angles ( @xmath44 and @xmath45 ) corresponding to the two field orientations ( perpendicular and parallel , respectively , to the @xmath16-planes ) which seem to determine the physics below and above @xmath5 , respectively . for the first field orientation ( @xmath46 planes ) ,
@xmath47 increases with increasing @xmath48 , reaches a maximum around 5 t , and decreases with further increasing @xmath48 .
we define the low ( high ) field region as the field region for which @xmath49 5 t ( @xmath50 5 t ) , based on this behavior of @xmath47 .
this @xmath47 dependence gives a qualitative explanation of the @xmath31 dependence [ shown in the two insets to fig .
2(a ) ] for @xmath51 , i.e. , at angles at which the dissipation is dominated by the perpendicular field component @xmath52 .
namely , at constant @xmath2 , the resistivity should decrease ( increase ) with increasing @xmath24 , i.e. , decreasing @xmath48 , for fields lower ( higher ) than 5 t as , indeed , shown by the bottom ( top ) inset to fig .
2(a ) for @xmath51 . for the second field orientation ( @xmath53-axis ) , the @xmath54 data of the inset to fig .
3 show an initial increase with increasing @xmath55 followed by a tendency to saturation for @xmath55 approaching 14 t. this monotonic @xmath54 dependence explains qualitatively the increase in @xmath31 [ shown in the two insets to fig .
2(a ) ] for @xmath56 , i.e. , at angles at which the dissipation is dominated by the parallel field component @xmath57 , in both low and high field regimes . in summary ,
the @xmath47 and @xmath54 dependences of the inset to fig .
3 explain the non - monotonic @xmath31 dependence at low fields and the monotonic @xmath31 dependence at high fields , as shown by bottom and top inset , respectively , to fig .
2(a ) .
figure 3 shows that the two resistivity curves of its inset , @xmath47 and @xmath54 , scale for @xmath58 t if @xmath54 is replaced by @xmath59 , with @xmath60 .
the physical meaning of @xmath61 is the anisotropy of the material .
the scaling of fig .
3 implies that the same physics is responsible for the two scaling behaviors displayed by figs . 1 and 2 for @xmath58 t , i.e. , @xmath62 and @xmath63 , respectively .
the scaling relationship of fig .
3 also allows one to determine the critical angle @xmath5 below ( above ) which the resistivity data scale when plotted vs the perpendicular ( parallel ) component of the applied magnetic field ( figs .
2(a ) and 2(b ) ) .
recall that eq .
( 1 ) gives : @xmath64 .
this latter relationship and the scaling of fig . 3 [ ( @xmath65 for @xmath66 4 t ) ] give @xmath67 .
therefore , @xmath68 with @xmath69 , as determined above .
this value of the critical angle is in excellent agreement with the @xmath70 value obtained experimentally , which proves the consistency of our analysis of the angular and field dependent data .
so , the critical angle is given by the anisotropy of the material .
the scaling of the magnetoresistivity curves for @xmath71-axis ( for which the lorentz force is maximum ) and for @xmath72-axis ( for which the lorents force is zero ) shown in fig .
3 implies that the spin effect , rather than the orbital effect , is responsible for the magnetoresistivity in region i. the fact that in this region the resistivity is also linear in @xmath3 , indicates that this linear @xmath3 dependence is as well a result of spin fluctuations .
this conclusion that spin fluctuations dominate the charge transport in the non - fermi liquid regime is consistent with recent in - nqr and co - nmr experiments@xcite , which have revealed that the magnetic nature in cecoin@xmath0 is characterized by strong antiferromagnetic spin fluctuations in the vecinity of the quantum critical point . in the following ,
we show that all the observed features of adr of cecoin@xmath0 in the non - fermi liquid regime , when a magnetic field is rotated from @xmath22axis to @xmath73axis with @xmath24 the angle between @xmath2 and the @xmath74axis , can be consistently discribed in terms of d - wave density waves in a magnetic field . the unconventional density wave udw or d - wave density wave d - dw is a kind of density wave in which the gap function @xmath75(@xmath76 ) vanishes on the line nodes .
therefore , the transition from the normal state to the udw is a metal to metal transition @xcite .
the udw exhibits two characteristics : angular dependent resistivity and giant nernst effect .
a udw in the low temperature phase of @xmath77-(et)@xmath78khg(scn)@xmath79 and the metallic phase of ( tmtsf)@xmath78x with x = pf@xmath80 and reo@xmath11 has been identified through adr @xcite .
we assume that the electrical conductivity in the non - fermi liquid regime is given by : @xmath81 where @xmath82 is the energy of all fermionic excitations ( holes and particles ) .
as shown below , this new formula appears to be more appropriate than the one used earlier @xcite . in the absence of a magnetic field , we assume that the quasiparticle energy spectrum is given by : @xmath83 where @xmath84 and @xmath85 ; here @xmath86 and @xmath87 are fermi velocities in the @xmath16-plane and along the @xmath18-axis , respectively , @xmath88 is the radial wave vector within the @xmath89plane , @xmath75 is the maximum value of the d - dw energy gap @xmath90 , and @xmath91 . here
we assume that @xmath90 has @xmath92-wave symmetry . in the vicinity of the nodal points , it is convenient to replace @xmath93 by @xmath94 , where @xmath95 is perpendicular to @xmath88 within the @xmath16 plane and @xmath96
. then in a magnetic field which makes an angle @xmath24 with the @xmath74axis , the energy spectrum becomes @xcite : @xmath97 @xmath98 here @xmath99 and @xmath100 are the two branches of the landau levels , n = 0 , 1 , 2,@xmath101 , and @xmath102 is the chemical potential .
we note that in ref .
@xcite only @xmath99 is considered .
the electrical conductivity is then given by : @xmath103 where @xmath104 and @xmath105 .
for simplicity , we take only the n = 0 and n = 1 landau levels . equation ( 6 ) fits very well the @xmath31 data over the whole range of the applied magnetic field , both in the low and high field regimes .
representative plots of the resistivity data measured in 4 , 8 , and 10 t along with the fitting curves are shown in fig .
4 . in these fits
we assume that @xmath106,\ ] ] with @xmath27 = 0 , 1 .
the fitting parameters determined by fitting the data measured in different magnetic fields are almost constant , hence field independent , as they should be .
their values are : the anisotropy @xmath61 = @xmath107 , @xmath108 m/s , @xmath109 k and @xmath110 t. this value of the anisotropy @xmath61 of 1.9 is very close to the value of 1.7 required to scale the data shown in fig .
3 . with @xmath111
m/s , obtained by fitting the @xmath112 curve of cecoin@xmath113 @xcite and the above value of @xmath114 , we get @xmath115 = @xmath116 0.1 .
provided that @xmath117 k , this gives @xmath118 k , a value expected from the weak - coupling theory for d - dw with @xmath119 = 20 k. @xcite this overall consistency between the obtained and expected values of the fitting paramenters further attests to the appropriateness of the model used to analyze the experimental data . moreover , the d - dw model given by eq .
( 6 ) also gives the scaling observed experientally at @xmath120 5 t. figure 5 is a plot of the @xmath31 data ( open symbols ) , the fitting with eq .
( 6 ) ( solid line ) , and the two scalings ( dashed lines ) calculated from eq .
( 6 ) in which @xmath99 and @xmath100 , eq . ( 4 ) and ( 5 ) , respectively , include only the @xmath38 or @xmath41 terms , for @xmath121 and @xmath122 , respectively .
note the excellent agreement between the fitting line and the two scaling lines .
therefore , even though the model proposed here is quite complex , it gives the two experimentally observed scaling laws at low magnetic field values . in summary , comprehensive angular dependent resistivity adr measurements in the non - fermi liquid region of cecoin@xmath0 at @xmath123 k
have shown the presence of two different scaling behaviors in the low - field region ( region i ) which involve the @xmath2 and @xmath24 dependence and are due to spin fluctuations .
the two scaling regions are separated by a critical angle @xmath5 , which is given by the intrinsic anisotropy . in the scaling region ,
the resistivity is linear in @xmath3 . at higher fields ,
the adr data are governed by the same physics , even though the two scalings fail .
a possible explanation for this anomalous adr in terms of @xmath7-wave density waves with landau quantization of the quasiparticle spectrum was presented .
this approach describes very well the salient features of adr data : ( i ) the distinctive angular dependences observed in the low ( @xmath124 ) and high ( @xmath125 ) field regimes follow naturally from eq .
( 7 ) ; ( ii ) the low field scaling behaviors follow from the quasiparticle spectrum given by eqs .
( 4 ) and ( 5 ) .
in addition , the quasi two dimensional aspect of d - wave density waves is explored here for the first time .
this research was supported by the national science foundation under grant no .
dmr-0406471 at ksu and the u. s. department of energy under grant no .
de - fg02 - 04er46105 at ucsd .
the authors acknowledge useful discussions with dr .
v. sandu
. 33 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , , , , , , , , * * , ( ) . , , , , , * * , ( ) .
, , , , , , , , , , * * , ( ) . , , , , , , , * * , ( ) . , , , , , , , * * , ( ) . , , , , , , , * * , ( ) . , , , , , , * * , ( ) . , , , , , , , , * * , ( ) . , , , , , * * , ( ) . , , , , , , , * * , ( ) . , , , , * * , ( ) . , , , , , , , , , , * * , ( ) . , , , , , , , , , , * * , ( ) . , , , , , , , , , * * , ( ) . , * * , ( ) . , , , , , , , , * * , ( ) . , , , , * * , ( ) . , , , , , * * , ( ) . , , , , , , , * * , ( ) . , , , , , , , , * * , ( ) . , , , , , , , , * * , ( ) . , , , , , , * * , ( ) . ,
* * , ( ) . , , , * * , ( ) . , , , , * * , ( ) . , , ,
, * * , ( ) . , , , , , , ,
, * * , ( ) . , ,
, , , * * , ( ) . , , , * * , ( ) . , , , , ,
, * * , ( ) . , ,
, , * * , ( ) . , , , * * , ( ) . , , , , , ,
* * , ( ) . |
in the cold dark matter framework , large spiral galaxies are built hierachically , and there is much observational evidence of galaxy interactions to support this .
however , we still face many challenges in understanding the details of galaxy evolution . in this regard ,
it is interesting to study the hot , massive stellar populations in nearby galaxies which can be a robust tracer of recent star formation activity due to their short lives .
young massive stars contribute to the global characteristics at the current epoch of their host galaxy and have a major role in the galaxy evolution .
young massive stars emit powerful ultraviolet ( uv ) radiation , therefore uv imaging is ideal to detect and characterize these stars , which are confused with older stellar populations in observations at longer wavelengths ( * ? ? ?
* and references therein ) . in most cases ,
young massive stars are formed in associations in the galactic disk .
we define these stellar associations from uv imaging , in order to study the spatial structure and intensity of the recent star formation .
_ galaxy evolution explorer _
( _ galex _ , * ? ? ?
* ) imaging in far - uv ( fuv ) and near - uv ( nuv ) passbands is particularly useful to study massive stellar populations .
specifically , the integrated fuv @xmath9 nuv color of young stellar populations is very sensitive to the age of the population , owing to the rapid evolution of the most massive stars ( e.g. bianchi 2007 , 2009 : fig .
consequently , uv imaging data allow us to unambiguously identify the young stellar populations and to estimate their ages and masses from colors and extinction - corrected uv luminosities , respectively .
these results provide the means of probing the history and modality of recent star formation in galaxies .
m31 is one of the two largest spiral galaxies in the local group , along with the milky way .
however , there has been growing evidence that the milky way and m31 have different properties ( * ? ? ?
* and references therein ) .
m31 shows a lower star formation rate ( sfr ) than the milky way @xcite .
furthermore , there are suggestions that m31 appears to be a more typical spiral galaxy than the milky way @xcite . as with the bulk of local spirals , m31 shows evidences for a formation and evolution history affected by merging and/or accretion events , including substructures in its halo (
* and references therein ) . in this respect ,
it is important to investigate the star - forming ( sf ) regions in m31 .
we present the first extensive study of the young stellar populations from uv imaging , covering the entire m31 disk within 26 kpc radius , and extending beyond this radius in some fields . by studying the sf regions in m31 traced by uv imaging
, we investigate the recent star formation history of this galaxy . in section 2 , we describe the _ galex _ uv data used .
we describe the detection and photometry of sf regions , from uv imaging , in section 3 , and estimate their interstellar extinction in section 4 . in section 5
, we select ob associations from ground - based photometry of stellar sources and compare them with our uv - defined sf regions . the age and mass of the sf regions
are derived in section 6 , the star formation history in section 7 , and the conclusions are presented in section 8 .
we focus on the disk region of m31 where most of the recent sf regions are located .
we adopt the disk semi - major axis value of r = 26 kpc from @xcite .
we considered 20 fields from the _ galex _ fourth data release ( gr4 ) .
_ galex _ observed the disk region of m31 as part of the nearby galaxies survey ( ngs , * ? ? ? * ) .
we rejected the fields which have only fuv observations and exposure time shorter than 2,000 s ( except for ps_m31_mos03 " ) , then selected the fields closer to the major axis when different pointings are located along the galaxy s major axis .
ps_m31_mos03 " field has a shorter exposure than 2,000 s but it is the only _ galex _ field observed in the south outermost disk region of m31 . as a result , we selected 7 _ galex _ fields covering the entire disk region ( see fig . 1 and table 1 ) .
each selected _ galex _ field has fuv ( 1344 - 1786 ) and nuv ( 1771 - 2831 ) imaging with the same exposure time .
_ galex _ fuv and nuv imaging has 4.2 and 5.3 arcsec resolution ( fwhm ) or about @xmath319 pc in m31 , and the field of view ( fov ) is about 1.27 and 1.25 degree respectively @xcite .
we used in our analysis only the central 1.1 degree diameter portion of the field , for best photometry .
_ galex _ images have a sampling of 1.5 arcsec pixel@xmath8 which corresponds to 5.67 pc , assuming a distance of 785 kpc @xcite .
the fuv images provide a direct measure of the flux from young massive stars not heavily embedded in interstellar dust .
most sf regions are gravitationally unbound systems and have irregular shapes . rather than using aperture photometry , we constructed contours of the sf regions to trace their morphology , and measure their uv flux and flux density .
the procedure was originally developed by @xcite in his dissertation , and we have modified the procedure for a more precise detection and photometry of the sf regions . our procedure consisted of three steps .
the first step was to detect all image pixels which have flux above a certain threshold in each fuv image .
an important factor in detecting and defining sf regions is the brightness limit .
the second step was to define the contours of each sf region from contiguous pixels detected above the threshold over a minimum area . in this way
, we can define contours of sf regions even if they have a complicated shape , and reject isolated stars which are smaller than the minimum size .
the third step was to estimate the background and to measure the flux of the defined sf regions .
even though the background in the uv image is much lower than in the optical , it is important to correctly subtract its contribution from the source photometry .
we performed various tests for determining the optimum flux threshold , minimum size of a sf region , and best background subtraction method .
we used the field nga_m31_mos0 " which has the longest exposure time ( 6,811 s ) in our selected _ galex _ data , to test and refine our procedures , which were then applied to all our selected fields .
this field is good for testing various types of sf regions because it contains portions from innermost to outermost spiral arms and the large ob association ngc 206 .
first of all , we compared various thresholds for the detection of source pixels .
we used the background - subtracted image ( -intbgsub " ) provided by the gr4 pipeline .
we estimated the mean background value by the sigma clipping method .
we examined the results using thresholds of two , three , and five sigma above the mean background value ( see fig .
we detect fainter objects if we use the lower thresholds , however the sf regions in the spiral arms merge together and the contamination by the background ( including older , diffuse populations ) is larger .
we can easily define the bright regions if we use higher thresholds , however we can not detect the faint ones .
the fuv magnitude limits of our detected sf regions from each threshold are shown in fig .
3 , they are @xmath321.5 ( low ) , @xmath321.0 ( mid ) , and @xmath320.4 ( high threshold ) mag in the ab magnitude system .
we adopted a threshold of three sigma above the mean background value , which showed in our analysis less contamination by background and marginally detects the faint regions .
this results in an average fuv flux threshold of @xmath30.0032 c s@xmath8 pixel@xmath8 , or @xmath325.9 mag arcsec@xmath4 .
then , we considered the minimum acceptable size of the sf regions , in order to eliminate contamination by background objects , artificial sources , foreground stars , and isolated bright stars in m31 .
we considered 3@xmath10fwhm of _ galex _
( @xmath313 arcsec or @xmath38.5 pixels ) as minimum diameter of a sf region , therefore we adopted a requirement of a minimum of 50 contiguous pixels ( @xmath31,600 pc@xmath2 or @xmath340 pc ) for the smallest sf region .
we contoured the contiguous pixels selected for measuring the defined regions and we estimated the center of each sf region by the mid - point of the maximum diameter of its contour .
these adopted threshold and minimum size of the sf regions , were then applied to all the selected fields .
however , the exposure time differences among fields induces variation of the detection limit .
therefore , we compared the detection fraction in overlapping image regions , one by one against the nga_m31_mos0 " field .
we first tried a fixed threshold ( @xmath30.0032 c s@xmath8 pixel@xmath8 ) for source detection in the nga_m31_mos0 " image . in this case , in images with shorter exposure time sources are over - detected and include noise .
a second method used the variable threshold of three sigma above the mean background value from each image .
this produced similar contours in the overlapping regions .
however , some sources in the shorter exposure time images were undetected because the background has larger sigma values .
finally , we adopted the ratio between detected and undetected pixels of overlapping regions .
this case produced slight over - detection in the shorter exposure time images .
therefore , we used the last method , manually varying the thresholds in each field , such as to obtain similar detections in overlapping regions .
the final selected fuv pixel maps are shown in fig .
4 . as a last step , we compared the detected sf regions within overlapping image portions by visual inspection to select a final catalog of unique sources . at the end , we obtained a catalog of 894 sf regions cleaned of artifacts , isolated stars , and overlapping objects .
we measured the flux within the contours of the sf regions as defined above , and subtracted the local background , estimated in a circular annulus surrounding the source .
we initially used an inner / outer size of the annulus of 1.5/3 times the size of each sf region .
however , for the largest sources , which are mostly found along the spiral arms , the annulus for background measurement scaled in this way becomes too large and includes unrelated stellar populations .
therefore , we tested three different procedures to estimate the background .
one was to measure the local background ( ap .
sky " in fig .
5 ) , adjusting the radius of the background annulus according to the size of the source .
we adopted a variable scaling of inner / outer radii for the background annulus , adjusted as 2/4 , 2/3 , 1.5/2.5 , 1.1/1.3 , and 1.05/1.2 times the source size ( r@xmath11 ) for the following ranges of source size respectively : r@xmath11
@xmath12 10 , 10 @xmath5 r@xmath11 @xmath12 20 , 20 @xmath5 r@xmath11 @xmath12 50 , 50 @xmath5 r@xmath11 @xmath12 100 , and 100 @xmath5 r@xmath11 ( [ pixels ] ) .
such ranges were found adequate to ensure a large enough area for the background measurement for the smallest sources , while preventing excessively large areas to be included in the calculations for large sources .
the second background estimate was obtained from the gr4 pipeline background image ( -skybg " ) , and the third measurement was performed by applying a median filter to the image . in the case of the gr4 pipeline background ( pipe .
sky " in fig .
5 ) , a sky background image is produced by a 5@xmath105 median filter size which is good for the case of poisson distribution @xcite .
the background measured using this image , underestimates the local contribution by the galaxy s background light , especially from surrounding stellar populations in the spiral arms . in the case of the median
filtered background ( med .
sky " in fig .
5 ) , a smaller filter size ( 3@xmath103 pixels ) than the standard pipeline was used to produce the background images .
the choice of a smaller filter size was driven by the consideration of spiral arm regions which host most of the sf regions .
the results from this background estimate are similar to those from the local background and reflect well the brightness of spiral arms .
the results from the three methods are compared in fig .
the background from the pipeline sky image is always underestimated because it measures the lowest sky level and not the local diffuse stellar population surrounding the source . in the right - side panels in fig . 5
, we see that this estimate is not sensitive to the spiral arm enhancements , as the local - background estimates are .
the local sky estimate is similar to the results from the median - filtered measurements but shows less scatter for sources with multiple observations in overlapping regions . in fig .
6 , we compare the photometry results using the three different methods for background subtraction in color - magnitude diagrams .
most sf regions have fuv @xmath9 nuv color between @xmath13 and @xmath14 in abmag .
the measurements from the pipeline and median - filtered backgrounds induce brighter nuv than fuv , especially for sources fainter than 20 mag .
we finally adopted the result from the local background .
the resulting catalog of sf regions and their _ galex _ photometry is given in table 2 .
in order to derive the physical parameters of the sf regions from the integrated photometry , we must take into account the interstellar extinction .
we estimated the reddening of each uv sf region from the reddening of the massive stars included within its contour . for massive stars we used the reddening - free parameter q @xcite .
we used the optical point - source measurements of m31 sources from p. massey ( priv .
comm . , 2009 ) , which is a revised m31 catalog from the noao survey data described by @xcite .
we selected sources with @xmath15 measurements having photometric errors lower than 0.1 mag in all bands ( 108,089 objects out of their 371,781 total sources catalog ) , which results in a magnitude limit of about 23rd mag in @xmath16-band .
this data is deep enough to select ob type stars which we used to estimate the interstellar reddening of our sf regions .
we selected ob type stars by comparing colors and brightness of the galactic ob type main sequence stars from @xcite .
we selected stars from o3 to b2v , because a b2v star has @xmath17 = @xmath18 which is @xmath16 = @xmath19 in m31 ( @xmath20 = 24.47 ) in absence of extinction .
we do nt know the internal reddening of m31 , therefore we used the reddening - free parameter @xmath21 to select the ob stars .
@xmath21 = @xmath22 - @xmath23@xmath24 the @xmath25 ratio is a constant ( 0.72 ) for milky way dust type @xcite and does not vary much for other dust types , within a reasonably small range of @xmath26 ( e.g. * ? ? ?
* ; * ? ? ?
* and references therein ) .
we selected the ob stars which have @xmath21 between @xmath27 ( o3v ) and @xmath28 ( b2v ) . in order to reduce contamination , we also adopted magnitude and color limits ; @xmath29 @xmath5 @xmath17 @xmath5 @xmath30 , @xmath31 @xmath5 @xmath32 @xmath5 0.46 , and @xmath33 @xmath5 @xmath34 @xmath5 @xmath35 @xcite .
the color and magnitude limits are derived assuming a maximum reddening value of @xmath26 @xmath5 0.7 .
the location of the selected ob stars in color - magnitude diagrams are presented in fig .
7 . with these restrictions
, we finally selected 22,655 o - b2 stars from the data of @xcite .
the spatial distribution of these stars , shown in fig .
1 , represents well the spiral structure of m31 .
the interstellar extinction of the selected ob stars was estimated from the reddening free parameter q@xmath36 , using the empirical relationship for giant and main - sequence stars by @xcite .
@xmath37 the estimated @xmath26 has mean , median , and mode value of about 0.34 , 0.32 , and 0.29 , respectively , from our selected ob stars .
the mean reddening value of ob stars is larger than @xcite s typical value @xmath26 = 0.13 which is estimated visually from the location of the `` blue plume '' in the color - magnitude diagram .
this difference may be caused by our selection of blue stars for the reddening estimate .
we also explore in this paper more reddened regions than the average entire stellar population .
we estimated the extinction by interstellar dust for each sf region from the average reddening of the ob stars within the sf region contour .
for the sf regions outside of @xcite survey , we assumed @xmath26 = 0.20 .
the spatial distributions of estimated interstellar reddening of ob stars and sf regions are shown in fig
the interstellar extinction decreases from the inner disk region outwards .
in particular , inner - most and south - west areas where we could not detect sf regions , have high interstellar extinction .
we will return to this point in the next section .
our detection method is based on the fuv flux , which could vanish in high interstellar extinction regions .
for comparison with our fuv - defined sf regions , we also used a path linkage criterion ( plc : * ? ? ? * ) method ( explored by * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) to detect ob associations using o - b2 stars .
we applied the plc method varying the minimum number of stars ( n@xmath38 ) and maximum link distance ( d@xmath39 ) ( see fig .
the best choice of n@xmath38 and d@xmath39 was found to be n@xmath38 = 5 stars and d@xmath39 = 10.4 arcsec ( @xmath340 pc ) .
@xcite found 174 ob associations and estimated a total number of @xmath3420 associations in m31 from a similar method but different optical photometry . with this method
, we found 650 ob associations in m31 from the o - b2 stars selected by us from @xcite photometry , which is @xmath3 2 mag deeper than what @xcite used .
we compared these ob associations with the 894 sf regions selected from fuv imaging .
they mostly overlap with the fuv - detected regions ( see fig .
10 ) , however some of fuv - selected sf regions have a larger area than ob associations defined from stellar photometry , and some additional ob associations are found from stellar photometry in high interstellar extinction regions .
the observation fields of @xcite cover a smaller area than our _ galex _ imaging , therefore we compared our results within 17 kpc de - projected distance ( about 1 deg@xmath2 ; dashed ellipse in fig .
10 ) from the center of m31 .
the total area of the sf regions derived from the fuv contours ( section 3 ) , and of the ob associations derived from stellar photometry ( above ) , are 4.1 % and 3.5 % of the area of the 17 kpc disk , respectively .
the numbers of o - b2 stars are 9094/11350 and 8670/11774 inside / outside of the uv - defined sf regions and of ob associations .
the average projected density of ob stars in the associations is 0.017 and 0.018 stars arcsec@xmath4 from uv - selected sf regions and ob associations selected from stellar photometry , respectively .
the projected density of stars inside sf regions is about a factor of 20 higher than in general field .
the comparison between the two methods is interesting , because the fuv - selected contours are more affected by interstellar reddening than the optical stellar photometry , on the other hand optical bands are not as sensitive as the fuv is to the t@xmath40 of the hottest stars ( e.g. * ? ? ?
* and references therein ) .
the similarity of the total number of ob stars detected inside young associations and in the field , by the two methods is remarkable .
the slight difference in the estimated area of the associations may be due , at least partly , to the low ( @xmath415 arcsec ) spatial resolution of the _ galex _ imaging .
we estimated the ages of the sf regions by comparing the measured ( fuv @xmath9 nuv ) colour with synthetic simple stellar population ( ssp ) models , reddened by the extinction amount estimated for each region .
we explored effects of metallicity and dust type on the results ( e.g. * ? ? ?
* ; * ? ? ?
* ) . then we estimated the masses of the sf regions from the reddening - corrected uv luminosity and the derived ages .
we used two sets of ssp models , one from @xcite ( bc03 ) and the other from padua ( pd : a. bressan , priv .
comm . , 2007 ) .
the ages derived from the two grids of models do not differ significantly ( see lower left panel of fig .
ages estimated using the bc03 models tend to be slightly younger than those derived using the pd models , below 30 myrs , and do not differ at all for older populations ( fig .
11 , lower left panel ) .
the small age difference propagates to the derived masses , as shown in the lower - right panel of fig .
the differences between results from the two model grids is not significant .
we used the pd models in our analysis .
most sf regions have uv colour between @xmath13 and 1.0 ( see upper left panel of fig .
. the estimated ages of most sf regions in our sample are younger than 400 myrs , reflecting our fuv - based selection .
our detection limit , plotted with a line in fig .
11 , indicates quantitatively how the flux - detection limit translates into mass detection limit , as a function of age , and shows that we can not detect low mass sf regions at older ages , as expected .
however , we also notice a lack of massive sf regions at younger ages . this will be discussed later .
we explored three metallicity values : subsolar ( z=0.008 ) , solar ( z=0.02 ) , and supersolar ( z=0.05 ) metallicity , although m31 is believed to have a typical metallicity about twice higher than the mw ( e.g. * ? ? ?
* and references therein ) .
our census of young stellar populations based on wide - field fuv imaging has an unprecedented extent , while direct metallicity measurements from spectroscopy are confined to limited samples .
therefore , we wanted to assess in general the dependence of our results on metallicity , which may vary in some environments .
we also examined the effect of four types of interstellar dust : milky way ( @xmath42 = 3.1 ; mw , @xcite ) , average large magellanic cloud ( avglmc ) , 30 doradus ( lmc2 ) @xcite , and small magellanic cloud ( smc , * ? ? ? * ) dust extinction .
the resulting ages and masses of the sf regions are plotted in figs 11 , 12 , and 13 .
the differences in derived ages and masses for three metallicity values and different types of interstellar dust are presented in fig .
we considered metallicity values of no less than z=0.008 because we expect the young sf regions not to be as metal poor as old globular clusters . in fig .
12 ages and masses derived from models with solar metallicity , and assuming mw - type interstellar reddening ( r@xmath43 = 3.1 ) , are compared to results from subsolar ( z=0.008 ) and supersolar ( z=0.05 ) metallicities ( top four panels ) .
the derived ages are older for subsolar metallicity , and younger for supersolar metallicity , with respect to solar metallicity results .
the differences are most significant for ages younger than 100 myrs ( see also fig .
9 of * ? ? ?
* ) and are up to a factor of @xmath33 at most . because the effect is stronger at certain ages ,
the number distribution of sf regions with ages also differs , as shown in fig .
12 . this will be taken into account in the following analysis , where we derive the global sf in m31 as a function of time .
the difference in the derived masses is not conspicuous , considering the uncertainties .
the uncertainty of the ages derived by comparing the photometry to synthetic population models , reported in table 2 , is derived by propagating only the photometric errors , because we investigated and showed separately the effects of different metallicity values and dust types .
the uncertainty on the derived masses reflects the uncertainty on the photometry and the age .
reddening corrections , as derived in section 5 , are applied .
the lower four panels of fig .
12 show the effects of the correction for reddening
. if the selective extinction by dust is steeper in the uv than the mw dust , as observed for example in the lmc ( average ) or the extreme starburst regions 30 dor ( labelled as avglmc " and lmc2 " , respectively , in fig . 12 ) , the dereddened uv luminosity will be higher but the dereddened fuv @xmath9 nuv color bluer , resulting in much younger ages and consequently lower masses .
an lmc - type dust is however not likely in m31 .
@xcite report uv extinction curves in m31 similar to the average mw extinction , from uv spectra of ob stars . moreover , if we apply uv extinction steeper than mw dust , most measured fuv @xmath9 nuv colors appear to be over - corrected , when compared to ssp model predictions . out of 847 sf regions
whose ( fuv @xmath9 nuv)@xmath44 is within the model color range when dereddened with mw dust type , only 569/227/26 sf regions have ( fuv @xmath9 nuv)@xmath44 colors within the model range if the progressively steeper dust types avglmc / lmc2/smc are applied . therefore , uv - steep dust extinction seems to be not realistic in most cases .
the fuv - selected sf regions follow the disk structure of m31 and their spatial distribution traces the recent star formation in m31 ( fig .
13 ) . as can be seen in fig
. 13 , most sf regions have de - projected distances between 40 and 75 arcmin ( 9 and 17 kpc ) from the center of m31 , with two peaks in this region .
this region is well - known as the ring of fire or the star - formation ring ( * ? ? ?
* and references therein ) . in this star formation ring between 40 and 75 arcmin from the galaxy center ,
the number of young ( @xmath12 10 myrs ) sf regions is similar to that of older ( @xmath45 10 myrs ) ones .
one thing of interest is that the number of younger sf regions ( 82 ) is larger than the number of older sf regions ( 40 ) outside of this ring ( @xmath46 @xmath45 75 arcmin ) . inside of the star formation ring ,
the number of younger sf regions ( 14 ) , however , is less than that of older ones ( 89 ) .
this suggests that the m31 disk formed stars continuously during the last few hundreds myrs at least and , furthermore , the outer disk shows more recent star formation .
the size of some younger sf regions is larger than that of older sf regions , however most of them are less massive ( as derived from the uv flux ) than older regions .
young , large sf regions may be broken into several sf regions with time , while some of the dense , small sf regions may survive longer than the others .
we noticed a lack of massive sf regions younger than 50 myrs in fig .
we would expect to find some massive young sf regions if star formation was constant .
we estimated the total sfr in m31 using the ages and masses of the sf regions derived in section 6 .
we added the estimated initial masses of the sf regions separated in four time bins , to investigate the sfr time evolution .
the results are shown in fig .
14 , and reported in table 3 , for three metallicity values .
although in each age interval the derived sfr depends on the assumed metallicity , in all cases fig.14 shows an apparent decrease of sfr in the recent epoch ( @xmath12 10 myrs ) with respect to the average value in the interval 10 - 100 myrs ago , the difference being smallest ( and probably not significant ) for supersolar metallicity , which is currently believed to be the most probable value for m31 .
the dashed line across the whole time interval is the average from the sf regions of all ages .
when interpreting this diagram , we must first of all recall that the masses are estimated from the uv flux above a certain threshold , and the corresponding limit for mass detection increases at older ages ( fig .
11 , continuous line ) .
therefore , the total stellar mass formed over 100 myrs ago may be relatively underestimated compared to that of younger populations , due to our fuv selection .
this bias makes the apparent decrease in sfr at young ages more robust .
most of the time - binned and mean sfrs are lower than one solar mass per year .
these values can be compared to the m31 s sfrs from ir and h@xmath6 .
@xcite estimated a sfr of 0.4 @xmath47 yr@xmath8 from 8 @xmath48 m non - stellar emission , higher than our uv - derived sfr in the @xmath124 myrs bin , if we assume solar metallicity , but comparable to our value for supersolar metallicity .
this shows that each indicator alone , uv or ir , may miss about half of the most recently formed stellar mass .
@xcite estimated 0.05 @xmath47 yr@xmath8 from h@xmath6 luminosities , lower ( within a factor of two ) than our estimate for the @xmath1210 myrs bin ( solar metallicity ) .
we used 7 _ galex _ fields covering the entire disk of m31 ( out to @xmath3 26 kpc radius ) to study its young stellar population .
we detected 894 sf regions from the fuv imaging , and measured their integrated fuv and nuv fluxes .
we estimated the interstellar extinction in each sf region from the ob stars within its contour , using the ground - based stellar photometry of @xcite .
we estimated ages and masses of our sf regions in m31 , detected from fuv imaging , by comparing the fuv and nuv measurements to population synthesis models .
most are younger than 400 myrs ( uv is not sensitive to older ages ) .
interestingly , there are no massive sf regions at young ages ( age @xmath12 50 myrs in fig .
11 ) , suggesting an apparent decrease of the sfr , as detected from the fuv imaging . there may be a potential bias , due to our definition of the sf contours from the fuv image , concerning the size distribution and average surface brightness .
younger sf complexes tend to be more compact , and older stellar associations are less dense , in general @xcite .
therefore , the same flux detection algorithm may break a young sf area into several peaks , but these contiguous regions may appear connected if they spread out at older ages .
this bias , however , does not appear significant from visual inspection of the complex sf areas along spiral arms , and younger associations tend to have larger size in our selection . in any case , the bias would only affect the apparent distribution [ with age ] of sizes and masses of the individual sf regions , and not affect the results when we add all individual masses in a wide age interval , in order to derive the total sfr in m31 .
because we have estimated ages and masses of the individual sf complexes , and the uv flux is sensitive to a much broader age range than e.g. h@xmath6 or ir , we were also able to derive the recent sfr in m31 as a function of time .
we estimated the sfr in four time intervals adding the masses of the sf regions of corresponding ages .
the resulting masses are restricted by our detection limit which is shown in fig .
the plot in fig .
14 shows a recent apparent decrease of the sfr , as derived from the uv flux .
however , we know that the youngest and compact sf regions are often still embedded in dust ( e.g. * ? ? ?
* and references therein ) .
these escape detection from uv imaging , but are instead revealed by ir emission from heated dust .
therefore , for a complete account of star formation in the very young age bins ( @xmath12 10 myrs ) , we added the ir - measured star formation from @xcite .
the total sfr at young ages ( uv + ir estimates ) is shown as thick lines at the top of vertical arrows in fig .
this more realistic and complete estimate of the star formation in the recent few million years significantly reduces the apparent recent decrease in sfr , as derived from uv flux only .
we also show the average stellar mass formed in a @xmath5 10 myrs bin , from our uv measurements , for comparison to h@xmath6 estimates . because h@xmath6 is an indirect measurement of the ionizing photons from the short - lived o - type stars , the sfr derived from this method
must be compared to our age bin of @xmath5 10 myrs .
the uv and h@xmath6 sfr estimates agree within a factor of two , for solar metallicity , as shown in fig . 14 .
if we assume supersolar metallicity for all sources ( z=0.05 ) , then the h@xmath6 underestimates the sfr by a factor of ten with respect to the sfr derived from the uv sources in the recent 10 myrs .
we note that the results shown in fig .
14 are derived by dereddening the uv fluxes with mw - type dust extinction .
if uv - steeper reddening applies to the intrinsic dust extinction in some sf regions , their derived ages would be younger ( see fig .
12 ) , resulting in a sfr lower for older ages and higher for younger ages than the values shown in fig.14 , and increasing the discreapancy with the h@xmath6 estimate .
even when the ir- and uv - derived sfrs are added at the youngest epochs , there seem to be a recent decrease of sfr , or a peak of sfr between @xmath3 10 - 100 myrs ( although the uv estimate of sf at older ages is a lower limit , as previously explained ) .
this seems to suggest that m31 had a starburst during this interval .
a possible scenario is that galaxy interaction may have induced violent star formation around this time interval .
@xcite considered a collisional event with m32 around 20 myrs ago to explain the ring structure by dynamical models .
@xcite also postulated a head - on collision event with m32 and estimated it happened around 210 myrs ago .
we now have a possible evidence of a recent starburst in m31 , which may have constructed the ring structure .
this work presented the first estimate of the recent sfr based on measurements of individual sf regions across the entire galaxy , from uv imaging .
@xcite provide a comparison of sf , based on h@xmath6 measurements , among local group galaxies , and several other authors give estimates of sf in nearby galaxies .
however , their results are mostly derived under the assumption of continuous star formation , translating global fluxes into sf .
our study provides a time - resolved sfr over the past few hundred million years .
we will perform a similar analysis on other local group galaxies , for a consistent comparison of results from our method within a range of physical environments .
such comparison should also clarify the relative calibration of uv , ir , and h@xmath6 as sf indicators as a function of galaxy physical conditions . in future works , we will also compare the parameters describing the properties of sf regions from the integrated measurements with resolved studies of their stellar populations ( from ground - based and bianchi s hst programs data ) .
we will estimate ages and masses from deeper _ galex _ images to test the limit of our current detection threshold .
ybk was supported by the korea research foundation grant funded by the korean government(moehrd ) ( krf-2007 - 612-c00047 ) .
scr acknowledges support from the kosef through the astrophysical research center for the structure and evolution of the cosmos ( arcsec ) .
we are grateful to d. thilker for discussions about the background subtraction , to k. kuntz for a careful reading of the manuscript and useful comments , and a. tolea for providing some of the procedures he developed for his dissertation with luciana bianchi and k. kuntz .
we are also grateful to p. massey for discussions about the reddening value and for providing the revised photometry catalog .
this work is based on archival data from the nasa galaxy evolution explorer ( galex ) which is operated for nasa by the california institute of technology under nasa contract nas5 - 98034 .
the galex data presented in this paper were obtained from the multimission archive at the space telescope science institute ( mast ) .
support for mast for non - hst data is provided by the nasa office of space science via grant nag5 - 7584 and by other grants and contracts .
galex ( galaxy evolution explorer ) is a nasa small explorer , launched in april 2003 .
we gratefully acknowledge nasa s support for construction , operation , and science analysis of the galex mission , developed in cooperation with the centre national detudes spatiales of france and the korean ministry of science and technology .
fields covering the m31 disk region from the gr4 data release .
black circles are our selected @xmath0 fields and grey circles are additional @xmath0 observations , not selected .
black dots are ob type stars selected from the ground - based photometric catalog of @xcite , as described in section 4.[fig1],width=551 ] ) , blue contours using the mid threshold ( mean background + 3@xmath49 ) , and orange contours using the high threshold ( mean background + 5@xmath49 ) .
the solid circle is the fov used in our analysis and the dashed circle is the whole @xmath0 fov .
enlargements of sample regions are shown in the rectangular panels .
background images are the grey scaled fuv images of the nga_m31_mos0 " field.[fig2],width=627 ] ) , blue dots are for the mid threshold ( mean background + 3@xmath49 ) , and orange dots are for the high threshold ( mean background + 5@xmath49 ) . dashed lines are the magnitude limit of each threshold.[fig3],width=313 ] bands lower than 0.1 mag .
blue points are our selected ob stars .
the lines represent the intrinsic color as a function of t@xmath40 for luminosity classes main - sequence ( v , red ) , giant ( iii , green ) , and supergiant ( ib and ia , blue and purple ) from @xcite .
the red arrow in each panel is the direction of reddening with @xmath50 .
[ fig7],width=566 ] ) and the maximum link distance ( d@xmath39 ) between stars .
the black arrow indicates the maximum peak of this function , which defined the parameters adopted for our selection of ob associations.[fig9],width=313 ] imaging ( blue contours ) and ob associations from optical stellar photometry ( red contours ) .
black dots are the selected ob stars .
the solid ellipse has a 26 kpc de - projected radius and the dashed ellipse has a 17 kpc de - projected radius.[fig10],width=627 ] 4 , @xmath1210 , 10 - 100 , and 100 - 400 myrs , and the dashed lines are mean values over the past 400 myrs . the red line is the sfr from h@xmath6 @xcite , shown in the @xmath1210 myrs age bin which traces only the youngest stars capable of ionizing the ism .
the thick horizontal lines above vertical arrows on the @xmath124 myrs age bin and @xmath1210 myrs indicate the total sfrs obtained by adding the sfr from ir measurements @xcite to our uv - based estimates .
[ fig14],width=313 ] rrrrrrr 1 & nga_m31_mos11 & 12.204914 & 42.956084 & 3340.35 & 0.00397 & 5533 + 2 & nga_m31_mos8 & 12.164794 & 42.030947 & 2702.75 & 0.00384 & 36791 + 3 & nga_m31_mos18 & 11.403102 & 42.370208 & 3244.45 & 0.00365 & 77217 + 4 & nga_m31_mos4 & 11.253077 & 41.863775 & 3589.70 & 0.00373 & 209395 + 5 & ps_m31_mos00 & 10.683594 & 41.277741 & 3760.10 & 0.00353 & 188512 + 6 & nga_m31_mos0 & 10.173990 & 40.836523 & 6811.30 & 0.00324 & 188847 + 7 & ps_m31_mos03 & 9.957439 & 40.358496 & 1182.20 & 0.00446 & 152669 + rrrrrrrrrrrrrrrrr 1003 & 12.013985 & 42.690975 & 18.870 & 0.020 & 19.158 & 0.013 & 0.20 & 112.5 & 1.6 & 1.3/ 1.9 & 3.7e+02 & 4.4e+02/ 3.7e+02 & -99.0 & -99.0/ 1.2 & -9.9e+01 & -9.9e+01/ 3.8e+02 + 2040 & 11.951906 & 42.073174 & 19.124 & 0.025 & 19.331 & 0.016 & 0.06 & 126.0 & 4.1 & 2.9/ 5.7 & 1.4e+02 & 9.0e+01/ 2.2e+02 & 2.3 & 1.8/ 2.7 & 8.2e+01 & 8.2e+01/ 1.1e+02 + 3069 & 11.725769 & 41.968216 & 19.770 & 0.034 & 19.986 & 0.023 & 0.05 & 155.2 & 3.7 & 2.1/ 5.8 & 6.8e+01 & 5.4e+01/ 1.1e+02 & 2.1 & 1.4/ 2.7 & 4.2e+01 & 5.6e+01/ 5.6e+01 + 3071 & 11.715304 & 41.971684 & 18.494 & 0.018 & 18.420 & 0.010 & 0.13 & 328.5 & 58.7 & 49.9/ 66.5 & 1.5e+04 & 1.2e+04/ 1.9e+04 & 22.4 & 17.6/ 30.8 & 5.0e+03 & 5.0e+03/ 8.2e+03 + 3074 & 11.641906 & 41.971554 & 20.070 & 0.039 & 20.057 & 0.023 & 0.19 & 117.0 & 40.5 & 24.5/ 49.7 & 3.2e+03 & 1.4e+03/ 4.4e+03 & 10.1 & 6.8/ 17.5 & 7.4e+02 & 4.5e+02/ 1.8e+03 + 3077 & 11.618323 & 41.983509 & 18.498 & 0.019 & 18.633 & 0.012 & 0.30 & 452.2 & 7.1 & 5.7/ 8.8 & 3.3e+03 & 2.5e+03/ 4.9e+03 & 3.1 & 2.8/ 3.6 & 1.2e+03 & 1.2e+03/ 1.8e+03 + 3081 & 11.635375 & 41.991177 & 16.531 & 0.007 & 16.393 & 0.004 & 0.19 & 2072.2 & 79.4 & 76.8/ 82.0 & 2.2e+05 & 2.2e+05/
2.2e+05 & 48.0 & 45.3/ 50.4 & 1.6e+05 & 1.6e+05/ 1.6e+05 + 3082 & 11.658252 & 41.985649 & 18.267 & 0.016 & 18.131 & 0.008 & 0.26 & 229.5 & 74.8 & 68.9/ 80.6 & 6.4e+04 & 6.4e+04/ 7.6e+04 & 43.3 & 34.6/ 49.3 & 4.2e+04 & 2.8e+04/ 5.6e+04 + 3083 & 11.625877 & 41.988590 & 19.501 & 0.032 & 19.476 & 0.019 & 0.46 & 207.0 & 19.9 & 14.8/ 29.8 & 1.8e+04 & 7.8e+03/ 3.0e+04 & 6.2 & 4.8/ 8.1 & 4.0e+03 & 2.6e+03/ 6.8e+03 + 3085 & 11.648593 & 41.992626 & 20.509 & 0.055 & 20.507 & 0.033 & 0.30 & 112.5 & 28.4 & 15.2/ 47.0 & 3.6e+03 & 2.2e+03/ 6.8e+03 & 7.7 & 4.8/ 15.5 & 8.1e+02 & 3.1e+02/ 2.7e+03 + 3086 & 11.654299 & 41.999249 & 17.305 & 0.011 & 17.136 & 0.006 & 0.20 & 1055.2 & 89.3 & 85.2/ 94.1 & 1.4e+05 & 1.4e+05/ 1.4e+05 & 54.9 & 52.2/ 57.6 & 8.9e+04 & 8.9e+04/ 1.2e+05 + ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... / ... & ... & ... / ... & ... & ... / ... & ... & ... / ... + | we present a comprehensive study of star - forming ( sf ) regions in the nearest large spiral galaxy m31 .
we use @xmath0 far - uv ( 1344 - 1786 , fuv ) and near - uv ( 1771 - 2831 , nuv ) imaging to detect young massive stars and trace the recent star formation across the galaxy .
the fuv and nuv flux measurements of the sf regions , combined with ground - based data for estimating the reddening by interstellar dust from the massive stars they contain , are used to derive their ages and masses .
the @xmath0 imaging , combining deep sensitivity and coverage of the entire galaxy , provides a complete picture of the recent star formation in m31 and its variation with environment throughout the galaxy .
the fuv and nuv measurements are sensitive to detect stellar populations younger than a few hundred myrs .
we detected 894 sf regions , with size @xmath1 1600 pc@xmath2 above an average fuv flux limit of @xmath326 abmag arcsecond@xmath4 , over the whole 26 kpc ( radius ) galaxy disk .
we derive the star - formation history of m31 within this time span .
the star formation rate ( sfr ) from the youngest uv sources ( age @xmath5 10 myr ) is comparable to that derived from h@xmath6 , as expected .
we show the dependence of the results on the assumed metallicity .
when star formation detected from ir measurements of the heated dust is added to the uv - measured star formation ( from the unobscured populations ) in the recent few myrs , we find the sfr has slightly decreased in recent epochs , with a possible peak between 10 and 100 myrs , and an average value of sfr @xmath30.6 or 0.7 m@xmath7 yr@xmath8 ( for metallicity z=0.02 or 0.05 respectively ) over the last 400 myrs . |
high contrast imaging at small angles is a very useful tool to access inner regions around stars .
while instrumental limits are constantly improved by the development of exoplanet imaging and characterization techniques , such capabilities can be used in stellar astrophysics to discover new systems or put constraint on known ones . moreover
, this small - angle capability allows a significant reduction in the size of potential space telescopes aimed at detecting and characterizing exoplanets , which is a cost - effective way of building exoplanet missions .
we have used a 1.5-m subaperture of the 5.1-m hale telescope at palomar with a small inner working angle ( iwa ) phase - mask coronagraph to detect a new candidate stellar companion to the @xmath1 scuti star @xmath0 cephei .
the result is interesting for two reasons .
first , @xmath0 cephei was claimed to have an infrared excess measured by _ iras _ , and recently detected by near - infrared ground - based interferometry . such infrared excesses are generally attributed to dusty debris disks @xcite .
we show here with an image that the origin of the near - infrared excess is a second object that is likely a late - type stellar companion .
moreover , we demonstrate that the previously reported 60 @xmath5 m _ iras _ excess is attributable to an unrelated field object .
second , the potential companion was imaged at 1.1 diffraction beamwidths ( @xmath3 , with the working wavelength @xmath6 @xmath5 m , and the telescope diameter @xmath7 m ) from the star .
such a small angle detection was only possible because of our phase mask vector vortex coronagraph ( vvc ) , demonstrating that such coronagraphs can effectively provide very small iwa .
@xmath0 cephei is a bright ( @xmath8 , @xmath9 , @xmath10 ) and nearby ( @xmath11 pc , see @xcite ) f0 iv star with a temperature @xmath12 k , of the @xmath1 scuti type ( table 1 ) .
privileged targets in asteroseismologic studies @xcite , @xmath1 scutis are multi - periodic variable stars lying at the base of the cepheid instability strip , where it crosses the main sequence . their small amplitude ( 0.0011 mag ) variability is due to simultaneous radial and non - radial pulsation modes with periods between 0.25 and 5 hours . the spectral type of @xmath1 scuti stars ranges from a0 to f5 , with nominal mass and temperature around 2 @xmath13 and 7500 k , respectively . for @xmath0 cephei , @xcite derived a mass of 1.8 @xmath14 @xmath13 .
@xmath0 cephei was claimed to be detected by _ iras _ at 12 , 25 , and 60 @xmath5 as psc 22132 + 5647 .
the reported 60 @xmath5 flux density of @xmath15 jy is @xmath16 times the expected photospheric value ( the ratio of excess infrared luminosity divided by the total energy output from the photosphere was reported to be @xmath17 ) . on this basis
, @xcite , @xcite , and @xcite identified the system as hosting a bright debris disk . in @xcite , this excess
was fitted with a single blackbody giving a putative ring of dust at 62 au with a temperature of 65 k. using space velocities uvw , lithium abundance and location on a hr diagram , @xcite also estimated the age of @xmath0 cephei to be a very uncertain @xmath18 myr .
lc [ table1 ] coordinates ( hms ) & ra=22 15 02.19 dec=+57 02 36.91 + galactic coordinates ( deg ) & l=102.87 b=+00.39 + proper motion ( @xmath19 ) & @xmath1ra=0.47645 @xmath1dec=0.04999 + radial velocity ( km / s ) & -0.6 + parallax ( @xmath19 ) & 0.3886 + spectral type & f0 iv + @xmath20 mag & 4.2 + @xmath21 mag & 3.5 + age & @xmath18 myr + distance & @xmath22 pc + mass & 1.8 @xmath13 + radius & 2 @xmath23 + temperature & 7350 k in the course of an on - going survey for bright exozodiacal disks around main sequence stars , lead by one of us , the chara - fluor interferometer detected a @xmath24-band excess for @xmath0 cephei ( absil et al . 2011 , in preparation ) . one possible interpretation of such near - infrared excesses is emission from hot dust close to the star @xcite .
the grain populations usually derived from such observations are quite intriguing , as they point towards very high dust replenishment rates , high cometary activity or unlikely major collisional events .
note that due to sparse ( u , v ) coverage , chara - fluor poorly constrains the nature of the excess or its location inside the fluor @xmath25 fwhm field of view , suggesting that a high resolution image might be able to distinguish between a close companion and dust .
we observed @xmath0 cephei in the @xmath26 band on 2010 june 23 , as part of our program of adaptive optics ( ao ) high contrast imaging of nearby young bright stars at palomar .
we took 50 frames of 4.238 s each , for a total of 212.4 s on the target .
we used the palomar well - corrected subaperture ( wcs ) and the vvc , which have been both described in details earlier @xcite .
briefly , a clear , off - axis wcs is provided by a set of relay optics upstream of the palomar ao system . the off - axis relay magnifies and shifts an off - axis sub - aperture pupil onto the deformable mirror , yielding roughly 10 cm actuator spacing in the selected pupil .
this configuration reduces the aperture size from 5 m to 1.5 m , but allows operation in the extreme ao regime .
providing unprecedented image quality on a ground - based telescope ( @xmath26-band strehl ratios @xmath27 ) , this setup is ideal for use with the vvc @xcite .
the vvc is based on a transparent phase - mask , with an azimuthal phase ramp centered on the image of the central star . upon propagation to the lyot plane ,
the central starlight is redistributed outside the geometric pupil area . there , a simple diaphragm known as a lyot stop is used to reject the starlight , leaving any off - axis feature mostly unaltered in the process : the 50% off - axis throughput , which is the commonly adopted definition of the iwa , is reached at 0.9 @xmath3 for a second order vvc such as the one we used at palomar @xcite .
the vvc allowed us to detect a second object near @xmath0 cephei at 330 @xmath28 mas from the host star , which corresponds to @xmath29 times the width of the diffraction beam ( @xmath30 mas at @xmath26 ) , making it the smallest angle detection ever reported with a coronagraph in @xmath3 units .
the projected separation of the potential companion places it at @xmath31 au .
figure [ fig1 ] ( a ) shows the raw image after sky subtraction , flat fielding , bad pixel replacement and cosmic ray removal . a faint secondary point source is clearly detected on this image to the east ( @xmath32 ) . a reference star ( hd 213558 )
was also imaged right after @xmath0 cephei , with no companion present ( figure [ fig1 ] , b ) .
the reference star was chosen with a @xmath33 color and brightness as similar as possible to the target .
matching @xmath20 magnitudes is important to ensure similar ao corrections . @xmath21
magnitudes , and integration time also need to be matched to ensure a proper subtraction . also , to reduce thermal , seeing and flexure - induced variations as much as possible , the calibration star was chosen and observed as close as possible to the hour angle and elevation of the target star , and just a few minutes after the latter . finely scaling the reference star image to exactly match @xmath0 cephei s apparent magnitude , and subtracting it from @xmath0 cephei , leaves a clean detection of the second object ( figure [ fig1 ] , c ) .
the subtracted image was used to derive the astrometry and photometry with minimal contamination from the residual errors left after imperfect wavefront correction .
note that the @xmath34 detection level after subtraction is calculated to be @xmath35 at @xmath36 @xmath3 , mostly limited by residual centering errors on the vvc .
the 2mass @xmath21 magnitude of @xmath0 cephei is not very well constrained : @xmath37 .
however , @xcite give an independent and accurate ( few percent accuracy ) measurement of the optical photometry ( @xmath38 , @xmath39 , @xmath40 , @xmath41 ) which is nominal for an f0 star .
in this case , the @xmath33 color is 0.7 @xcite , and the @xmath21 magnitude extrapolated from @xmath20 should be 3.49 , very close to the 2mass value .
aperture photometry using a radius of 300 mas ( our fwhm ) yields a flux ratio of @xmath42 in the @xmath26 band .
the apparent magnitude of the potential companion is then @xmath43 .
note that we conservatively propagated the 2mass error bars , knowing that they must be pessimistic by an order of magnitude .
for an assumed distance of @xmath22 pc , the absolute @xmath26 magnitude of the companion is therefore @xmath44 .
as yet , we have no second epoch measurement available because the palomar ao system was taken offline in mid-2010 to enable the palm-3000 upgrade . the probability for the potential companion to be a background object , given its proximity to @xmath0 cephei , the galactic coordinates of the system ( in the galactic plane but well away from the bulge , see table 1 ) and the number of objects present with @xmath45 in a series of regions of radii 1 - 10 deg centered around the star ( measured population in this zone of the sky ) , is very small , @xmath46 .
thus it is likely that the candidate companion is physically associated with @xmath0 cephei .
however , astrometry and/or spectroscopy will be needed to confirm this .
lc spectral type & k8-m2 + @xmath21 mag & 7.8 + age & @xmath18 myr + distance & @xmath478.6 au from primary + mass & 0.5 @xmath13 + radius & @xmath48 @xmath23 + temperature & 3650 k
assuming the candidate companion is bound to @xmath0 cephei , we will derive its physical properties based on an evolutionary model .
we will also discuss the consequence of this discovery on the interpretation of near - infrared and far - infrared excesses .
an interesting question one could then ask is why this second source has largely remained undetected so far . according to the bcah98 evolutionary model @xcite , and following the age estimation in @xcite
the candidate companion would have a mass of @xmath49 @xmath13 , and a temperature of 3650 k ( table 2 ) .
the stellar type of the candidate companion is thus likely to be at the transition between late - type k stars and early m stars .
this result is fairly independent of the age between 200 myr up to well above 1 gyr , making this mass determination quite robust even with big error bars ( figure [ baraffe ] ) . .
our photometry with conservative error bars is overplotted .
[ baraffe ] ] the interferometrically detected near - infrared excess is likely due to the companion , and not hot dust .
however , the candidate companion would only contribute @xmath50 additional flux at 60 @xmath5 .
it thus can not account for the large excess detected in the _ iras _ beam , which was interpreted as thermal emission from cold dust at a separation of @xmath51 au @xcite . in this picture
, the cold dust belt would thus likely be circumbinary which is not unprecedented , see for instance the case of gg tau @xcite .
however , as mentioned earlier , @xmath0 cephei lies away from the galactic bulge but still on the galactic plane ( @xmath52 deg , see table 1 ) , which raises the possibility of source confusion given the large beam size of _ iras _
( @xmath53 at 60 @xmath5 ) . to confirm the excess emission ,
the star was observed with spitzer / mips on 2009 feb 20 ( pi : george rieke ) .
the mips 70 @xmath5 bcd ( basic calibrated data ) image retrieved from the _ spitzer _ heritage archive ( see figure [ fig2 ] ) shows only a very weak detection at the stellar position , and a much stronger far - infrared source located @xmath54 to its nne .
photometry in a @xmath55 circular aperture centered on the stellar position using aperture corrections as given in the mips instrument handbook finds a stellar 70 @xmath5 flux density of @xmath56 80 mjy .
this value is consistent with the expected stellar photospheric emission ( as extrapolated from the _ iras _ 12 and 25 @xmath5 flux densities ) .
we therefore conclude that the _ iras _ 60 @xmath5 measurement was very likely confused by the bright adjacent source . indeed , the reported position of the _ iras _ source psc 22132 + 5647 is shifted @xmath57 e from the known position of @xmath0 cephei at the time of the measurement .
thus , it was never right to associate @xmath0 cephei and psc 22132 + 5647 , which is likely a virtual combination of two real sources : @xmath0 cephei , detected at 12 and 25 @xmath5 by _ iras _ , and the far - infrared source to its nne , dominating at 60 @xmath5 . in conclusion ,
@xmath0 cephei lacks any far - infrared excess and thus does not possess a debris disk sculpted by the putative companion .
bcd ( basic calibrated data ) image of @xmath0 cephei retrieved from the spitzer heritage archive shows only a very weak detection at the stellar position ( green circle ) , and a much stronger far - infrared source located @xmath54 to its nne .
[ fig2 ] ] using speckle interferometry in the visible , @xcite , while classifying @xmath0 cephei as a spectroscopic binary ( that companion would be much closer in ) , did not report the detection of an outer companion down to the detectivity limit of the technique used , which is @xmath58 , or a flux ratio of @xmath59 . note that for a late k / early m type , the @xmath33 color is @xmath60 mag @xcite .
@xmath0 cephei s own @xmath33 color should be only 0.7 mag @xcite .
given our measured flux ratio of @xmath42 in the @xmath26 band , the contrast will be @xmath61 mag greater at @xmath20 band , i.e. an overall flux ratio of @xmath62 . thus it is not at all surprising that optical measurements ( speckle interferometry or spectra ) would fail to show a double - lined binary , or that _ hipparcos _ would fail to see the companion as a separate source .
a 0.5 @xmath13 companion on a circular 8.6 au orbit would cause an @xmath63 mas reflex motion of @xmath0 cephei over a period of @xmath64 years .
such a large astrometric signature would have been easily detected by _
hipparcos _ , which observed @xmath0 cephei more than 100 times over its 3.5 year mission lifetime with a typical accuracy of 0.5 mas per observation @xcite .
@xmath0 cephei is however classified as a single star in the _ hipparcos _ catalog .
this either suggests that our detected object is not physically associated , which seems unlikely based on a statistical argument ( see section [ probability ] ) , or that the orbital semi - major axis is actually much larger than the apparent separation of @xmath4 au , resulting in a much larger period than the _ hipparcos _ mission duration .
a first possibility is that the companion is on a significantly eccentric orbit .
while no proper motion variation was observed between short - term _ hipparcos _ and long - term tycho observations , some proper motion acceleration was detected by _
hipparcos _ around @xmath0 cephei @xcite .
according to the same authors , such behavior not only suggests that @xmath0 cephei is an astrometric binary , but also that the system orbital period should be typically less than @xmath65 years , unless there is significant eccentricity in the system . following this interpretation , our measured minimum orbital period of @xmath64 years supports a substantial eccentricity .
additionally , the system may be seen under a high inclination , resulting in a strongly elliptical apparent orbit whatever its true eccentricity .
evidence for such a significant departure from pole - on inclination comes from the observed @xmath66 of 91 km / s @xcite , which is to be compared with a break - up velocity around 400km / s for a sub - giant star like @xmath0 cephei .
the candidate companion did not leave any detected radial velocity ( rv ) signature either . indeed , @xcite report a very rich spectrum of pulsating modes and beats for @xmath0 cephei , consistent with a @xmath1 scuti ( see also @xcite ) .
the stellar pulsations induce rv signatures at the @xmath67 km / s level , making it very difficult to detect faint companions using rv measurements . indeed , assuming a semi - major axis @xmath68 au , @xmath69
@xmath13 , @xmath70 @xmath13 , we get an rv amplitude of @xmath71 km / s , with a period of @xmath64 years .
this letter reports the detection of a candidate dim companion to @xmath0 cephei , a @xmath1 scuti star .
this is the first image of a candidate companion , seen at 330 @xmath28 mas separation , only 1.1 resolution elements from the central host star ( @xmath4 au ) . if physically associated , the companion has a mass of @xmath49 @xmath13 , and a temperature of 3650 k , likely placing it at the transition between late - type k stars and early m stars .
the interferometrically detected near - infrared excess is likely due to the companion , and not hot dust , while the previously reported _ iras _ 60 @xmath5
m excess was due to source confusion .
finally , the most plausible reason why this candidate companion remained hidden so far is that _ hipparcos _ failed to detect it because of its long period , highly eccentric orbit , while the stellar pulsations of the @xmath1 scuti primary precluded an rv detection .
this detection was made possible by state - of - the - art high contrast imaging techniques ( wavefront control and vortex coronagraphy ) envisioned for upcoming next - generation ground - based extreme ao systems and future space - based coronagraphs . we have demonstrated a @xmath34 detection capability of the order of @xmath35 at @xmath72 , which is sufficient to directly image secondary stars and brown dwarfs to an order of magnitude fainter than speckle interferometry . at @xmath73 ,
we routinely reach @xmath74 contrast levels , and with careful calibration @xmath75 @xcite , two to three orders of magnitude better than current interferometric techniques .
extreme ao coupled to vector vortex coronagraphy has thus great potential for the characterization of spectroscopic binary and brown dwarf orbits , especially with a larger telescope .
it can also dramatically improve imaging capabilities even with rather small telescopes .
such techniques thus open the door to the direct imaging of spectroscopic binary companions , as well as brown dwarfs and exoplanets .
this work was carried out at the european southern observatory ( eso ) site of vitacura ( santiago , chile ) , and the jet propulsion laboratory ( jpl ) , california institute of technology ( caltech ) , under contract with the national aeronautics and space administration ( nasa ) .
the data presented in this letter are based on observations obtained at the hale telescope , palomar observatory , as part of a continuing collaboration between caltech , nasa / jpl , and cornell university .
this work is also based ( in part ) on observations made with the _
spitzer space telescope _
, which is operated by the jpl , caltech , under a contract with nasa .
this research has made use of the nasa / ipac / nexsci star and exoplanet database , which is operated by the jpl , caltech , under contract with nasa , and nasa s astrophysics data system and of the simbad database , operated at cds ( strasbourg , france ) . | using a vector vortex coronagraph behind the 1.5-m well - corrected subaperture ( wcs ) at palomar , we detected a second object very close to @xmath0 cephei , a @xmath1 scuti f0 iv star .
the candidate companion , @xmath2 times fainter than @xmath0 cephei , if physically associated , is a late - type k or early m star , and lies at an angular separation of 330 mas , or 1.1 @xmath3 for the wcs , making it the smallest angle detection ever realized with a coronagraph in terms of @xmath3 units .
the projected separation of the putative companion is @xmath4 au , most likely on a highly eccentric orbit .
the recently detected near - infrared excess is thus likely not due to hot dust .
moreover , we also show that the previously reported _ iras _ 60
@xmath5 m excess was due to source confusion on the galactic plane . |
let s consider the hamiltonian @xmath111 and rewrite it in the following way @xmath112 where we compact the position and momentum operators in a vector formalism , i.e. @xmath113 and @xmath114 .
moreover we have an @xmath115 matrix @xmath116 with elements @xmath117 , where the connection matrix @xmath118 has the following form @xmath119 since @xmath116 is symmetric , it can be diagonalized by an orthogonal transformation @xmath23 , i.e. @xmath120 where @xmath59 is a the diagonal matrix containing the eigenvalues @xmath121 of @xmath116 . thus defining the new variables @xmath122 and @xmath123 , we can rewrite the hamiltonian as @xmath124\ ] ] where the eigenfrequencies are @xmath125,and again @xmath126 and @xmath127 .
thus we passed from a chain environment into the equivalent star model .
the starting point is the total hamiltonian @xmath128 after the diagonalization of the environment .
since @xmath129 is also quadratic in position and momentum operators , it can again be written as @xmath130 where , contrary to the notation of last section , we have @xmath131 and @xmath132 .
the @xmath133 matrix @xmath134 has elements @xmath135 for @xmath136 , @xmath137 and @xmath138 .
we can again perform a diagonalization of @xmath134 through an orthogonal matrix @xmath139 , i.e. @xmath140 , and upon defining the new system - environment normal modes @xmath141 and @xmath142 , we rewrite the full hamiltonian as @xmath143\ ] ] where @xmath144 are the eigenvalues of @xmath134 contained in the diagonal matrix @xmath60 . in this picture , the time evolution for each normal mode is trivial , @xmath145 now , remembering the new variables are connected to the old ones through the orthogonal transformation @xmath139 at any time @xmath146 , we easily get @xmath147\\ & p_i(t)=\sum_{j=1}^{n+1}\bigl[\mathbf{b}^{pq}_{ij}(t)q_j(0)+\mathbf{b}^{pp}_{ij}(t)p_j(0)\bigl ] \end{split}\ ] ] where @xmath148 where @xmath149 is a diagonal matrix with @xmath150 , @xmath151 is also diagonal such that @xmath152 and @xmath153 .
provide the time evolution of the position and momentum operators for the system and the normal modes when their form at time zero is given .
the simulations for @xmath80 have been done for two - mode vacuum squeezed states between system and ancilla , with a squeezing parameter @xmath70 unless otherwise stated . as shown in fig.3 in the main paper ( extra black dot ) , increasing this parameter yields higher values for the measure . however , there are not qualitative changes in the behavior of @xmath80 ( see fig .
[ figapp2 ] ) .
the main issue with this measure is that for low system - bath couplings it is inconclusive ( i.e. @xmath155 ) , but this can be easily fixed by increasing @xmath20 .
typical time steps for integration of this measure @xmath156 ( e.g. for @xmath157 in fig.3 ) have been used , which is more than enough to resolve the positive slopes of the dynamics for @xmath158 ( see eq .
( 4 ) in main work ) .
the simulations for @xmath78 have been performed between pairs of one - mode vacuum squeezed states . in fig.3 of main work ,
one of them having squeezing parameter @xmath159 and phase - space angle @xmath160 , the other state having @xmath161 $ ] , @xmath162 and @xmath163 $ ] with @xmath164 ( same for fig .
4c ; for figs .
4d and 5 we have used @xmath165 $ ] with @xmath166 ) . the bold ( blue ) point in fig .
3 in the main paper was obtained by increasing @xmath167 $ ] , @xmath162 and @xmath163 $ ] with @xmath164 . instead , for figures 4a , b we have used a much more thorough scan with @xmath168 , @xmath160 , @xmath169 $ ] , @xmath170 and @xmath171 $ ] with @xmath172 .
we stress that the behavior against temperature is quite sensitive to the range of squeezings used and needs to be scanned intensively , because the optimal pair for this measure depends on the bath temperature ( mostly for low t ) .
we show an example in fig .
[ figapp2 ] .
the behavior of @xmath78 under the non - ohmic spectral density @xmath173 was shown in last figure of main work , fig .
this is a spectral density which pivots around @xmath77 and therefore keeps the coupling strength ( @xmath174 ) fixed , while at the same time the extreme points ( @xmath12 and @xmath86 ) it differs in values .
however , choosing for example @xmath175 ( which keeps @xmath176 constant but differs at @xmath177 for different @xmath96 ) would change the value of @xmath78 , as shown in fig .
[ figapp3 ] . with @xmath178 , @xmath179 , @xmath180 and @xmath181 ( black for @xmath100 to gray for @xmath182 .
bottom : blp nm versus @xmath96 evaluated at @xmath183 ( sparse dashed , tight dashed , continuous ) for the spectral density above .
[ figapp3 ] ] with @xmath184^s \theta(\omega_r-\omega)$ ] , @xmath185 , @xmath180 and @xmath186 ( continuous , dashed , dotted ) .
we have drawn the spectral density to guide the eye .
inset : blp nm versus @xmath96 evaluated at @xmath187 .
[ figapp4 ] ] deeper differences are found considering different power law spectra densities with constraint coupling strengths at both @xmath12 and @xmath86 , as results from @xmath188^s\theta(\omega_r-\omega).\ ] ] in fig .
[ figapp4 ] it can be seen that the behavior of @xmath78 is again different .
it does actually decrease in the super - ohmic case ( larger @xmath96 values ) .
we notice that in the latter two cases the spectral density at the system frequency @xmath189 differs in value when changing @xmath96 , unlike the case discussed in the main work .
these examples show that normalization has to be carefully taken into account when drawing general conclusions about non - markovian aspects of dissipation in presence of different power law in the spectral density . | non - homogeneous chain environments ( e.g. segmented ion traps ) are investigated through an exact diagonalization approach .
different spectral densities , including band - gaps , can be engineered to separately assess memory effects .
environment non - markovianity is quantified with recently introduced measures of information flow - back and non - divisibility of the system dynamical map . by sweeping the bath spectrum via tuning of the system frequency
we show strongest memory effect at band - gap edges and provide an interpretation based on energy flow between system and environment .
a system weakly coupled to a stiff chain ensures a markovian dynamics , while the size of the environment as well as the local density of modes are not substantial factors .
we show an opposite effect when increasing the temperature inside or outside the spectral band - gap .
further , non - markovianity arises for larger ( negative and positive ) powers of algebraic spectral densities , being the ohmic case not always the most markovian one . _
introduction_. achievements in preparation , control and measurements of quantum systems require a deep understanding of the mechanism of interaction between a given open quantum system and the surrounding environment @xcite . from a theoretical point of view ,
popular approaches , e.g. derivations of master or quantum langevin equations @xcite , are based on the assumption of an infinite heat bath with some given spectral density @xmath0 embedding all information about the real couplings and frequencies in the complex environment , and the structure of the coupling to the system .
typical approximations to simplify the treatment , such as negligible memory effects ( markovian approximation ) , system time coarse graining , weak system - bath coupling , large frequency cut - off , ohmic spectral density , drastically constrain the possible frequency dependence of @xmath0 .
although simplified spectral densities allow an analytical treatment , important deviations from these simple instances are common in several systems like , for instance , electric circuits , acoustic polarons in metals and semi - conductors , radiation damping of charged particles @xcite , photonic crystals @xcite , but also in ion traps suffering electric field noise @xcite , micro - mechanical oscillators @xcite or polarized photons for broad frequency spectra @xcite . memory effects and deviations from markovian dynamics have been widely explored considering the time dependence of master equation - coefficients and deviations from exponential decays @xcite but only in the last few years several approaches have been proposed to systematically distinguish markovian evolution in terms of properties of the dynamical map @xcite .
different measures allow to quantify non - markovianity ( nm ) in terms of deviations from the lindblad form of the generator of the master equation @xcite , flow - back of information from the environment @xcite , and entanglement decay with an ancilla @xcite , to mention some of them .
the recent endeavor to better characterize memory effects in open systems not only aims to a deeper understanding of dissipative dynamics in physical , biological and chemical systems , but is also explored as a resource in quantum technologies @xcite . the aim of this work is to identify non - markovian effects originating in the structure of the system and bath couplings as well as in the distribution of energies , as given by the form of the spectral density @xmath0 . to this end
we consider a microscopic model given by an inhomogeneous harmonic chain , avoiding the limitations of approximate approaches .
finite models of coupled oscillators have been used to assess entanglement dynamics @xcite and its generation when attaching ions to a chain @xcite and can also provide an insightful test - bed to establish the regimes of validity of approximated master equations @xcite .
the case of an oscillator attached to a homogeneous chain was already studied by rubin to determine the statistical properties of crystals with defects : this configuration leads to a ohmic dissipation ( thus markovian , at least for large temperature ) @xcite . moving to non - homogeneous chain configurations , we inquire the origin of nm to distinguish among several independent features quantifying separately different sources of nm .
when focusing on periodic systems ( e.g. dimers ) , we can engineer spectral densities with finite band - gaps , like in semiconductors or photonic crystals @xcite . for suitable couplings
we show that the system is actually influenced by the resonant portion of the environment .
memory effects @xcite are then evaluated by sweeping the spectral density for a structured bath allowing us to show the effects of the local form of the spectrum . before introducing the model ,
we point out that experimental implementation of a tunable chain of oscillators can be obtained through recent progress in segmented paul traps @xcite also allowing tunability of ions couplings and onsite potentials .
other possible setups are based on photonic crystal nanocavities , microtoroid resonators ( see @xcite for a review ) or optomechanic resonators @xcite .
furthermore , correlations spectra of the system can be measured to gain insight on the spectral density induced by the rest of the chain ( see @xcite ) .
the non - homogeneous harmonic chain we consider in this work represents then a structured and controllable reservoir amenable to experimental realization .
_ the system : non - homogeneous chain_. we consider an open quantum system consisting of an harmonic oscillator @xmath1 , where @xmath2 and @xmath3 are the system momentum and position operators , interacting with an environment ( @xmath4 ) that consists of an harmonic chain of @xmath5 elements interacting through a spring - like coupling non - homogeneous along the chain .
a particularly dramatic deviation from the ohmic spectrum obtained with a homogeneous chain ( rubin model @xcite ) is found when considering a periodic configuration ( dimer ) of identical oscillators with alternate values of couplings @xmath6 and @xmath7 ( see fig .
[ fig1 ] ) @xmath8 with @xmath9 ( @xmath10 ) and @xmath11 @xcite . for any @xmath12 and @xmath13
we build a band gap model , characterized by a frequency spectrum distributed in an acoustic and an optical band separated by a finite gap @xcite . and
@xmath14 ( black line ) , gapped spectral density with @xmath15 , @xmath16 and @xmath17 ( red line ) , and increasing coupling density with @xmath18 and @xmath14 ( blue line ) .
bath frequency @xmath19 ( in unspecified , but fixed , frequency units ) in all examples .
schematic for the three plotted examples in the corresponding colors ( top left ) and of a star configuration ( top right ) also represented .
[ fig1 ] ] the system is attached to the first element of the e chain and @xmath20 is a coupling constant with an interaction term of the form @xmath21 .
we consider an initial factorized state between system squeezed vacuum and a thermal bath at temperature @xmath22 .
the chain configuration for real oscillators is mapped by diagonalization of the environment ( through an orthogonal transformation @xmath23 ) into a star configuration of independent oscillators ( bath s eigenmodes ) of frequency @xmath24 interacting with the system with strength @xmath25 ( see fig . 1 and @xcite ) .
_ generalized langevin equation_. the reduced dynamics of the system is governed by a langevin equation , typically derived starting from a star environment @xcite , @xmath26 where @xmath27 is langevin forcing of the system @xcite and @xmath28 is the damping kernel accounting for dissipation and memory effects .
the spectral density can be obtained from the damping kernel as @xmath29 . for a finite number @xmath5 of normal modes ,
the dissipative dynamics suffers recurrence for times @xmath30 ( related to reflection into the system of the fastest signals traveling along the chain ) , whose value depends on the number of modes @xmath5 and on the frequency spectrum @xcite .
if we look at times @xmath31 the spectral density can be described by a smooth function of frequency .
the form of this function can be directly related to the dynamical properties of the system .
for instance , a linear ohmic spectral density , i.e. @xmath32 , leads to a markovian equation with time - independent friction kernel @xmath33 . in the rubin model
it is possible to show that @xmath34 with @xmath35 the highest ( cutoff ) frequency in the spectrum of normal modes .
other non - homogeneous chain examples are shown in fig .
[ fig1 ] .
_ resonance conditions_. in order to investigate the role of the spectral density s shape , we first need to know to what extent the system is affected by different bath eigenmodes depending on their relative detuning .
to do so we compare and bandwidth @xmath36 , for a band gap density with @xmath15 , @xmath16 and @xmath37 at temperature @xmath38 .
integration performed up to @xmath39 for @xmath40 oscillators .
( b ) system excitation number at @xmath41 as a function of the proper frequency @xmath42 for a band gap model with @xmath15 , @xmath16 and @xmath37 ( @xmath0 as shaded area ) at @xmath38 .
system - bath coupling @xmath43 ( red solid ) and @xmath44 ( blue dashed).[fig2 ] ] the state of the system @xmath45 with @xmath46 obtained by allowing the system to interact only with those normal modes whose frequencies lie within a range @xmath36 to system frequency @xmath42 , i.e. such that @xmath47 .
the time evolution is obtained by a full diagonalization of @xmath48 @xcite .
deviation between @xmath45 and @xmath46 can be witnessed by the uhlmann fidelity measure @xcite , @xmath49 .
this quantity averaged in time , @xmath50 , is shown in fig .
2a for a band gap model with system frequency larger than the optical band , @xmath51 .
here and in the following we choose @xmath52 but big enough to explore the dissipative dynamics .
if we take the value @xmath53 as a guide for the eye , a weakly coupled oscillator ( @xmath54 ) interacts only with a small vicinity of modes in the nearest band ( right / optical band in fig . [ fig1 ] ) , while already for @xmath55 we need to take into account modes in the furthest band ( left / acoustic band ) .
this rough estimation allows to appreciate that for weak couplings @xmath56 a reduced number of resonant bath modes suffice to determine the system evolution . the effect of the environment band - gap is clearly shown by looking at the system excitation ( average energy normalized by its frequency ) @xmath57 at @xmath58 by varying the system frequency @xmath42 to sweep the spectral density ( fig . [
fig2]b ) .
energy is dissipated continuously when @xmath42 is resonant with the bath while it can not propagate into the chain for @xmath42 within the band - gap ( leading to oscillatory behavior and the formation of bound dressed states ) @xcite .
_ non - markovian dynamics_. among the different quantifiers of nm appeared recently in the literature we consider hereafter the breuer - laine - piilo ( blp ) @xcite and rivas - huelga - plenio ( rhp ) @xcite measures .
the first ( blp ) gives an interpretation of nm in terms of a back - flow of information from the environment into the system .
its definition exploits the contractivity property of the quantum trace distance @xmath59 under completely positive and trace preserving maps @xcite . for continuous variable systems , and within gaussian states ,
the definition has been extended by substituting the trace distance with the fidelity @xmath60 @xcite , and the associated measure for the degree of nm reads @xmath61 where the maximization is taken among all pairs of gaussian states @xmath62 and integration is performed up to @xmath63 .
the rhp criterion witnesses the non - divisibility of the dynamical map by preparing the system in an entangled state with an ancilla and evaluating the non - monotonic time evolution of the entanglement .
the measure reads @xcite @xmath64 where @xmath65 denotes a proper system - ancilla entanglement measure , such as logarithmic negativity . we should stress that when eq .
gives zero , the map can be either divisible or non - divisible . for details on the numerical evaluation of these measures
see @xcite . in fig .
[ fig3 ] we show both nm measures for a dimer chain environment as a function of the system frequency @xmath42 for temperature @xmath38 .
we find that local nm maxima are present at the edges of the band - gap spectral density for both measures , while rapidly decreasing both inside the band and the gaps of the spectral density .
sharp changes in frequency lead to long times in the fourier transform , and are responsible for broad noise and dissipation kernels , for long time bath correlations @xcite and for pronounced nm signatures , fig .
[ fig3 ] .
full markovian behavior is achieved only inside the band where @xmath66 . even if the null value of the rhp measure is inconclusive , we mention that increasing the system - bath coupling @xmath20 , it raises to finite values , yielding a shape very much in agreement with the blp measure .
this is one the few comparisons of two different nm measures in the literature @xcite . , @xmath16 , @xmath67 , @xmath68 as a function of @xmath42 , for @xmath38 .
the red ( continuous ) curve represents the spectral density used in arbitrary units .
the extra dots are evaluated at @xmath69 increasing the two - mode squeezing parameter from @xmath70 to @xmath71 ( for rhp nm ) and increasing the squeezing parameter range for the blp nm maximization ( from @xmath72 to @xmath73 ) @xcite .
[ fig3 ] ] we also notice that in correspondence with the edges of the spectral density @xmath0 , where @xmath74 is larger , we find a higher density of modes . to exclude its connection with nm effects
, we artificially engineered the spectral density to obtain a constant density of modes throughout the spectrum .
this allows us to establish that the enhancement of nm at the edges is actually not related to the normal mode density of states .
the origin of nm is elsewhere .
interaction between system and environment ( @xmath75 ) leads to energy exchange .
when the system is in resonance with a band of the spectral density ( optical or acoustical ) , energy is exchanged with many normal modes . in the real chain picture
this corresponds to energy allowed to travel along the chain .
this leads to an ever increasing indistinguishability of the states in eq .
[ nmblp ] ( all loosing energy and relaxing ) and thus to a low value of nm . on the other hand , at the band edges and in the band - gaps
, the energy lost by the system can not travel freely along the chain , but bounces back and forth from the first elements of the chain to the system .
this implies a periodical increase / decrease of distinguishability of the states whose fidelity we are integrating , hence we witness a higher nm value .
the higher the detuning ( deeper in the band - gap ) , the less excitations / energy are exchanged , resulting in a diminishing value for the blp nm .
this result is in accordance with ref .
@xcite , where they study a generic bosonic / fermionic reservoir , and show that band - gaps generate localized modes ( thus dissipationless oscillatory behavior ) plus nonexponential decays ( identified there with nm ) . according to ref .
@xcite , fig .
[ fig3 ] provides a quantitative measure of nm at band - gap edges as flow back of information .
_ what makes a bath markovian ?
_ besides the resonance conditions and the back reflection of information / energy when the system is out of resonance with the bath , what are the main factors leading to more markovian dynamics ?
we have tried to answer this question in a fivefold approach , namely , with respect to the bath s size , its temperature , the width of its spectral density , its shape ( sub - ohmic , ohmic , super - ohmic ) and the strength of its coupling to the system . for 50 oscillators , @xmath15 , @xmath16 , @xmath67 and @xmath37 , for temperatures @xmath76
$ ] and @xmath77 ( the limits of the acoustic and optical bands are obvious from the discontinuity of @xmath78 ) .
( b ) temperature dependence of @xmath78 for @xmath79 in black ( gray ) .
( c ) @xmath78 ( dashed ) and @xmath80 ( continuous ) at @xmath81 for an homogeneous 50 oscillators chain , @xmath82 and @xmath37 .
( d ) no - gap ( star - configuration ) spectrum @xmath83 with 40 oscillators , @xmath84 , @xmath85 and varying @xmath86 .
( inset shows @xmath0 for three values of @xmath86 . )
[ fig4 ] ] it is known that at low temperatures , markovian approximations are generally not valid @xcite and non - exponential decays of the correlation functions do arise @xcite . in fig .
[ fig4]a we show that the back - flow of information @xmath78 is similarly present only at low temperatures ( @xmath87 ) when the system is dissipating within a band of the spectral density .
surprisingly , when we move to a non - resonant configuration , with the frequency of the system in the band - gap , nm is found to increase with temperature , fig .
[ fig4]b .
when increasing temperature to higher values a linear increase of nm is observed within the gap , while in the band @xmath78 tends to vanish .
most prominent nm effects are expected in the strong coupling regime between system and bath @xcite : indeed reduction to markovian dynamics is provided by a decrease of the coupling , which reduces nm by two orders of magnitude by one order of magnitude decrease in coupling . for the @xmath80
we see that it tends to zero very fast ( remember that a zero value is inconclusive ) fig .
[ fig4]c .
the importance of frequency cut - off in the spectral density is relevant not only to avoid unphysical divergences but also when discussing nm in terms of relevant time scales .
interestingly , a chain environment allows to engineer the value of the frequency cut - off of an ohmic environment by increasing the coupling strengths within the chain . as shown in fig .
[ fig4]d an increase of the width of the spectrum ( stiff chains ) allows for a monotonic flow of information into the environment leading to a decrease of nm .
we also checked the dependence of nm with the bath size , seeing none ( except for the role of the recurrence time ) .
finally we built an artificial star model ( fig .
[ fig1 ] ) with equally spaced frequencies and couplings @xmath88 in order to isolate the effects of different algebraic behavior of the spectrum , looking , as common in the literature , at deviations from the ohmic case . in fig .
[ fig5 ] we show @xmath78 for algebraic spectral densities @xmath89 with fixed sharp cutoffs at @xmath90 .
we consider both positive and negative algebraic forms @xcite and observe an increase of nm in both cases when departing from @xmath91 at temperatures @xmath92 ( @xmath93 not shown ) . for the star - configuration density @xmath94 ( @xmath95 is the step function ) as a function of @xmath96 at @xmath97 , for temperatures @xmath98 .
[ fig5 ] ] surprisingly , the lowest nm is achieved for @xmath99 at zero temperature , instead of @xmath100 ( ohmic case ) . for higher temperatures the lowest value is achieved for @xmath100 or a bit higher . also , a bath with @xmath101 is more non - markovian than one with @xmath102 .
we also stress that different scaling and constraints of power functions densities lead to significant influence on memory effects @xcite . _ summary and conclusions . _
we have considered different ( non - phenomenological ) spectral densities attainable by tuning non - homogeneous oscillators chains that can be implemented in segmented ion chains traps and also with nano - oscillators . a system attached at one extreme of the chain dissipates in this finite bath and exhibits memory effects .
the importance of our analysis is the separate assessment , without approximations , of environment features that can be microscopically engineered and the quantitative comparison of two nm measures @xcite to establish their influence in retaining environment memory effects .
the possibility to tune the chain in a dimer configuration allows us to assess the influence of band - gaps and to obtain a more detailed picture on the origin of markovianity in relation to specific features of the spectral density .
the main role describing the behavior of nm is played by a resonance condition : if the system is resonant with the normal modes of the bath , energy transfer along the chain is allowed and therefore information and energy flow irreversibly from system to bath leading to a markovian dynamics .
the largest flow back of information is found at the edges of the gaps where the energy bounces between the system and bath . a high frequency cutoff and weak coupling , the first obtained by increasing the stiffness of the chain and the latter by decreasing the system - bath coupling , are major factors for ensuring markovian dynamics .
indeed , nm effect of strong coupling or frequency cut - off , have already been discussed in the literature and we find consistent results .
a relevant point when dealing with finite systems is that actually neither the size of the bath is important ( only matters in limiting the recurrence time ) nor the local density of modes has any significant influence on memory effects .
on the other hand , unexpected results have been found when increasing temperature leading to opposite behavior inside and outside the band with a ( linear in t ) nm increase in the band - gap while inside the band memory effects become negligible for temperatures larger than the environment frequencies .
finally we have quantified nm when departing from the ohmic case , considering positive and negative algebraic spectral densities @xmath103 for @xmath104 ( being @xmath105 a recently measured value @xcite ) showing an enhancement of memory effects for both positive and negative values of @xmath96 , with the ohmic case ( s=1 ) not being necessarily the most markovian one . financial support from micinn , mineco , csic , eu commission and feder funding under grants fis2007 - 60327 ( fisicos ) , fis2011 - 23526 ( tiqs ) , post - doctoral jae program and cost action mp1209 is acknowledged . 10 u. weiss , _ quantum dissipative systems _ world scientific publishing , 2008 ) ; h .-
p .
breuer and f. petruccione , _ the theory of open quantum systems _
( oxford univ . press , 2007 ) ; gardiner and p. zoller , _ quantum noise _ , ( springer , berlin 2004 ) p. hanggi and g .- l .
ingold , chaos * 15 * , 026105 ( 2005 ) . s. john and t. quang ,
phys .
rev .
a * 50 * , 1764 ( 1994 ) .
h .-
t .
tan , w .- m .
zhang , and g .- x .
li , phys .
rev .
a * 83 * , 062310 ( 2011 ) .
j. prior , i. de vega , a. w. chin , s. f. huelga , m. b. plenio , phys .
rev .
a * 87 * , 013428 ( 2013 ) .
ulrich hoeppe _ et .
al . _ ,
phys .
rev .
lett . * 108 * , 043603 ( 2012 ) .
d. a. hite et al . , phys . rev .
lett . *
109 * , 103001 ( 2012 ) .
s. groeblacher , a. trubarov , n. prigge , m. aspelmeyer and j. eisert , arxiv:1305.6942 ( 2013 ) .
bi - heng liu , li li , yun - feng huang , chuan - feng li , guang - can guo , elsi - mari laine , heinz - peter breuer , and jyrki piilo , nature physics * 7 * , 931 ( 2011 ) w.m .
zhang , p. y. lo , h. n. xiong , m. w. y. tu and f.nori phys .
rev .
lett . * 109 * , 170402 ( 2012 ) .
f. haake and r. reibold , phys .
rev .
a * 32 * , 2462 ( 1985 ) . b. l. hu , j. p. paz , and y. zhang , phys .
rev .
d * 45 * , 2843 ( 1992 ) .
m. m. wolf , j. eisert , t. s. cubitt , and j. i. cirac , phys .
rev .
lett .
* 101 * , 150402 ( 2008 ) .
h .-
p .
breuer , e .-
m .
laine , and j. piilo , phys .
rev .
lett . * 103 * , 210401 ( 2009 ) .
a. rivas , s. f. huelga , and m. b. plenio , phys .
rev .
lett . * 105 * , 050403 ( 2010 ) .
chruciski d. and kossakowski phys .
rev .
lett . * 104 * , 070406 ( 2010 ) .
b. bellomo , r. lo franco , and g. compagno , phys .
rev .
lett .
99 , 160502 ( 2007 ) ; chin , a. w. , huelga , s. f. , and plenio , m. b. phys .
rev .
lett .
109 , 233601 ( 2012 ) ; b. bylicka , d. chruciski , s. maniscalco , arxiv:1301.2585 m. b. plenio , j. hartley , and j. eisert , n. jour . phys . * 6 * , 36 ( 2004 ) . a. wolf , g. de chiara , e. kajari , e. lutz , and g. morigi , eur . lett . * 95 * , 60008 ( 2011 ) .
a. rivas , a. d. k. plato , s. f. huelga , and m. b. plenio , n. jour .
phys .
* 12 * , 113032 ( 2010 ) .
r. j. rubin , phys . rev . * 131 * , 964 ( 1963 ) , and references therein . n. w. ashcroft and n. d. mermin ,
_ solid state physics _
( saunders college , 1976 ) .
a. walther _
et al . _ ,
phys .
rev .
lett . * 109 * , 080501 ( 2012 ) .
r. bowler _
et al . _ ,
phys .
rev .
lett .
* 109 * , 080502 ( 2012 ) .
m. j. hartmann , f. g. s. l. brandao and m. b. plenio , laser & photon .
rev .
, 130 ( 2008 ) .
m. aspelmeyer , s. grblacher , k. hammerer , and n. kiesel , j. opt .
soc . am .
b * 27 * , a189a197 ( 2010 ) .
translation invariance of the environment is recovered by imposing an extra local potential to the first and last masses . when @xmath106 and @xmath107 , in the limit @xmath108 a rubin chain is recovered @xcite .
a rough estimate of the recurrence time can be given as the time it takes for the fastest phonon to travel back and forth along the chain .
the group velocity of each phonon is simply @xmath109 .
see the supplementary information at the end of the manuscript for further details . for the sake of completeness we provide the expression of the langevin force @xmath110 .
k. zyczkowski and i. bengtsson , _ geometry of quantum states _ , cambridge university press 2008 .
s. john and j. wang , phys .
rev .
lett .
* 64 * , 2418 ( 1990 ) .
r. vasile , s. maniscalco , m. g. a. paris , h .-
p .
breuer , and j. piilo , phys .
rev .
a * 84 * , 052118 ( 2011 ) . in respect to the trace distance , the fidelity increases under any cpt evolution
. a. w. chin , a. rivas , s. f. huelga and m. b. plenio , j. math . phys . * 51 * , 092109 ( 2010 ) .
d. chruscinski , a. kossakowski , and a. rivas , phys .
rev .
a * 83 * , 052128 ( 2011 ) ; p. haikka , j. d. cresser , and s. maniscalco , phys . rev .
a * 83 * , 012112 ( 2011 ) ; |
nowadays the most accepted model in cosmology which explains the evolution of the universe is known as @xmath0cdm . in this model 4 per cent of the total content of the universe is baryonic matter , 22 per cent is non - baryonic dark matter ( dm ) and the rest is in some form of cosmological constant .
@xmath0cdm has achieved several observations with outstanding success , like for example the fact that the cosmic microwave background radiation can be explained in great detail and that it provides a framework within one can understand the large - scale isotropy of the universe and important characteristics on the origin , nature and evolution of density fluctuations which are believed to give rise to galaxies and other cosmic structures .
there remain , however , certain conflicts at galactic scales , like the cusp profile of central densities in galactic halos , the overpopulation of substructures predicted by @xmath4-body numerical simulations which are an order of magnitude larger than what has been observed , etc ( see for example @xcite , @xcite , @xcite , and @xcite )
. there are other important issues of the @xmath0cdm model which require further considerations .
observations point out to a better understanding of the theory beginning with the local void , which contains far fewer galaxies than the expected .
another problem arises for the so called pure disk galaxies , which do not appear in numerical simulations of structure formation in the standard model .
these problems would be solved , if the structure grew faster than it does in the standard paradigm @xcite . on the other hand
, @xcite also found that the collision velocity of 3000km / s at r200 for the bullet cluster is very unlikely and might no be explained by the standard model .
a final example of inconsistencies can be seen in a paper by @xcite , who found anomalies in the mass power spectrum obtained by sdss and the one obtained with the cdm model . with these and other results
it seems necessary to change the @xmath0cdm paradigm to try and explain the formation of structure in the universe .
the incorporation of a new kind of dm , different from the one proposed by the @xmath0cdm model into the big bang theory holds out the possibility of resolving some of these issues .
several authors have proposed interesting alternatives in where they try to solve the latter explained difficulties that the cdm scenario seem not solve .
some of these models consider warm dark matter ( @xcite , @xcite ) , self interacting dark matter ( @xcite,@xcite ) , and other exotic scenarios as braneworld models @xcite .
in fact , there are models that do not include dm but instead modify the newtonian force law @xcite .
other alternative scenario that has received much attention in the last years is the scalar field dark matter ( sfdm ) model .
the main idea is simple ( see @xcite ) , the nature of the dm is completely determined by a fundamental scalar field @xmath5 .
the sfdm model proposes that galactic haloes form by bose - einstein condensation of a scalar field ( sf ) whose boson has an ultra - light mass of the order of @xmath6ev . from this mass
it follows that the critical temperature of condensation @xmath7tev is very high , therefore , they form bose - einstein condensates ( bec ) drops very early in the universe . in addition , the compton length @xmath8 associated to this boson is about @xmath9 kpc that corresponds to the dark halo - size of typical galaxies in the universe .
thus , it has been proposed that these drops are the haloes of galaxies ( see @xcite ) , @xmath10 , that haloes are huge drops of sf . on the other hand ,
big structures form like in the @xmath0cdm model , by hierarchy @xcite , thus , all successful predictions of the standard model at large scales are well reproduced by sfdm . in other words , in the sfdm model the haloes of galaxies do not form hierarchically , they are formed at the same time and in the same way when the universe reaches the critical temperature of condensation of the sf . from this
it follows that all galaxies must be very similar because they formed in the same manner and at the same moment @xcite .
therefore , from this paradigm we have to expect that there exist well formed galaxy haloes at higher redshifts than in the @xmath0cdm model . in this model
the scalar particles with that ultra light mass are such that their wave properties avoid the cusp problem and reduce the high number of small satellites by the quantum uncertainty principle @xcite which is another problem that is still present in the @xmath0cdm model @xcite .
summarizing , it is remarkable that with only one free parameter , the ultra - light scalar field mass ( @xmath11ev ) , the sfdm model fits : 1 . the evolution of the cosmological densities @xcite .
the acoustic peaks of the cosmic microwave background @xcite .
3 . the scalar field has a natural cut off , thus the substructures in clusters of galaxies is avoided naturally . with a scalar field mass of @xmath12ev
the amount of substructure is compatible with the observed one @xcite .
we expect that sfdm forms galaxies earlier than the cdm model , because they form becs at a critical temperature @xmath13tev .
so if sfdm is right , we have to see big galaxies at high redshifts with similar features @xcite .
the rotation curves of big galaxies and lsb galaxies @xcite .
6 . with this mass ,
the critical mass of collapse for a real scalar field is just @xmath14 , i.e. , the one observed in galaxies haloes @xcite .
the observed properties of dwarf galaxies , i.e. , the minimum length scale , the minimum mass scale , and their independence from the brightness @xcite .
and recently it has been demonstrated that if the scalar mass is @xmath15ev , the sfdm haloes would have cores large enough to explain the longevity of the cold clump in ursa minor and the wide distribution of globular clusters in fornax ( see @xcite ) .
the idea was first considered by @xcite and independently by @xcite suggested bosonic dark matter as a model for galactic halos @xcite . in the bec model , dm haloes can be described , in the non - relativistic regime , as newtonian gravitational condensates made up of ultra light bosons condensated in a single macroscopic wave function .
thus , becs haloes can be described as a coherent scalar field @xmath5 .
several authors have introduced a dynamic scalar field with a certain potential @xmath16 as a candidate to dark matter , although there is not yet an agreement for the correct form of the potential of the field .
one interesting work pointing this way was done by @xcite and independently by @xcite where they used a potential of the form @xmath17 $ ] to explain the core density problem for disc galaxy halos in the @xmath0cdm model ( see also @xcite , @xcite and @xcite ) .
@xcite presented a model for the dm in spiral galaxies , in which they supposed that dm is an arbitrary sf endowed with a scalar potential .
other scalar potential widely used to describe dm is @xmath18 @xcite .
this potential is very interesting because it can mimic the cosmological evolution of the universe predicted by the @xmath0cdm model .
if we consider a sf self - interaction , we need to add a quartic term to the sf potential @xcite , in this case the equation of state of the sf is that of a polytope of index n=1 ( see @xcite ) .
different issues of the cosmological behavior of the sfdm / bec model have been studied ( see for example @xcite ) .
@xcite proposed fuzzy dark matter that is composed of ultra light scalar particles who are initially in the form of a bec . in their work
@xcite used a bosonic dark matter model to explain the structure formation via high - resolution simulations , @xcite reviewed the key properties that may arise from the bosonic nature of sfdm models .
recently @xcite developed a further analysis of the cosmological dynamics of sfdm / bec as well as of the evolution of their fluctuations ( see also @xcite ) .
in the same direction , @xcite studied , the growth of scalar fluctuations and large - scale structure formation with the fluid approach of the sfdm / bec model . on the other hand ,
sfdm / bec model has provided to be a good candidate for dm haloes of galaxies in the universe because it can explain many aspects where the standard model fails ( @xcite ) .
in addition , many numerical simulations have been performed to study the gravitational collapse of sfdm / bec model @xcite .
@xcite found approximate analytical expression and numerical solutions of the mass - radius relation of sfdm / bec haloes .
recently , @xcite give constraints on the boson mass to form and mantain more than one vortex in sfdm / bec haloes .
these constraints are in agreement with the ultra light mass founded in previous works ( see also @xcite ) .
lately , @xcite performed n - body simulations to study the dynamics observed in ursa minor dwarf galaxy .
they model the dark matter halo of ursa minor as a sfdm / bec halo to establish constraints for the boson mass .
moreover , they introduce a dynamical friction analysis with the sfdm / bec model to study the wide distribution of globular clusters in fornax . an overall good agreement is found for the ultra light mass @xmath19ev of bosonic dark matter .
the main objective of this review is to introduce the framework of the sfdm / bec model that assumes that dm is a scalar field that involves a self - interacting potential of the form @xmath20 , where @xmath6ev is the mass of the scalar field , @xcite , @xcite and @xcite .
we solve the friedmann equations for the sfdm / bec model to show that it behaves just like the cdm model . also , we show that the sfdm / bec predicts galaxy formation earlier than the cdm model , because they form bec at a critical temperature @xmath21 tev .
so , if sfdm is right , this would imply that we have to see big galaxies at high redshifts . in order to do this ,
we study the density fluctuations of the scalar field from a hydrodynamical point of view , this will give us some information about the energy density of dark matter halos necessary to obtain the observational results of large - scale structure .
here we will give some tools that might be necessary for the study of the early formation of structure .
in section [ cosmology ] we study the cosmological evolution of the sf with a field and a fluid approach .
we obtain a numerical solution for the scalar density as well as analytical expressions for the kinetic and potential energies of the sf . also , we study the evolution for the density contrast of scalar fluctuations in the linear regime . in section [ galacticsfdm ]
we study the rotation curves and the cusp / core discrepancy with sfdm haloes to compare with the predictions of cdm .
finally we give our conclusions in section [ sec : conclusions ] .
the framework of the standard cosmological model is of a homogeneous and isotropic universe whose evolution is best described by friedmann s equations that come from general relativity and whose main ingredients can be described by fluids whose characteristics are very similar to those we see in our universe .
of course , the universe is not exactly homogeneous and isotropic but this standard model does give us a framework within which we can study the evolution of the expansion rate of the universe as well as the evolution of small fluctuations that give rise to gravitational structures in the universe as galaxies and clusters of galaxies . in this section ,
we study the cosmological dynamics of the sfdm / bec model and the evolution of their scalar perturbations with two different frameworks : the field and fluid approach .
we use the friedmann - lematre - robertson - walker ( flrw ) metric with scale factor @xmath22 .
our background universe is composed by sfdm ( @xmath23 ) endowed with a scalar potential @xmath24 , radiation ( @xmath25 ) , neutrinos ( @xmath26 ) , baryons ( @xmath27 ) , and a cosmological constant ( @xmath0 ) as dark energy .
we begin by recalling the basic background equations . from the energy - momentum tensor @xmath28 for a scalar field ,
the scalar energy density @xmath29 and the scalar pressure @xmath30 are given by @xmath31 @xmath32 where the dots stand for the derivative with respect to the cosmological time and @xmath33 is the kronecker delta .
thus , the equation of state ( eos ) for the scalar field is @xmath34 with @xmath35 the radiation fields , the baryonic component and the cosmological constant are represented by perfect fluids with baryotropic equation of state @xmath36 , where @xmath37 is a constant , @xmath38 . for example ,
@xmath39 for radiation and neutrinos , @xmath40 for baryons , and for a cosmological constant @xmath41 .
the einstein - klein - gordon equations that describe this universe are ( in units @xmath42 ) .
@xmath43 [ eq : back ] with the friedmann constraint @xmath44 being @xmath45 , @xmath46 the hubble parameter and the commas stand for the derivative with respect to scalar field .
notice that background scalar quantities at zero order have the subscript @xmath47 . in order to solve the system of equations ( [ eq : back ] ) , we define the following dimensionless variables @xmath48 here we take the quadratic scalar potential @xmath49 with @xmath50ev . using these variables ,
the equations ( [ eq : back ] ) for the evolution of the background universe are transformed into @xmath51 [ eq : dsb ] where the prime denotes a derivative with respect to the e - folding number @xmath52 , and @xmath53 is defined as @xmath54 we have introduced the variable @xmath55 , where @xmath56 is a constant , as a control parameter for the dynamics of @xmath57 . in eq .
( [ eq : dsb_s ] ) , @xmath58 is a constant and the exponent @xmath59 is @xmath60 ( see @xcite for more details of the control parameter ) . with these variables ,
the density parameters @xmath61 for each component @xmath62 can be written as @xmath63 subject to the friedmann constraint @xmath64 in addition , we may write the eos of the scalar field as @xmath65 since @xmath66 is a function of time , if its temporal average tends to zero , this would imply that @xmath67-dark matter can be able to mimic the eos for cdm .
we solve the system of equations ( [ eq : dsb ] ) for the background universe numerically with a four order adams - bashforth - moulton ( abm ) method .
we take as initial conditions the best estimates from 5-years @xcite and 7-years wmap @xcite values to @xmath68 , @xmath69 , @xmath70 , @xmath71 , @xmath72 and the exponent @xmath73 . in fig.[fig : phi ] we see the evolution of the scalar field @xmath74 . this figure shows how the scalar field oscillates very stark about the minimum of the scalar potential @xmath75 , on the bottom of this figure we show a zoom of these oscilations . in fig.[fig : ox ] we show the evolutions of the kinetic ( @xmath76 ) and potential energy ( @xmath77 ) of the scalar field , as expected , the sf oscilations are translated into very stark oscilations for the kinetic and the potential energies of the sf .
however , observe the evolution of the dark matter density of the scalar field , that means @xmath78 in fig.[fig : back ] .
note the following crucial point , although the kinetic and potential energies show very stark oscillations , the sum of both energies does not oscillates at all , this sum is the density parameter @xmath79 , which does not display any oscillation .
the important point we have to take from now on into account is that the oscillations are not physical observables at all , they are a feature of the sf , what we observe in fact is the density of the sf which does not oscillates .
fig.[fig : back ] shows the numerical evolution of the density parameters in our model today , so that it relates to the redshift @xmath25 by @xmath80 . ] . at early times , radiation dominates the evolution of the universe .
later on , the universe has an epoch where the energy density radiation is equal to the dark matter density , at @xmath81 , then dark matter begins to dominate the evolution .
the recombination era in sfdm model occurs at @xmath82 . at later times , the cosmological constant dominates the dynamics of the universe at @xmath83 .
the behavior is exactly the same as in the @xmath0cdm model .
furthermore , in the recombination era of the sfdm model , the neutrinos made up @xmath84 per cent of the universe completely in agreement with the measurements of wmap .
fig.[fig : w0 ] shows the evolution of the eos for the sf .
although the eos varies with time ( oscillations ) , the temporal average , @xmath85 , drops to zero .
therefore , sfdm is like a pressureless fluid and behaves as cdm at cosmological scales ( @xcite , @xcite , @xcite , @xcite ) . in this section
we perform a transformation in order to solve the friedmann equations analytically with the approximation @xmath86 . here
we take the scalar potential as @xmath87 , ( we remain that we use units where @xmath42 unless it is stipulated ) .
then for the ultra - light boson particle we have that @xmath6 ev .
now we express the sf , @xmath88 , in terms of the new variables @xmath89 and @xmath90 , where @xmath89 is constant in the background and @xmath91 will be the energy density of the fluid also in the background .
so , our background field is proposed as @xmath92 where , @xmath93 and with this our sf in the background can be finally expressed as , @xmath94 with this we obtain @xmath95 ^ 2 .
\label{backsf } \end{aligned}\ ] ] to simplify , observe that the uncertanty relation implies that @xmath96 , and for the background in the non - relativistic case the relation @xmath97 is satisfied .
notice also that for the background we have that the density goes as @xmath98 but we also have that @xmath99 ev @xmath100 ev , so with these considerations at hand for the background , in ( [ backsf ] ) we have @xmath101 finally , substituting this last equation and equation ( [ tri ] ) into ( [ rhophi0 ] ) when taking @xmath102 , we obtain @xmath103=2\frac{m^2}{\hbar^2}\hat{\rho}_0 .
\label{trigo}\ ] ] comparing this result with ( [ eq : dens ] ) we have that the identity @xmath104 holds for the background , so comparing with ( [ trigo ] ) , @xmath105 @xmath106 in terms of the two analytic results , we show the evolution of the kinetic and potential energies ( [ eq : cinet ] ) and ( [ eq : potencial ] ) in fig .
[ fig1 ] , where for the evolution we used the e - folding number @xmath4 and the fact that @xmath107 . observe the excelent accordance with the numerical results in @xcite for the kinetic and potential energies of the background respectively .
therefore , we expect that if we sum the analytical kinetic and the potential energies we will obtain the same behavior for the sfdm density shown in fig .
[ fig : back ] .
in what follows we will use the analytic expressions ( [ eq : cinet ] ) and ( [ eq : potencial ] ) instead of the numerical ones .
nowadays it is known that our universe is not exactly isotropic and spatially homogeneous like the flrw metric describes . there
exist small deviations of this model that were the cosmological seeds to lead the large - scale structure formation in the universe . in this section
we compute the growth of the sfdm overdensities @xmath108 at the stage when the density contrast @xmath109 is much smaller than unity .
it is believed that the universe was almost uniform after inflation , with very small density contrast .
as the universe expanded , the small over - densities grew until they began to collapse , leading to the formation of structure in the universe .
here we consider small enough deviations in the flrw model , so that they can be treated by linear perturbation theory . in the sfdm paradigm ,
the dark haloes of the galaxies are becs composed of ultra light scalar particles with @xmath15ev then it would be about @xmath110 gev@xmath111 scalar particles to follow in a single galaxy and the occupation numbers in galactic haloes are so big that the dm behaves as a coherent classical scalar field @xmath5 that obeys the klein - gordon equation @xmath112 , where @xmath113 is the dalambertian .
therefore , to study the structure formation in the universe we describe a non - interacting sfdm model such that : i ) it can describe it more as a field than as particles and ii ) we find a function that only depends on the three spatial coordinates and time .
first , we introduce the perturbed metric tensor in the flrw background , we only consider scalar perturbations .
we then give the equation of energy - momentum conservation and the einstein field equations for the perturbed metric . by definition , a perturbation done in any quantity ,
is the difference between its value in some event in real space - time , and its corresponding value in the background .
so , for example for the sf we have @xmath114 where the background is only time dependent , while the perturbations also depend on the space coordinates .
similar cases apply for the metric ; @xmath115 .
\end{aligned}\ ] ] here @xmath116 is a perturbation associated to the curvature and e is asociated to the expansion .
we will work under the newtonian gauge , which is defined when @xmath117 .
an advantage of using this gauge is that here the metric tensor @xmath118 is diagonal , and so the calculations become much easier
. we will only work with scalar perturbations , vector and tensor perturbations are eliminated from the beginning , so that only scalar perturbations are taken into account .
another advantage in using this gauge is that @xmath119 will play the role of the gravitational potential , this will help us to have a simpler physical interpretation , i.e. , both potentials @xmath119 and @xmath116 are then related .
this metric has already been used in other works ( see for example @xcite , @xcite and @xcite ) .
the perturbed energy - momentum tensor for the sf can be written as the background value @xmath120 plus a perturbation @xmath121 where @xmath122 $ ] , i. e. @xmath123 we now derive the perturbed evolution equations for the different quantities mentioned above ; the scalar perturbation @xmath124 and the scalar potential @xmath116 .
for the perturbed energy - momentum tensor , we get @xmath125 [ eq : tperturbado ] where in equations ( [ eq : tperturbado ] ) the dot denotes differentiating with respect to cosmological time @xmath126 , which is related to conformal time by the simple relation @xmath127 in the newtonian gauge , the metric tensor @xmath128 becomes diagonal and from this , in the trace of the einstein s equations , the scalar potentials @xmath116 and @xmath119 are identical and , therefore , @xmath116 plays the role of the gravitational potential @xmath129 usually this equation contains a term of anisotropic stress , which vanishes in the case of a sf . altogether , the perturbed einstein s equations @xmath130 to first order for a sf in the newtonian gauge are @xmath131 .
\label{eq : sfmet}\end{aligned}\ ] ] these equations describe the evolution of the scalar perturbations . for the evolution of the perturbations in the sf we use the perturbed klein - gordon equation
@xmath132 in order to solve the equations ( [ eq : sfmet ] ) and ( [ eq : kgl ] ) , we will turn to fourier s space .
the beauty of this expansion relaying on the fact that each fourier mode will propagate independently .
to first order , the derivation of fourier s components is straightforward .
the perturbation @xmath124 relates to its fourier component @xmath133 by @xmath134 where @xmath59 is the wave number . here
the wave number is defined as @xmath135 , being @xmath136 the length scale of the perturbation .
the perturbed equations ( [ eq : sfmet ] ) altogether with the sf transformation read @xmath137&= & 8\pi g(\dot{\phi}_0\dot{\delta\phi_k } \nonumber\\&&-\phi\dot{\phi}_0 ^ 2-v_{\phi}\delta\phi_k)\label{eq : fsc},\end{aligned}\ ] ] [ eq : fsp ] and the klein - gordon equation ( [ eq : kgl ] ) transforms into @xmath138 these set of equations describe the evolution of the perturbations .
( [ eq : fsa ] ) makes reference to the evolution of the energy density , eq .
( [ eq : fsb ] ) to the evolution of the gravitational potential and finally , eq .
( [ eq : kgfou ] ) refers to the perturbations over the sf .
now , from the time derivative of ( [ rhopert ] ) , we have @xmath139 performing a fourier transformation of the above equation and combining it with the unperturbed and perturbed klein - gordon equations ( [ eq : nopertkg ] ) and ( [ eq : kgfou ] ) respectively , and with the use of equation ( [ eq : fsp ] ) we arrive at @xmath140 and on the other hand we have , @xmath141 thus , @xmath142 this last equation can be expressed in terms of the density contrast @xmath143 making use of the equations from the background , we have @xmath144 where we have defined the function @xmath145 as @xmath146 it is convenient to define the function @xmath147 as well , @xmath148 taking the time average of equation ( [ pro1 ] ) we obtain @xmath149 in equation ( [ eq : deltlini ] ) for the radiation and matter dominated eras the first term in the parenthesis goes as @xmath150 , see for example @xcite , also we can see from fig.[fig : w0 ] that @xmath151 .
it is easy to see the temporal average of the terms @xmath152 and @xmath145 of the equation ( [ eq : deltlini ] ) . from fig.[fig : fphi0 ] we observe that that @xmath153 tends to one . on the other hand
, since we are using post - newtonian approximation @xmath154 drops to zero , meaning that the second term on the right - hand side of eq .
( [ eq : deltlini ] ) dissapears .
so , due to the scalar oscillations around the minimum of the scalar potential , the unperturbed sf behaves very similar to the cdm model , but equation ( [ eq : deltlini ] ) tell us that their perturbations too , all the growing behavior for the @xmath59 modes are recovered and preserved so far .
finally , we conclude that eq .
( [ eq : deltlini ] ) resembles the equation for the density contrast as in the standard cdm model @xcite .
this means that the sfdm perturbations in the linear regime grow exactly as cdm perturbations .
we now study the cosmological evolution of our model through a dynamical system . in order to solve the system of equations for the perturbations given in eqs .
( [ eq : fsp ] ) and ( [ eq : kgfou ] ) , we define the following dimensionless variables given by @xmath155 using the variables defined in eq .
( [ eq : dvlr ] ) , the equations can be transformed into an autonomus system with respect to the e - folding number @xmath4 , then the density contrast can be written in terms of these variables @xmath156}{\omega_{\phi_{0}}}. \label{eq : deltala}\ ] ] therefore the transition between the linear and non - linear regime of the sfdm perturbations can be studied through a numerical code .
we rewrite equation ( [ eq : kgfou ] ) in the following form @xmath157 equation ( [ eq : kgfou_2 ] ) is the one of a harmonic oscillator with a damphing term @xmath158 and an extra force .
this equation contains oscillating , growing and decaying solutions , depending on the factor @xmath159 .
if this factor is positive , we expect that the fluctuations oscillate , only if the factor is negative we will have growing up fluctuations .
this factor contains the scale @xmath160 , which is always positive .
thus , if we expect to have growing up fluctuation of the scalar field , this happen only when the potential @xmath161 is negative
. then equation ( [ eq : kgfou_2 ] ) will have growing up solutions for certain values of the scale @xmath162 , this is , only for enough small values of the scale , which means for enough big fluctuations .
this is the way we can fix the mass of the scalar field , the potential depends on the value of the scalar field mass .
observing the smallest structure we see in the universe , we are able to fix the mass of the scalar field .
using this observations one can avoid the problem of substructure of the @xmath0cdm . in order to do this , in @xcite the mass of the scalar field
was fixed in @xmath163ev .
the sf @xmath5 has very stark oscillations from the beginning , these oscillations are transmitted to the fluctuations which apparently seems to grow very fast and are too big .
nevertheless , this behavior is not physical , because we only see the oscillations of the fields , but we can not see clearly the evolution of its density , @xcite . in order to drop out these oscillations , in what follows we perform two transformations .
the first one changes the perturbed klein - gordon equation into a kind of schrdinger equation and the second transforms this last equation into a hydrodynamical system , where we can interpret the physical quantities easier and the observable quantities become much clear .
first , we consider the perturbed klein - gordon [ eq : kgl ] where we have set @xmath164 @xmath165 now we express the perturbed sf @xmath124 in terms of the field @xmath166 , @xmath167 term which oscillates with a frequency proportional to @xmath168 and @xmath169 which would be proportional to a wave function of an ensamble of particles in the condensate . with this equation and
the expresion for the potential of the scalar field , ( [ k - g ] ) transforms into @xmath170 where we have defined @xmath171 notice that this last equation could represent a kind of gross - pitaevskii equation in an expanding universe .
the only modification of equation ( [ schrodinger ] ) in comparison to the schrdinger or the gross - pitaevskii equation is the scale factor @xmath172 associated to the co - moving spatial gradient and that the laplacian @xmath173 transforms into the d alambertian @xmath113 . to explore the hydrodynamical nature of bosonic dark matter
, we will use a modified fluid approach . then , to make the connection between the theory of the field and the condensates waves function , the field is proposed as ( madelung transformation @xcite ) @xmath174 where @xmath166 will be the condensates wave function with @xmath175 and @xmath176 , @xcite .
here we have separated @xmath166 into a real phase @xmath89 and a real amplitude @xmath177 and the condition @xmath178 is satisfied . from ( [ psi ] )
we have @xmath179 ^ 2\nonumber\\ & -&\frac{\hbar^2}{2}(\dot s / m)^2=0.\label{hidro1 } \label{hidro } \end{aligned}\ ] ] now , taking the gradient of ( [ hidro1 ] ) then dividing by @xmath180 and using the definition @xmath181 we have , @xmath182 where in ( [ hidro1 ] ) @xmath183 and in ( [ eq:1 ] ) we have defined @xmath184 .
it is worth noting that , to this moment this last set of equations do not involve any approximations with respect to equation ( [ schrodinger ] ) and can be used in linear and non - linear regimes .
now , neglecting squared terms , second order time derivatives and products of time derivatives in this last set of equations we get , @xmath185 where the equation for the gravitational field is given by the poisson s equation ( [ poisson ] ) . in these equations
we have introduced @xmath186 , such that @xmath187 .
equation ( [ vel ] ) shows the proportionality between the gradient of the phase and the velocity of the fluid .
note that @xmath188 can represent the velocity field for the fluid and @xmath189 will be the particles density number within the fluid .
also there exists an extra term of third order for the partial derivatives in the waves amplitude which goes as the gradient of @xmath190 , this term would result in a sort of quantum pressure that would act against gravity .
we remain that @xmath119 represents the gravitational field .
these two sets of equations ( [ navier ] ) and ( [ navier1 ] ) would be analogous to the euler s equations of classical fluids , with the main difference that there exists a quantum term , which we will call @xmath191 and will be given by @xmath192 which can describe a force or a sort of negative quantum pressure .
for equation ( [ navier ] ) we have that @xmath189 will represent the mass density or the particles density number of the fluid , where all the particles would have the same mass .
finally , these equations describe the dynamics of a great number of non - interacting identical particles that manifest themselves in the form of a fluid , also equation ( [ schrodinger ] ) can describe a great number of non - interacting but self - interacting identical particles in the way of a bose gas , when the probability density is interpreted as the density number .
now , these hydrodynamical equations are a set of complicated non - linear differential equations . to solve them
we will restrict ourselves to a vecinity of total equilibrium .
for this let @xmath91 be the mass density of the fluid in equilibrium , the average velocity @xmath193 will be taken as zero in equilibrium , so we will only have @xmath194 out of equilibrium .
then , the matter in the universe will be considered as a hydrodynamical fluid inside an universe in expansion .
this system will then evolve in this universe and later on they will collapse because of their gravitational attraction . then from ( [ navier ] ) for the mass density of the fluid in equilibrium we have , @xmath195 with solutions of the form @xmath196 where as we know , in general if we have an equation of state of the form @xmath197 and consider cdm or dust as dark matter such that @xmath198 it holds that @xmath199 then , when the scale factor was small , the densities were necessarily bigger .
now , the particles density number are inversely proportional to the volume , and must be proportional to @xmath200 , therefore the matter energy density will also be proportional to @xmath200 , result that is consistent with our expression ( [ back ] ) .
in addition , the cosmological behavior of @xmath91 is in agreement with the numerical results show in fig . [
fig : back ] .
now for the system out of equilibrium we have @xmath201 equations that are valid in a universe in expansion . in order to solve system ( [ outeq ] ) we look for solutions in the form of plane waves , for this the convenient ansatz goes as @xmath202 where @xmath203 is the position vector and @xmath59 is a real wavevector which corresponds to a wavelength @xmath136 . if we substitute these ansatz in the set of equations ( [ outeq ] ) , we then have @xmath204 where we have defined the velocity @xmath205 to solve the system is convenient to rotate the coordinate system so that the propagation of the waves will be along the direction of one of the axes . for this we know that the velocity vector can be divided into longitudinal ( parallel to @xmath59 ) and transverse ( perpendicular to @xmath59 ) parts , such that @xmath206 , where @xmath207 is the vector perpendicular to the wave propagation vector @xmath208 . in terms of @xmath207 for equations ( [ eq1])-([modos ] ) we have @xmath209 in addition to an equation for @xmath207 , @xmath210 , with solutions
@xmath211 with @xmath212 a constant of integration , i.e. , perpendicular modes to the wave vector are eliminated with the expansion of the universe .
now , if we use the result ( [ back ] ) , then equation ( [ eq2 ] ) can be written as @xmath213 system ( [ eq2])-([lambda ] ) can be treated as in the case of a universe with no expansion , so combining the two equations and with the aid of ( [ lambda1 ] ) , we get @xmath214\delta=0 , \label{delta}\ ] ] where @xmath215 is defined as the density contrast .
this will be a fundamental equation in the understanding of the evolution of the primordial fluctuations .
first we will give a brief summary of the results for the @xmath0cdm model , this will enable us to make a direct comparison with our results . for cdm the equation for the evolution of the density contrast
is given by , @xmath216 where @xmath217 is defined as the sound velocity ( which in our case it is not ) .
now lets analyze equation ( [ deltacdm ] ) at the beginning of the matter dominated era a time just after the epoch of equality , and just before recombination when the radiation has cooled down and the photons do not interact with the electrons anymore , for a relativistic treatment see @xcite . in this era , @xmath218 , practically all the interesting fluctuation modes are well within the horizon , and the evolution of the perturbations can be well described within the newtonian analysis . at this time
, matter behaves like dust with zero pressure .
so we have @xmath219 , @xmath220 and @xmath221 therefore @xmath222 . for equation ( [ deltacdm ] )
we have @xmath223 the solutions to this equation are of the form @xmath224 where @xmath225 and @xmath226 are integration constants , from this solution we can see that we have modes that will disappear as time goes by , and modes that grow proportionally to the expansion of the universe .
this is an important result , because then the density contrast will grow proportionally to the expansion of the universe when this is dominated by matter .
then , these fluctuations can maybe grow and give life to the galaxies , clusters of galaxies and all the large - scale structure we see now a days .
now lets see what happens to the sfdm at this epoch ( @xmath218 ) .
the evolution of the perturbations in this case will be given by equation ( [ delta ] ) .
in general we have that in equation ( [ delta ] ) the term @xmath227 is very small throughout the evolution of the pertubations ( @xmath228 for small @xmath59 ) , so it really does not have a significant contribution on its evolution .
when the condition @xmath102 is taken we can have a bec that might be or might not be stable , if there exists stability the results of sfdm are consistent with those obtained from cdm ( in this case both equations ( [ delta ] ) and ( [ deltacdm ] ) are almost equal ) , the condition of stability for the bec in the sfdm case will come from the study of @xmath229 together with @xmath191 .
@xmath230 as we can see in fig .
[ fig2 ] the perturbations used for the @xmath0cdm model grow in a similar way for the sfdm model , when @xmath102 , in this case both perturbations can give birth to structures quite similar in size , and this will happen with all the fluctuations as long as @xmath59 is kept small . when @xmath231 the results are quite different , so when discussing the evolution of the density perturbations , there are two different cases : i ) in the case of @xmath232 the amplitude of the density contrast tends to decrease as @xmath229 grows bigger and bigger away from zero until the amplitude of the density takes negative values ( around @xmath233 ) , telling us that this kind of fluctuations can not grow in time , and hence do not form a bec .
ii ) on the other hand if @xmath234 the fluctuations for the density contrast alway grow despite their size , this results means either than the fluctuations grow and form a stable bec or than the density grows because it is collapsing into a single point and our bec might be unstable , the study of the stability of this fluctuations needs then to be studied with non - linear perturbation theory .
these results are shown in fig .
[ fig3 ] , in both figures [ fig2 ] and [ fig3 ] the initial condition for @xmath235 goes as @xmath236 in accordance with the data obtained from wmap .
if these fluctuations result stable and because they are big in size , this means that they can only give birth to large structures .
these fluctuations can then help for the formation of large clusters or other large - scale structure in the universe at its early stages ( around @xmath218 ) .
then , as these kind of sfdm can only interact with radiation in a gravitational form it is not limited by its interaction with radiation , and the dark matter halos can then create potential wells that will collapse early in time giving enough time for the structures to form .
then if dm is some kind of sfdm , the luminous matter will follow the dm potentials giving birth to large - scale structure .
finally , a further analysis is required to obtain better results of the growth of scalar perturbations with self - interacting sfdm as well as to study the transition between linear and no linear regime of scalar fluctuations .
nevertheless , we conclude that the sfdm model at cosmological scale is a plausible alternative to dm nature because it preserves all the success of the @xmath0cdm model .
in this section we study the sfdm / bec model at galactic scales , we will focus on the cusp / core discrepancy between the numerical predictions of the standard model and the astronomical observations .
as already said before , recent observations in far and nearby galaxies have shown that the @xmath0cdm model faces serious conflicts when trying to explain the galaxy formation at small scales @xcite .
for instance , in the @xmath0cdm simulations the halos present rising densities towards the central region behaving as @xmath237 well within 1 kpc @xcite . on the other hand , several observations suggest that the dynamics ( rotation curves ) of dwarf and lsb galaxies are more consistent with a constant central density @xcite , this is most commonly known as the cusp / core problem @xcite . studying a wide range of galaxies of different morphologies and with magnitudes in the interval
@xmath238 @xcite fit their rotation curves ( rc ) using a burkert profile for the dm @xcite and found that @xmath239 remains approximately constant , where @xmath240 with @xmath241 the central dm density and @xmath242 the core radius .
similar result where found in @xcite .
exploring further the constant value of @xmath243 for the dm , @xcite found that within @xmath242 the dm central surface density in terms of the mass inside it ( @xmath244 ) is @xmath245m@xmath246 pc@xmath247,the gravitational acceleration due to dm felt by a test particle at the radius @xmath242 was found to be @xmath248 additionally they reported the acceleration due to the luminous matter at @xmath242 to be @xmath249 . in the @xmath0cdm model
the galaxies have evolve through numerous mergers and grew in different environments , the star formation and basic properties of the galaxies are not expected to be a common factor among them . therefore giving both the constancy of @xmath243 and the core in the central regions of galaxies seems very unlikely in this model .
any model trying to become a serious alternative to @xmath0cdm has to succeed in reproducing observations in which the standard model fails but also has to keep the solid description at large scale .
it is always necesary to test the sfdm / bec model with the two observations mentioned above , the cusp / core problem and the constant dm central surface density . in order to do this , in @xcite
it was used the thomas - fermi approximation and a static bec dm halo to fit rotation curves of a set of galaxies .
however , so far there was no comparison between the density profile and the data , in that work this was done by fitting rotation curves of 13 high resolution low surface brightness ( lsb ) galaxies and additionally compare the fits to two characteristic density profiles 1 ) the cuspy navarro - frenk - white ( nfw ) profile that results from n - body simulations using @xmath0cdm and 2 ) the pseudo isothermal ( pi ) core profile .
the comparison allow to show that the model is in general agreement with the data and with a core in the central region .
for the second goal the meaning of a core is somewhat ambiguous .
large scale @xmath4-body simulations using collissionless cdm the internal region of dm halos show a density distribution described by a power law @xmath250 with @xmath251 , such behaviour is what is now called a cusp .
on the other hand , observations mainly in dwarf and lsb galaxies seem to prefer a central density going as @xmath252 .
this discrepancy between observation and the cdm model receives the name of cusp / core problem . among the empirical profiles most frequently used to describe the constant density behavior in this galaxies are the pi @xcite , the isothermal @xcite and the burkert profile @xcite .
even though their behavior is similar in the central region and is specified by the central density fitting parameter , their second parameter called the core radius does not represent the same idea .
for instance , in the pi profile ( eq.([eq : pi ] ) ) we see that the core radius will be the distance in which the density is half the central density . for the burkert profile
the core radius @xmath253 will be when @xmath254 and for an isothermal profile ( i ) @xcite @xmath255 .
hence , we see an ambiguity in the meaning of the core radius , they get the same name but the interpretation depends on the profile . if we want to compare the central density of lsb galaxies with that of nfw , it usually suffice to have a qualitative comparison , so far this is what we have been doing by fitting empirical profiles .
however , high resolution rotation curves demand a more quantitative comparison . indeed ,
if we want to test models by fitting rcs we have to know the specific meaning and size of the core , then we will be able to tell if a model is consistent with a cusp or not by making a direct comparison with the data .
in @xcite it was proposed a new definition for the core and core radius that allow us to decide when a density profile is cusp or core . using this definition in the sfdm / bec model discussed above it is possible to find that the sfdm / bec model can reproduced the constant value of @xmath243 and as a crosscheck with the pi profile , in @xcite
was found that the results are in very good agreement with observations .
this argues in favor of the model and the core definition .
following @xcite we provide the dark matter density profiles that will be used for the analysis . the case in which the dark matter is in the form of a static bec and
the number of dm particles in the ground state is very large was considered in @xcite . following this paper and assuming the thomas - fermi approximation @xcite which neglects the anisotropic pressure terms that are relevant only in the boundary of the condensate , the system of equations describing the static bec in a gravitational potential @xmath75 is given by @xmath256 @xmath257 with the following eos @xmath258 where @xmath259 , @xmath260 is the mass density of the static bec configuration and @xmath261 is the pressure , as we are considering zero temperature @xmath261 is not a thermal pressure but instead it is produced by the strong repulsive interaction between the ground state bosons .
assuming spherical symmetry and denoting @xmath262 as the radius at which the pressure and density are zero , the density profile takes the form @xcite @xmath263 where @xmath264 and @xmath265 is the bec central density , @xmath168 is the mass of the dm particle and @xmath180 is the scattering lenght .
the mass at the radius @xmath266 is given by @xmath267 from here the tangential velocity @xmath75 of a test particle at a distance @xmath266 , is @xmath268 the logaritmic slope of a density profile is defined as @xmath269 using ( [ eq : becrho ] ) in ( [ eq : alpha ] ) we obtain @xmath270.\ ] ] we study now the pseudo isothermal ( pi ) profile .
the empirical core profiles that exist in the literature fit two parameters , a scale radius and a scale density .
a characteristic profile of this type is @xmath271 this is the pi profile @xcite . here
@xmath272 is the scale radius and @xmath273 is the central density .
the rotation curve is @xmath274 the nfw profile emerges from numerical simulations that use only cdm and are based on the @xmath0cdm model @xcite .
in addition to this , we have chosen this profile because it is representative of what is called the cuspy behavior ( @xmath251 ) in the center of galaxies due to dm .
the nfw density profile@xcite and the rotation curve are given respectively by @xmath275 @xmath276},\ ] ] @xmath277 is related with the density of the universe at the moment the halo collapsed and @xmath278 is a characteristic radius . to solve the core radius ambiguity and to unify the concept for future comparison , in @xcite it was found that a good definition for the @xmath279 is a region where the density profile presents logarithmic slopes @xmath280 and the core radius will be the radius at which the core begins , that is to say , for radius smaller than the core radius we will have @xmath280 , this means that its value @xmath281 is determined by the equation @xmath282 the advantages of this definition are that the interpretation is independent of the profile chosen ( also notice that it applies to the total density profile and is not restricted to that of dm ) and in virtue of the same definition we can directly tell if a dm model profile is cored or cuspy . with this new definition the specific distance at which the core radius occurs still depends on the profile chosen but now the physical interpretation is only one . in the following when we refer to both the core and core radius
we adopt the previous interpretation . applying the definition to ( [ eq : becrho ] ) we get the core radius for the sfdm / bec profile @xmath283 and for comparison we use ( [ eq : pi ] ) because in turns out that the parameter @xmath272 corresponds to the core radius as defined above .
finally , fitting the nfw profile provides a direct comparison between a cusp and core and hence to the cusp / core problem .
we see from ( [ eq : becrho ] ) that the sfdm / bec model satisfies @xmath284 near the origin , but a priori this does not imply consistency with observed rcs .
therefore we reproduce the fit of the profiles to thirteen high resolution observed rcs of a sample of lsb galaxies given in @xcite .
the rcs were taken from a subsample of @xcite , we chose galaxies that have at least 3 values within @xmath285 kpc , not presenting bulbs and the quality in the rc in h@xmath286 is good as defined in @xcite .
the rcs in this work omit galaxies presenting high asymmetries and included in the error bars are experimental errors in the velocity measurement , inclination and small asymmetries . because the dm is the dominant mass component for these galaxies
we adopt the minimum disk hypothesis ( neglects baryon contribution to the observed rc ) . as the difference between a core and a cusp
is not overlapped only for data values inside @xmath287kpc and given that in the interval @xmath285 to @xmath288kpc the slopes of core and cusp profiles are very similar , which can lead to the wrong conclusion that cuspy halos are consistent with observations , we determined the logarithmic slope and the uncertainty following @xcite with the difference that we fit only the data within @xmath287 kpc and that there is no need of an uncertain `` break radius '' . in fig .
[ fig : rotcur ] , we show the fits to the rc data and the density profiles , also shown are the core radius in the bec ( magenta ) and in pi ( blue ) profiles .
the gray arrow is the fit that determines @xmath286 , the length denotes the fitted region and is bounded by @xmath289 that denotes the nearest radius to @xmath287kpc where a data point is given .
we show @xmath286 by fitting values inside @xmath289 and we also show the core radius for the sfdm / bec profile @xmath283 in order to compare it with @xmath272 ( see @xcite for more details of the fitting parameters ) . [ cols= "
< , < , < " , ] the fits of the rcs in fig [ fig : rotcur ] prove that the solid - body like behavior characterized by a linear increase in the velocity in the central region is more consistent with the core pi and sfdm / bec profiles than the cuspy nfw . the fits within @xmath289 give an average value of @[email protected] @xmath3 0.18 consistent with those obtained in @xcite @[email protected] @xmath3 0.2 and with @[email protected] @xmath3 0.07 reported by @xcite analyzing 7 things dwarf galaxies .
the case of eso1870510 might be considered to be consistent with nfw profile , however it is the innermost value that considerably decreases @xmath286 , being an irregular galaxy more central data near the innermost region is required to discard the possibility of any violent event that might have caused such a slope value .
the density profiles corresponding to the rcs fits are also shown in fig [ fig : rotcur ] for each galaxy .
we see that the sfdm / bec fit slightly deviates from the farthest data points as a result of having a finite radius @xmath262 that is fixed by the same data .
this discrepancy is due to the fact that the halo might be more extended than the value @xmath262 . as a matter of a fact , the more the extended the `` flat '' outer region in the rcs the more conspicuous the discrepancy .
@xcite recently found finite temperature corrections to ( [ eq : pi ] ) , this suffice to alleviate the latter problem in lsb galaxies and dwarfs but not for bigger galaxies .
some solutions have been proposed including vortex lattices @xcite and adding more nodes @xcite in the solution of system ( [ eq : bec1 ] ) and ( [ eq : bec2 ] ) ) , but so far no final conclusion has been reached .
when comparing the sfdm / bec and pi core radius we find a general difference of @xmath291kpc , the core size in the pi is approximately @xmath292 kpc smaller than the bec core size , but the pi central density is larger . in u4115 , u11557 and u11583
both profiles are very similar which results in a similar core and central density values , this can also be taken as a consistency check for our core definition . comparing the values of @xmath283 we did not find a tendency to a common value . assuming that the core radius determines the transition where the dm distribution changes from the outer region to the inner constant central density
, the lack of the unique value means that there is not a common radius at which this transition takes place .
for the second test @xcite uses @xmath283 to calculate ( [ eq : mu ] ) .
we have already seen that @xmath272 and @xmath283 are generally different and @xmath283 is not a fit parameter .
hence _ a priori _ @xmath283 is not expected to correlate with @xmath293 . with the fitting parameters it was obtained @xmath294 @xmath295 for the average values in the sfdm / bec and pi profiles respectively .
we see the excellent agreement of ( [ eq : mub ] ) with ( [ eq : mupi ] ) that was used as a crosscheck and with ( [ eq : logmu ] ) in which a much bigger sample was used .
the agreement has shown that the sfdm / bec model is capable of reproducing the constancy of the value @xmath243 , something that because of the cuspy nature is not possible in the nfw profile . in fig .
[ fig : logs ] we plot the above values for each galaxy .
if we define the dm central surface density mentioned in the introduction , we have for the bec profile @xmath296 where @xmath297 is obtained from ( [ eq : becm ] ) evaluated at @xmath283 .
we show that for u11748 the value @xmath298 is considerably above the rest and with the largest uncertainty .
for this reason , in the following analysis we omit both , this value and the smallest one that corresponds to eso1200211 . doing this
we get @xmath299 m@xmath300pc@xmath301 , and for the acceleration felt by a test particle located in @xmath283 due to dm only we have @xmath302 5.2 @xmath303 cms@xmath247 broadly consistent with ( [ eq : gdm ] ) . the fact that all galaxies present approximately the same order of magnitude in @xmath304 might suggest that @xmath283 represents more than a transition towards a constant density , it can give us information about the close relation between dm and the baryons . moreover , in view of the lack of a unique core radius , we can interpret the transition in dm distribution as an effect of crossing a certain acceleration scale instead of a radial length scale . such interpretation reminds us that given in mond but with the big difference that the acceleration scale found is for dm and is not a postulate of the model . to determine which interpretation causes the transition whether an acceleration scale or a length scale we will need to study the properties of larger samples of galaxies with the new telescopes .
in this work , we revisit an alternative paradigm to dark matter nature known as scalar field dark matter or bose - einstein condensate dark matter model . in this model a fundamental scalar field plays the role of dark matter in the universe .
the hypothesis is that this scalar field undergoes a phase transition very early in the universe leading to the formation of bose - einstein condensate drops .
therefore , the dark haloes of the galaxies in the universe are huge drops of scalar field . to explore if this hypothesis is a viable alternative to the standard model we developed both cosmological and galactic studies . in the cosmological regime , we have shown , with a field approach , that the sfdm / bec model with an ultralight mass of @xmath163 ev mimics the behavior of the cosmological expansion rate predicted with the @xmath0cdm model . moreover , although in general a scalar field is not a fluid , it can be treated as if it behaved like one .
thus , we performed a fluid approach for the cosmological evolution of the scalar field and we found that the analytical expressions for the kinetic and potential energies of the scalar field is in excellent agreement with our previous numerical results .
the interesting cosmological behavior of the scalar field indicates that their scalar fluctuations can be the appropriate for the purpose of structure formation , because overdense regions of sfdm / bec can support the formation of galactic structure .
thus , we have revisited the growth of sfdm / bec fluctuations in the linear regime . within the linear theory of scalar perturbations
we obtain an equation for the evolution of the density contrast .
this equation differs from the density contrast equation for cdm , however the extra terms tend to the values of the standard equation .
therefore , as a goal of this model , we obtained that the scalar perturbations grow up exactly as the cdm paradigm .
in addition , we study the liner growth of the scalar fluctuations with a fluid approach . in the case of the sfdm / bec with @xmath102
we have shown that for the matter dominated era and for big structure this model simulates the behavior of cdm because in general in a matter dominated universe for low-@xmath59 , @xmath305 tends to be a very small quantity tending to zero , so from ( [ delta ] ) we can see that on this era we will have the cdm profile given by ( [ deltacdm ] ) , i.e. , the sfdm density contrast profile is very similar to that of the @xmath0cdm model ( see fig .
[ fig2 ] ) . on the contrary for @xmath231
both models have different behavior as we can see from fig . [ fig3 ] .
the numerical results suggest that linear fluctuations on the sfdm can grow faster than those for cdm around @xmath306 . here
an important point is that although cdm can grow it does so in a hierarchical way , while from fig .
[ fig3 ] we can see that sfdm can have bigger fluctuations just before the @xmath0cdm model does , i.e. , it might be that no hierarchical model of structure formation is needed for sfdm and is expected that for the non - linear fluctuations the behaviour will be quite the same as soon as the scalar field condensates , in a very early epoch when the energy of the universe was about @xmath9 tev .
thus , the standard and the sfdm / bec model can be contrasted in their predictions concerning the formation of the first galaxies .
if in the future we see more and more well formed and massive galaxies at high redshifts , this could be also a new indication in favour of the sfdm / bec paradigm .
on the other hand , we also studied the implications of a sfdm / bec model at galactic scales .
we find that the sfdm / bec model gives a constant density profile that is consistent with rcs of dark matter dominated galaxies .
the profile is as good as one of the most frequently used empirical core profiles but with the advantage of coming from a solid theoretical frame .
we fit data within @xmath287kpc and found a logarithmic slope @xmath307 0.27@xmath3 0.18 in perfect agreement with a core .
it is important to notice that the cusp in the central regions is not a prediction that comes from first principles in the cdm model , it is a property that is derived by fitting simulations that use only dm .
in addition , we review a new definition for the core and core radius that allows a definite distinction when a density profile is core or cusp . using this definition
we find the core radius in the sfdm / bec profile to be in most cases over @xmath292kpc bigger than the core radius in the pi profile . as a second result of the core definition ,
we show the constant value of @xmath243 which is proportional to the central surface density .
this result is one of several conflicts that jeopardize the current standard cosmological model .
finally , with all these intriguing results sfdm / bec model could be a serious alternative to the dark matter problem in the universe .
although the observational evidence seems to be in favor of some kind of cold dark matter , if we continue to observe even more galaxies at higher redshifts and if higher resolution observations of nearby galaxies exhibit a core density profile , this model can be a good alternative to @xmath0cdm .
we expect that the close future observations in galaxies surveys can decide the nature of the dark matter .
this work was partially supported by conacyt mxico under grants cb-2009 - 01 , no .
132400 , no . 166212 , and i0101/131/07 c-234/07 of the instituto avanzado de cosmologia ( iac ) collaboration ( http://www.iac.edu.mx/ ) .
arbey , a. , lesgourgues , j. and salati , p. , 2001
d * 64 * 123528 arbey , a. , lesgourgues , j. and salati , p. , 2002
d * 65 * 083514 arbey , a. , lesgourgues , j. and salati , p. , 2003
d * 68 * 023511 arbey , a. , 2006 eas publ .
ser . * 20 * 257 arbey , a. , 2006b phys .
d * 74 * 043516 arbey , a. , 2009 eas publ .
ser . * 36 * 161 athanassoula e. , bosma a. , papaioannou s. , 1987 , a&a , 179 , 2340 guzmn , f. siddhartha and matos , tonatiuh .
astron . nachr .
guzman f. s. , matos t. , 2000 , class .
quantum grav . , 17 , l9-l16 .
guzmn f.s .
, urea - lpez l. a. , 2003 , phys
d 68 * , 024023 .
guzmn f. s. , urea - lpez l.a .
, 2004 , phys . rev . *
d 69 * , 124033 .
guzmn f. s. , urea - lpez l. a. , 2006 , apj , 645 , 814 .
harko t. , 2011 , mnras , 413 , 3095 harko t. , 2011 , jcap , 05 , 022 harko t. , 2011 , phys .
d , 83 , 123515 harko t. , madarassy e. j. m. , 2012 , jcap , 01 , 020 .
hu w. , barkana r. , gruzinov a. , 2000 , phys .
lett . , 85 , 1158 | in the last time the cold dark matter ( cdm ) model has suggested more and more that it is not able to describe all the properties of nearby galaxies that can be observed in great detail as well as that it has some problems in the mechanism by which matter is more rapidly gathered into large - scale structure such as galaxies and clusters of galaxies . in this work
we revisit an alternative model , the scalar field dark matter ( sfdm ) model , which proposes that the galactic haloes form by condensation of a scalar field ( sf ) very early in the universe , i.e. , in this model the haloes of galaxies are astronomical bose - einstein condensate drops of sf . on the other hand , large - scale structures like clusters or superclusters of galaxies form similar to the @xmath0cdm model , by hierarchy ,
thus all the predictions of the @xmath0cdm model at cosmological scales are reproduced by sfdm .
this model predicts that all galaxy haloes must be very similar and exist for higher redshifts than in the @xmath0cdm model . in the first part of this review we revisit the cosmological evolution of sfdm model with a scalar potential @xmath1 with two different frameworks : the field and fluid approach .
we derive the evolution equations of the sf in the linear regime of perturbations as well .
the scalar fluctuations have an oscillating growing mode and therefore , this kind of dark matter could lead to the early formation of gravitational structures in the universe .
we also revisit how bec dark matter haloes exhibit a natural cut of the mass power spectrum . in the last part
, we study the core central density profiles of bec dark matter haloes and fit high - resolution rotation curves , we show a sample of some low surface brightness galaxies .
the mean value of the logarithmic inner density slopes is @xmath20.27 @xmath3 0.18 . using a model independent new definition of the core in the bec density profile
, we show that the recent observation of the constant dark matter central surface density can be reproduced .
we conclude that in light of the difficulties that the @xmath0cdm model is currently facing the sfdm model can be a worthy alternative to keep exploring further . |
the observed hierarchy in the masses and mixings of quarks and leptons is one of the most puzzling features of nature . a plausible explanation is provided by flavor dependent gauge symmetries . in a popular class of such models
one extends the standard model ( sm ) to include a family
dependent @xmath0 factor . upon spontaneous breaking of the @xmath0 symmetry effective yukawa couplings of the form @xmath5 are induced , where @xmath6 is a small parameter , and @xmath7 are positive integers related to the family dependent @xmath0 charges . even when the fundamental yukawa couplings @xmath8 are all of order one , a hierarchical spectrum is realized due to the suppression in powers of @xmath9 @xcite .
such flavor @xmath0 symmetries can be naturally identified with the anomalous @xmath10 symmetry of string theory @xcite .
models using anomalous @xmath10 symmetry for fermion mass and mixing hierarchy abound in the literature @xcite@xcite .
most models of this type also assume low energy supersymmetry ( susy ) to stabilize the higgs boson mass .
novel phenomena which are amenable to experimental tests can arise in such contexts .
the purpose of this paper is to analyze one such effect , viz . , the electric dipole moments of elementary fermions @xcite .
low energy supersymmetry can potentially induce excessive flavor violation in processes such as @xmath11 mixing and @xmath12 decay if the soft supersymmetry breaking lagrangian takes its most general form .
this potential problem is usually avoided by assuming a universal form for the soft susy breaking terms .
even with universality , the @xmath13violating phases present in the soft susy breaking lagrangian can induce electric dipole moments ( edm ) for the neutron and the electron at a level exceeding the current experimental limits .
these effects have been extensively studied in the literature @xcite@xcite .
we will assume in the present work a universal susy breaking spectrum that is also @xmath13invariant so that excessive edms are not induced from the fundamental soft susy breaking parameters .
the edms that we find in the context of models of fermion mass hierarchy are induced purely by complex yukawa couplings .
the phases in the yukawa couplings are believed to be the source for the observed @xmath13violation in the @xmath14 and @xmath15 meson systems ( ckm @xmath13violation ) .
it is thus reasonable to assume all yukawa couplings , including the leptonic yukawa couplings , to be complex . as we will see , it is natural that the flavor @xmath0 gauge symmetry responsible for explaining the fermion mass hierarchy breaks spontaneously at a scale @xmath16 slightly below the fundamental plank ( or string ) scale , @xmath17 . in the momentum regime
@xmath18 the flavor gauge sector will be active and will contribute to flavor violation in the squark and slepton sectors .
these effects would survive down to the susy breaking scale and can lead to observable phenomena . in a previous paper we have studied
leptonic rare decays @xmath19 and @xmath20 induced by the flavor gauge sector @xcite . with complex yukawa couplings ,
this flavor violation will also lead to edms for the electron ( @xmath21 ) , muon ( @xmath22 ) , the neutron ( @xmath23 ) , and the deuteron ( @xmath24 ) even with universal and @xmath13conserving soft susy breaking terms at the string scale . in a popular class of anomalous @xmath0 models which explains the fermion mass hierarchy , including bi large neutrino mixing , we find @xmath25 cm , @xmath2 cm , @xmath26 cm , and @xmath4 cm , which are within reach of next generation experiments
. there are proposals to improve the current limit on electron edm , @xmath27 cm @xcite , by about two to four orders of magnitude @xcite .
it is expected that the current limit on the muon edm , @xmath28 cm , will be improved by six orders of magnitude or even more in the not too distant future @xcite .
there are also proposals which would improve the current neutron edm limit from @xmath29 cm @xcite by a factor of 5 @xcite .
the deuteron edm is expected to be probed to the level of @xmath30 cm in the near future @xcite .
supersymmetry may reveal itself in these experiments before direct discovery at lhc , if the current ideas of solving the fermion mass hierarchy problem are correct .
lepton edms may arise even without flavor gauge symmetry from complex neutrino yukawa couplings responsible for the seesaw mechanism in the context of low energy susy .
this effect has received much attention recently @xcite .
we have computed such effects for @xmath21 and @xmath22 , but found them to be much less significant compared to the flavor @xmath0 induced effects .
for example , we find @xmath31 cm for large @xmath32 from the neutrino yukawa coupling effects , to be compared with @xmath33 cm from the flavor @xmath0 sector .
similar effects from gut threshold has been studied in ref .
@xcite .
the paper is organized as follows . in section 2
we review the class of models based on anomalous @xmath10 symmetry for fermion mass hierarchy @xcite . in section 3 the edms induced by the flavor @xmath10 gauge sector is analyzed . in 3.1 a qualitative discussion of the radiative corrections to the soft masses and @xmath34terms is given . in 3.2
we present our full numerical results for the edms .
section 4 has our conclusions . in appendix [ a1 ] we give the relevant expressions for the @xmath35functions for the soft susy breaking parameters including corrections from the @xmath10 gauge sector . in appendix [ a2 ]
the fermion mass fit for the model used in the numerical analysis is presented .
appendix [ a3 ] lists the formulas needed for the calculation of edms .
in this section we review briefly the idea of explaining fermion mass hierarchy with a flavor dependent @xmath0 symmetry .
we focus on a specific class of anomalous @xmath10 models discussed in ref .
@xcite to address the fermion edm .
most models of ref .
@xcite@xcite will also fall into this category and will lead to similar results . in these models families
are distinguished by their anomalous @xmath0 charges .
the @xmath10 symmetry is broken spontaneously by an mssm singlet flavon field @xmath36 which acquires a vacuum expectation value ( vev ) slightly below the string scale @xmath37 .
this provides a small expansion parameter @xmath38 needed for explaining the fermion mass hierarchy .
@xmath0 invariance forbids renormalizable yukawa couplings for the light families , but would allow them through effective nonrenormalizable couplings suppressed by a factor @xmath39 ( for the fermion mass operator connecting flavors @xmath40 and @xmath41 ) with @xmath7 being positive integers . even with all couplings being of order one , hierarchical masses for different flavors are naturally realized @xcite .
although this mechanism will work with any flavor @xmath0 , anomalous @xmath0 models are attractive since they would also provide a natural understanding for the smallness of @xmath42 @xcite , which arises from the one loop induced fayet
illiopoulos @xmath43term @xcite .
consider the following fermion mass matrices studied in ref @xcite : @xmath44 here @xmath45 , @xmath46 and @xmath47 are the up
quark , the down quark and the charged lepton mass matrices in the basis @xmath48 .
complex order one coefficients multiplying each entry of the matrices are not shown . @xmath49 and @xmath50 are the neutrino dirac and majorana mass matrices . the light neutrino mass matrix @xmath51 is derived from the seesaw mechanism . @xmath52 and
@xmath53 are integers and are chosen differently for different values of @xmath54 .
the choice of @xmath55 corresponds to large ( @xmath56 ) , medium ( @xmath57 ) and small ( @xmath58 ) values of @xmath32 respectively .
the quark and lepton masses and mixings arising from eq .
( [ massm1 ] ) are fully consistent with experimental observations if @xmath42 .
note that the ckm mixing angles are small , while the leptonic mixing angles relevant for solar and atmospheric neutrino oscillations are of order unity .
the lopsided nature of the matrices @xmath46 and @xmath47 of eq .
( [ massm1 ] ) enables such a disparity to be realized @xcite .
the superpotential which would lead to the yukawa coupling structure in eq .
( [ massm1 ] ) has the following general form near the fundamental scale @xmath37 : @xmath59 here @xmath60 are the generation indices and @xmath61 are positive integers fixed by the choice of the @xmath10 charge assignment .
we choose all charges to be integers with the charge of @xmath36 being negative .
@xmath62 are the yukawa couplings which we take to be complex and of order one for all @xmath63 .
@xmath64 stands for the mssm higgs doublets @xmath65 and @xmath66 .
@xmath36 is the mssm singlet flavon field whose vev determines the expansion parameter @xmath9 .
@xmath67 in eq .
( [ superp1 ] ) contains mssm singlet fields @xmath68 which would be needed for anomaly cancelation .
the @xmath0 charge assignment shown in table [ tcharge ] will lead to the texture of eq .
( [ massm1 ] ) . ' '' '' field & @xmath10 charge & charge notation + ' '' '' @xmath69 , @xmath70 , @xmath71 & @xmath72 & @xmath73 + ' '' '' @xmath74 , @xmath75 , @xmath76&@xmath77&@xmath78 + ' '' '' @xmath79 , @xmath80 , @xmath81&@xmath82&@xmath83 + ' '' '' @xmath84 , @xmath85 , @xmath86&@xmath87 & @xmath88 + ' '' '' @xmath89,@xmath90,@xmath91&@xmath92&@xmath93 + ' '' '' @xmath94 , @xmath95 , @xmath96 & @xmath97&@xmath98 + ' '' '' @xmath65 , @xmath66 , @xmath36&@xmath99&@xmath100 + the @xmath10 anomalies are cancelled by the green schwarz mechanism @xcite which requires @xmath101 here @xmath102 , @xmath103 , @xmath104 , @xmath105 and @xmath106 are @xmath107 , @xmath108 , @xmath109 , @xmath110 and @xmath111 anomaly coefficients . all other anomalies ( such as @xmath112 ) must vanish .
@xmath113 , @xmath114 are the kac - moody levels , with the non abelian levels @xmath115 and @xmath116 being integers . the factor @xmath117 in front of the cubic anomaly @xmath105 has a combinatorial origin owing to the three identical @xmath10 gauge boson legs .
we require string unification of all the gauge couplings including that of the @xmath10 , @xmath118 , at the fundamental scale @xmath37 @xcite : @xmath119 for a clear discussion of the coefficients in eqs .
( [ anomalygs])([unifgs ] ) see ref .
@xcite . here
@xmath120 are the @xmath121 , @xmath122 and @xmath123 gauge couplings for @xmath124 . with the choice @xmath125 ,
consistency with the observed unification of gauge couplings within mssm would require @xmath126 , corresponding to the @xmath127 normalization of hypercharge .
this will be an automatic consequence of the green schwarz anomaly cancelation conditions with our choice of texture .
the small discrepancy ( in the absence of a covering gut group ) between the unification scale derived from low energy data and the string scale may be understood in the context of @xmath128theory by making use of the radius of the eleventh dimension @xcite . with @xmath125
we find from table [ tcharge ] , @xmath129 and @xmath130 .
( [ anomalygs ] ) then requires @xmath131 , i.e. , a common exponent for the charged lepton and the neutrino dirac yukawa coupling matrices . with @xmath131 ,
the condition @xmath132 fixes @xmath133 to be @xmath134 , consistent with @xmath127 unification .
note also that the charges given in table [ tcharge ] become compatible with @xmath127 unification .
since @xmath135 for the fermion multiplets of @xmath127 , and since the higgs doublets carry zero @xmath10 charge , the anomaly coefficient @xmath136 ^ 2 \times u(1)_y$ ] vanishes , as required . the last equality in eq .
( [ anomalygs ] ) requires @xmath137 this can not be satisfied with the mssm fields alone , since @xmath138 , which does not match eq .
( [ agravity ] ) .
we cancel this anomaly by introducing mssm singlet fields @xmath68 obeying @xmath139 .
if all the @xmath68 fields have the same charge equal to @xmath140 , they will acquire masses of order @xmath141 through the coupling @xmath142 and will decouple from low energy theory . for other choices of the charge of @xmath68 these masses can be different .
for example , if the charge is equal to @xmath143 , their masses will be of order @xmath144 ; if the charge is @xmath145 the masses will be of order @xmath146 .
we will consider only the case where the @xmath68 fields have charge @xmath140 .
with the charges of all fields fixed , we are now in a position to determine the @xmath10 charge normalization so that @xmath147 at the string scale , ( we take @xmath125 . ) this normalization factor , which we denote as @xmath148 , is given by @xmath149 . all the charges given in table [ tcharge ] are to be multiplied by @xmath148 . from the green schwarz anomaly cancelation condition @xmath150 , we have @xmath151 from which we find the normalization of the @xmath10 charge @xmath149 to be @xmath152 the fayet iliopoulos term for the anomalous @xmath10 , generated through the gravitational anomaly , is given by @xcite @xmath153 where @xmath154 is the unified gauge coupling at the string scale ( see eq .
( [ unifgs ] ) ) . by minimizing the potential from the @xmath10 @xmath43term @xmath155 in such a way that supersymmetry remains unbroken
, one finds for the vev of @xmath36 @xmath156 for the fermion mass texture in eq .
( [ massm1 ] ) , corresponding to the @xmath10 charges given in table [ tcharge ] , we find @xmath157 the masses of the @xmath10 gauge boson and the corresponding gaugino are obtained from @xmath158 and found to be @xmath159 in the momentum range below @xmath37 and above @xmath16 , these gauge particles will be active and will induce flavor dependent corrections to the sfermion soft masses and the @xmath34terms .
it is these effects which induce edms for the electron , muon and the neutron at low energies .
in the standard model the electric dipole moments of the electron , muon and the neutron are predicted to be extremely small and beyond reach of planned experiments . in the presence of low energy supersymmetry
these edms can exceed the current experimental limits if soft susy breaking parameters are complex @xcite@xcite . to focus on the anomalous
@xmath0 induced effects we shall adopt the minimal supergravity scenario with universal and @xmath13conserving soft susy breaking parameters . specifically , at the string scale we assume a universal scalar mass @xmath160 , a common gaugino mass @xmath161 , and trilinear @xmath34terms proportional to their respective yukawa couplings .
we assume @xmath160 , @xmath161 and @xmath162 , and the higgs mass parameters @xmath163 and @xmath164 to be real .
thus the only source of @xmath13violation is in the complex yukawa couplings .
this is needed for the ckm @xmath13violation in the quark sector and it is natural to assume that the leptonic yukawa couplings are complex as well . in sec .
3.1 we give a qualitative estimate of the edms induced by the radiative corrections involving the @xmath10 gauge sector in such a susy context .
our numerical results are presented in sec .
we now give approximate expressions for the @xmath10 gauge sector rge corrections to the soft parameters between the string scale and the @xmath10 breaking scale @xmath16 .
the full rge expressions for the soft parameters in the presence of higher dimensional operators as in eq .
( [ superp1 ] ) have been derived in ref . @xcite . in appendix
[ a1 ] we summarize the relevant expressions .
the @xmath10 corrections to the soft masses for the left
handed slepton are obtained from eq .
( [ betamf ] ) to be @xmath165 and a similar expression for the right handed slepton masses with the interchange @xmath166 @xmath167 .
there are analogous corrections in the squark sector .
the corrections to the @xmath34terms are obtained from eq .
( [ beta2 ] ) as @xmath168 where @xmath169 are biliear combinations of the flavor charges given by @xcite @xmath170 numerical values of @xmath171 for our model are given in eq .
( [ ze1 ] ) of appendix [ a1 ] .
note that these corrections , eqs .
( [ delms ] ) and ( [ dela ] ) , are flavor dependent .
due to the flavor dependent nature of these corrections , the fermion and the corresponding sfermion mass matrices can not be diagonalized simultaneously .
this was the source of the flavor violation studied in ref .
for the same reason , with complex yukawa couplings @xmath172 , nonzero edms for the fermions will be induced .
let us now estimate the edm of the electron arising from the corrections in eqs .
( [ delms ] ) and ( [ dela ] ) .
there are three flavor dependent matrices in the leptonic sector , not including the neutrino yukawa matrix @xmath173 .
they are the leptonic yukawa matrix @xmath174 and the matrices of @xmath10 charges @xmath175 and @xmath176 for the lepton doublets and singlets ( see table [ tcharge ] ) . in the mass eigenbasis for
the charged leptons @xmath177 and @xmath178 will develop complex off diagonal entries , with the phases arising from @xmath174 through the unitary matrices that diagonalize @xmath174 .
this is the basic source for the edm .
the corrections given in eq .
( [ delms ] ) will generate edm of the electron through the product of slepton mixings in @xmath179 , @xmath180 and @xmath181 ( for @xmath182 ) .
the induced edm will be @xmath183 $ ] , where @xmath184 and @xmath185 are unitary matrices which diagonalize @xmath174 , @xmath186 .
there are additional corrections which are quadratic in @xmath177 and @xmath178 .
the corrections to the @xmath34terms in eq .
( [ dela ] ) will also induce edm directly through @xmath187 mixings . combining these effects with the formula for the edm given in eq .
( [ formuledm ] ) of appendix [ a3 ] we arrive at the following approximate expression for @xmath21 : @xmath188\mbox{im}\left[\frac{y^e_{1i}y^e_{i1}}{y^e_{ii}}\right],\end{aligned}\ ] ] where @xmath189 and @xmath190 denote the contributions from the soft masses and the @xmath34terms respectively .
they are given by @xmath191 here @xmath192 is obtained from @xmath193 by the replacement @xmath194 .
@xmath195 is the average slepton mass and @xmath196 is the bino mass . the function @xmath197 is given in eq .
( [ funcedm ] ) in appendix [ a3 ] .
we see explicitly that the complex yukawa couplings along with nonuniversal @xmath10 charges lead to nonzero edm . to estimate the size of this effect we choose the approximations @xmath198 . following the mass matrices given in eq .
( [ massm1 ] ) we take @xmath199 .
we consider here only the contribution from the @xmath200 mixing , since the @xmath10 charge difference is the largest between the first and the third generations in @xmath178 .
then we find @xmath201.\end{aligned}\ ] ] from this estimate we see that the electron edm induced by the @xmath10 gauge corrections is in the experimentally interesting range and already puts constraint on the soft susy breaking parameters .
the actual numerical result is quite sensitive to the choice of @xmath160 and @xmath161 . in our numerical calculations
we have chosen @xmath202 for low @xmath32 for cosmological reason . in this case the @xmath203 coefficient in eq .
( [ edmapprx2 ] ) will be reduced to an @xmath204 number . for large @xmath32
this coefficient will remain as @xmath203 .
let us now compare the anomalous @xmath0 induced edm with the right handed neutrino induced effects pointed out in ref . @xcite and studied further in ref .
the latter effects induce edm which are given by @xmath205,\end{aligned}\ ] ] where @xmath206 . here
@xmath207 is the mass of the right handed neutrino of flavor @xmath40 . with our texture for the neutrino mass matrices dictated by @xmath10 symmetry
we find the right handed neutrino induced edm to be @xmath208 cm , which is two to three orders magnitude smaller than the anomalous @xmath10 induced effects . in our numerical analysis
we present separately our results for the electron edm arising from the right handed neutrino effects .
in this sub - section we present our numerical results for the electron , muon , neutron and the deuteron electric dipole moments .
we adopt the minimal supergravity scenario for supersymmetry breaking . at the string scale ,
taken to be @xmath209 gev , we assume a universal scalar mass @xmath160 and a common gaugino soft mass @xmath161 .
all susy breaking parameters ( @xmath210 ) and the @xmath163 term are taken to be real at the string scale .
we choose @xmath211 for all cases except in fig .
[ fig1 ] where we also show results for @xmath212 .
the anomalous @xmath0 gauge coupling @xmath118 is chosen to be @xmath213 , consistent with string unification .
the soft susy breaking parameters are evolved from @xmath37 to the @xmath10 breaking scale @xmath214 ( see eq .
( [ mam1 ] ) ) including the @xmath10 gaugino / gauge boson corrections .
we present our results for the edm for three values of the parameter @xmath32 , small ( @xmath215 ) , medium @xmath216 and large @xmath217 .
we take @xmath202 for low and medium values of @xmath32 .
this is motivated by the requirement that the right abundance of cosmological cold dark matter be generated . with @xmath202 , the right handed charged sleptons will have masses slightly above that of the neutralino lsp .
the relic abundance of neutralinos is in the right range with such a spectrum , as a result of coannihilation @xcite . for large @xmath32
we also allow the choice @xmath218 , since alternative mechanisms for reproducing the right relic abundance of lsp become available in this case @xcite .
we vary @xmath161 in the range @xmath219 gev to @xmath220 tev .
the results are presented for two different values of @xmath162 , @xmath221 and @xmath222 gev .
the lepton edms induced by the flavor @xmath0 gaugino / gauge boson contribution are plotted against the universal gaugino mass @xmath161 in figures [ fig1]-[fig2 ] . in figure [ fig3 ]
the electron edm induced purely by the right handed neutrino effects is plotted . in figure [ fig4]-[fig5 ]
we plot the edm of the neutron and the deuteron arising from the flavor @xmath0 gauge boson / gaugino effects . as input at @xmath37
we choose the yukawa coupling matrices given in eqs .
( [ yukup])([yukneutrino ] ) of appendix [ a2 ] ( for @xmath223 ) .
these are obtained by extrapolating the low energy yukawa couplings to @xmath37 and applying bi - unitary transformations at @xmath37 to generate the texture given in eq .
( [ massm1 ] ) .
the low energy yukawa couplings and their extrapolation are discussed in appendix [ a2 ] . as for the neutrino dirac yukawa couplings ,
we choose @xmath173 to be such that in the flavor basis ( after the bi unitary rotations ) it exhibits approximately the structure given in eq .
( [ massm1 ] ) with @xmath224 . for a given choice of hierarchical light
neutrino spectrum this would uniquely fix the right handed neutrino mass matrix through the seesaw mechanism .
@xmath50 will then have the form given in eq .
( [ massm1 ] ) .
we set @xmath225 with @xmath226 gev .
the eigenvalues of the right handed neutrino mass matrix are important for the lepton edms induced by the right handed neutrino threshold effects .
it should be noted that the unitary rotations applied on the diagonal yukawa matrices at @xmath37 are not unique , except that they should conform to the fermion mass matrix structure shown in eq .
( [ massm1 ] ) .
so our fits should be taken only as indicative , and not definitive .
we expect differences of order one in our numerical results on edm arising from the arbitrariness in these unitary matrices . in figure [ fig1 ] the electron edm induced by the @xmath10 gaugino / gauge boson contributions to the soft masses and @xmath34terms
are plotted as a function of @xmath161 for three values of @xmath32 .
we see that some parts of the parameter space are already excluded by the current experimental upper bound @xmath227 cm and that the other parts are in the range which will be tested by next generation electron edm experiments @xcite . in figure [ fig2 ]
we plot the muon edm as a function of susy breaking parameters .
we find @xmath22 to be in the range @xmath228 cm for most of the parameter space .
this value is somewhat smaller than than @xmath229 , which would be the naive expectation based on the scaling of lepton masses .
this happens for the following reason .
the second and third family left
handed charged sleptons have the same @xmath10 charge , so the flavor gauge bosons / gaugino will not generate any mass splitting between these sleptons .
the mixing in the right
handed charged slepton sector is suppressed by a factor @xmath230 for all @xmath32 , compared to the suppression factor @xmath9 between the first and the second generations .
on the other hand , we find quite an enhancement of the muon edm for the choice @xmath218 and @xmath231 .
for this choice , the electron edm is well above the experimental bound . since the two edms are induced by independent phases , it is possible to choose the parameters such that the electron edm is below the experimental limit and at the same time the muon edm is at the level of @xmath232 cm , although we do not attempt such an explicit solution here .
it should also be pointed out that parts of the parameter space where @xmath22 is large is already ruled out by the experimental upper limit for the radiative decay @xmath12 for the numerical fits shown @xcite .
the remaining regions will be put to experimental scrutiny by future experiments @xcite . in figure [ fig3 ]
we present for comparison , the electron edm arising solely from the right handed neutrino threshold effects @xcite . with the proper decoupling of the right handed neutrinos @xcite we find our results to be in rough agreement with those in ref .
nevertheless these effects , which yield at most @xmath233 , are much smaller compared to the @xmath10 effects .
in figure [ fig4 ] we plot the neutron edm versus @xmath161 . in figure [ fig5 ]
we plot the deuteron edm .
details of the calculations are given in appendix [ a3 ] .
in both cases our numerical results are in the interesting range which should be accessible to proposed experiments in the near future .
we find the contributions from the ckm phase to be of the same order as the contributions from the @xmath10 gaugino / gauge boson sector .
figures [ fig4][fig5 ] include both these effects .
the flavor sector contribution to the neutron edm is somewhat smaller compared to the leptonic edm due to the gluino focusing effect .
( the squarks receive flavor universal contributions for their masses below @xmath16 from the gluino , which tends to suppress flavor violation and thus @xmath23 . )
we have also studied the constraint on the chromoelectric dipole moment for the strange quark @xmath234 arising from @xmath235hg edm @xcite .
this bound reads as @xmath236 this constraint is easily satisfied in our model .
the down type squark mixing in the @xmath237 sector is suppressed by a factor @xmath230 for the right handed squarks , and is vanishing to leading order for the left
handed squarks , similar to the case of @xmath238 mixing .
consequently , we find the chromoelectric edm of the strange quark to be about two to three orders of magnitude below the experimental limit . the soft susy breaking bilinear @xmath15term and the gaugino masses will develop complex phases via the one loop and two loop rge corrections respectively arising from the @xmath34term contributions . in our model we find these corrections to be negligible compared to the @xmath10 flavor gaugino / gauge boson effects .
in this paper we have studied the electric dipole moments of the electron , muon and the neutron induced by a flavor dependent @xmath0 symmetry which explains the hierarchy of fermion masses and mixings in a natural way via the froggatt - nielsen mechanism .
this @xmath0 symmetry may be identified as the anomalous @xmath0 of string theory .
this symmetry is broken spontaneously at a scale @xmath16 slightly below the string scale , @xmath239 . in the momentum regime @xmath240
, the flavor @xmath10 gauge boson sector will be active and will contribute to the soft susy breaking parameters in a flavor dependent fashion .
we adopt the minimal supergravity scenario for susy breaking , and assume that the soft susy breaking parameters are universal and real .
the complex yukawa couplings will still induce phases in the soft susy masses and the @xmath34-parameters , leading to the generation of edm .
this is the main source of the edm that we have studied here .
we have presented our numerical results for the electron , muon , neutron and the deuteron edms in figures [ fig1][fig5 ] as functions of supersymmetry breaking parameters . @xmath21 and @xmath23 are very close to the current experimental limits , @xmath241 cm and @xmath242 cm . for the deuteron ,
our prediction , @xmath4 cm , would make it within reach of proposed experiments .
for the case of the muon , although @xmath22 is rather small for low @xmath32 , in the case of large @xmath243 , for certain choices of phases in the yuakawa couplings , we have found the induced edm to be as large as @xmath244 cm , which might be accessible to future experiments @xcite . in the leptonic sector , these edms are much larger than the ones induced by the neutrino seesaw sector , which yields , for example , @xmath245 cm with our texture of fermion mass matrices dictated by flavor symmetries . in figure [ fig4 ]
we present our results for the induced @xmath21 arising from the neutrino seesaw sector .
discovery of electric dipole moments for the electron , muon and the neutron can shed light on one of the fundamental puzzles of nature , viz . , the origin of mass for elementary particles .
we thank professor lee roberts and the physics department of boston university for hospitality during our participation in the international symposium on lepton moments " , cape cod , ma , june 2003 . ts .
e. also acknowledges financial support for participation . this work is supported in part by doe grant
# de - fg02 - 04er46140 , and an award from the research corporation .
the one loop @xmath35functions for the soft sfermion masses are given by @xmath246 where the superscripts @xmath247 , @xmath34 and @xmath248 denote the contributions from the mssm sector , the anomalous @xmath0 gauge sector , and the right handed neutrino sector respectively .
@xmath249 are present only for the left handed slepton and the sneutrino soft masses and their explicit form is given elsewhere .
the anomalous @xmath0 part @xmath250 is @xcite @xmath251 here @xmath252 is defined as @xmath253 where @xmath254 are the soft masses of the extra particles @xmath68 introduced for anomaly cancelation via the green schwarz mechanism .
the trace is taken over family space . here
@xmath255 etc . , stand for the mssm @xmath35functions without the neutrino and the flavor @xmath10 contributions . introducing a notation @xmath256 the @xmath0 gaugino part of @xmath34term @xmath35functions
is given by @xmath257 where @xmath118 and @xmath258 are the @xmath10 gauge coupling and the gaugino mass respectively .
@xmath259 is the @xmath10 charge of the up type ( down type ) higgs doublet if @xmath260 is up type ( down type ) .
here we defined the combination of the @xmath10 charges @xmath261 as @xmath262 from the @xmath10 charge assignments for the mssm fields as given in table [ tcharge ] one has @xmath263
@xmath265 for the three different values of @xmath266 and @xmath267 independent of @xmath52 .
the terms we are interested in are the ones proportional to the @xmath10 gaugino mass @xmath258 and the term proportional to @xmath252 in the @xmath35functions . beside these ,
the @xmath34term @xmath35functions contain a flavor dependent piece which arises from the wave function renormalization .
since the corresponding yukawa @xmath35functions contain the same terms , these are simultaneously diagonalized , and do not lead to flavor violation . here
we present the numerical fits to the fermion masses and mixings adopted for the calculation of the edms . as input at low energy we choose the following values for the running quark masses @xcite : @xmath268 the ckm mixing matrix is chosen in the standard parametrization with @xmath269 , @xmath270 , @xmath271 and the complex phase @xmath272 .
we use two loop qed and qcd renormalization group equations to evolve these masses from the low energy scale to the scale of susy breaking .
we obtain the following running factor @xmath273 for the fermion masses at the susy breaking scale @xmath274 , initially chosen to be @xmath275 gev with @xmath276 : @xmath277 then these masses at @xmath274 are used to calculate the yukawa couplings in @xmath278 scheme .
using one loop susy rge evolution above @xmath274 we obtain the yukawa couplings at the @xmath10 breaking scale ( @xmath279 gev ) to be @xmath280 for @xmath223 , @xmath281 for @xmath282 and @xmath283 for @xmath231 .
we have chosen the dirac neutrino yukawa couplings such that in the flavor basis ( after the bi unitary rotations ) it exhibits the assumed structure in eq .
( [ massm1 ] ) . the light
neutrino masses and the lepton mixing matrix are chosen to be @xmath284 ev , @xmath285 ev and @xmath286 ev and @xmath287 which are compatible with the observed light neutrino oscillation parameters . from this
the right handed neutrino mass matrix is fixed uniquely with @xmath288 where @xmath289 gev .
the eigenvalues of @xmath50 are used to calculate the lepton edm induced by the right handed neutrino threshold effects . in the following , we present our fits to the texture of eq .
( [ massm1 ] ) which have been used in our numerical calculations for @xmath223 ( we have similar fits upto an overall factor for @xmath290 ) .
this fit is not unique , and so can lead to order one uncertainty in our edm results .
the following fit is found by applying bi - unitary transformations with complex phases on the diagonal yukawa coupling matrices at the @xmath10 breaking scale @xmath16 .
we introduce a notation for the yukawa couplings : @xmath291 at @xmath16 we find the following fit ( for @xmath292 : @xmath293 @xmath294 @xmath295 @xmath296 note that all coefficients multiplying @xmath297 in eqs .
( [ yukup])([yukneutrino ] ) are of order unity .
the edms of elementary fermions are sum of neutralino , chargino and for quarks gluino contributions which we denote as @xmath298 , @xmath299 and @xmath300 .
in addition to these , the quarks receive contributions from chromoelectric and purely gluonic dimension six operators @xcite .
we have not considered here the latter one , since these effects turn out to be small . the effective edm operator @xmath301 for a spin@xmath302 particle is given by @xmath303 the edm @xmath301 in general has the following components in a supersymmetric theory : @xmath304 where @xmath305 here @xmath306 , @xmath307 and @xmath308 are the neutralino , chargino and the gluino masses respectively .
@xmath309 ( @xmath310 ) are the eigenvalues of the sfermion mass matrices .
the coefficients @xmath311 , @xmath312 and @xmath313 are given by @xmath314{\nonumber}\\ & \times&\left[-\sqrt{2}\left\{\tan\theta_w\left(q_{f_i}- t_{f_i}\right)\left(o^n\right)_{a1}+t_{3f_i}\left(o^n\right)_{a2}\right\ } u^{f*}_{i , x}\right.{\nonumber}\\&-&\left .
k_f\left(o^n\right)_{a^\prime a}u^{f*}_{i+3,x}\right],{\nonumber}\\ c^u_{xb}&=&k_u\left(o^c_r\right)^*_{b2}u^d_{1,x } \left[\left(o^c_l\right)_{b1}u^{d}_{4,x}-k_d\left(o^c_l\right)_{b 2}u^{d}_{4,x}\right]^*,{\nonumber}\\ c^d_{xb}&=&k_d\left(o^c_l\right)^*_{b2}u^u_{i1 } \left[\left(o^c_r\right)_{b1}u^{u}_{1,x}-k_u\left(o^c_r\right)_{b 2}u^{u}_{4,x}\right]^*,{\nonumber}\\ g^{q_i}_x&=&u^q_{i , x}u^{q*}_{i+3,x},\end{aligned}\ ] ] where @xmath315 and @xmath316 .
@xmath317 and @xmath318 matrices diagonalize the neutralino and chargino mass matrices respectively .
the index @xmath319 of @xmath317 in the neutralino contribution formula takes value of @xmath320 for @xmath321 .
the chromoelectric dipole moments @xmath322 for quarks are defined as @xmath323 the contributions to @xmath322 from neutralino , chargino and gluino are given by @xmath324 where @xmath325 we use the qcd sum rule based estimate of ref .
@xcite to evaluate the neutron and the deuteron edms : @xmath326 here the running factors are @xmath327 and @xmath328 .
froggatt and h.b .
nielsen , nucl .
phys . * b147 * , 277 ( 1979 ) .
green and j.h .
schwarz , phys .
b149 * , 117 ( 1984 ) ; nucl . phys . *
b255 * , 93 ( 1985 ) ; m.b .
green , j.h .
schwarz and p. west , nucl .
b254 * , 327 ( 1985 ) .
l. e. ibanez , g. g. ross , phys . lett .
* b332 * , 100 ( 1994 ) ; p. binetruy and p. ramond , phys . lett .
* b350 * , 49 ( 1995 ) ; p. binetruy , s. lavignac and p. ramond , nucl . phys .
* b477 * , 353 ( 1996 ) .
babu and r.n .
mohapatra , phys .
lett . * 74 * , 2418 ( 1995 ) ; v. jain and r. shrock , phys .
b * 352 * , 83 ( 1995 ) ; y. nir , phys . lett .
* b354 * , 107 ( 1995 ) ; e. dudas , s. pokorski and c.a .
savoy , phys .
b356 * , 45 ( 1995 ) ; e. j. chun , a. lukas , phys . lett . *
b387 * , 99 ( 1996 ) ; y. nir and r. rattazzi , phys . lett . *
b382 * , 363 ( 1996 ) ; a. h. chamseddine and h. k. dreiner , phys . lett . * b389 * , 533 ( 1996 ) ; k. choi , e. j. chun and h. d. kim , phys . lett .
* b394 * , 89 ( 1997 ) ; z. berezhiani and z. tavartkiladze , phys . lett .
* b396 * , 150 ( 1997 ) , phys .
b409 * , 220 ( 1997 ) ; p. binetruy , s. lavignac , s.t . petcov and p. ramond ,
nucl . phys . *
b496 * , 3 ( 1997 ) ; p. binetruy , e. dudas , s. lavignac and c.a .
savoy , phys .
b422 * , 171 ( 1998 ) ; j. k. elwood , n. irges and p. ramond , phys .
* 81 * , 5064 ( 1998 ) ; y grossman , y. nir and y. shadmi , jhep * 9810 * , 007 ( 1998 ) ; g. altarelli and f. feruglio , jhep * 9811 * , 021 ( 1998 ) , phys . rept . * 320 * , 295 ( 1999 ) ; q. shafi and z. tavartkiladze , phys
* b451 * , 129 ( 1999 ) , phys . lett . *
b482 * , 145 ( 2000 ) ; g. eyal , phys . lett . *
b461 * , 71 ( 1999 ) ; m. e. gomez , g. k. leontaris , s. lola and j. d. vergados , phys . rev . *
d59 * , 116009 ( 1999 ) ; k. choi , k. hwang and e. j. chun , phys . rev .
* d60 * , 031301 ( 1999 ) ; j. feng and y. nir , phys . rev . *
d61 * , 113005 ( 2000 ) ; a.s .
joshipura , r. vaidya and s.k .
vempati , phys . rev . *
d62 * , 093020 ( 2000 ) ; m.s .
berger , k. siyeon , phys . rev . *
d62 * , 033004 ( 2000 ) , phys . rev . *
d64 * , 053006 ( 2001 ) ; n. maekawa , prog .
106 * , 401 ( 2001 ) ; j. sato and k. tobe , phys . rev . * d63 * , 116010 ( 2001 ) ; m. tanimoto , phys . lett . * b501 * , 231 ( 2001 ) ; i. gogoladze and a. perez - lorenzana , phys . rev . *
d65 * , 095011 ( 2002 ) ; i. jack , d.r.t .
jones and r. wild , phys .
* b535 * , 193 ( 2002 ) ; t. ohlsson and g. seidl , nucl . phys . *
b643 * , 247 ( 2002 ) ; s.m .
barr and i. dorsner , phys . rev .
* d65 * , 095004 ( 2002 ) . t. kobayashi , h. nakano , h. terao and k. yoshioka , prog .
. phys . * 110 * , 247 ( 2003 ) ; k.s .
babu , i. gogoladze and k. wang , nucl .
* b660 * , 322 ( 2003 ) ; h.k .
dreiner , m. thormeier , arxiv : hep - ph/0305270 ; h.k .
dreiner , h. murayama and m. thormeier , arxiv : hep - ph/0312012 .
babu , ts .
enkhbat and i. gogoladze , nucl . phys . *
b678 * , 233 ( 2004 ) . for a recent review
see n. fortson , p. sandars and s. barr , phys .
today * 56n6 * , 33 ( 2003 ) .
ellis and d.v .
nanopoulos , phys .
b110 * , 44 ( 1982 ) ; j.r .
ellis , s. ferrara and d.v .
nanopoulos , phys .
b114 * , 231 ( 1982 ) ; w. buchmuller and d. wyler , phys . lett . *
b121 * , 321 ( 1983 ) ; j. polchinski , phys . lett . *
b125 * , 393 ( 1983 ) .
f. del aguila , m.b .
gavela , j.a .
grifols and a. mendez , phys .
* b126 * , 71 ( 1983 ) , erratum - ibid .
* b129 * 473 ( 1983 ) ; e. franco and m. l. mangano , phys .
b135 * , 445 ( 1984 ) ; p. nath , phys .
* 66 * , 2565 ( 1991 ) ; y. kizukuri and n. oshimo , phys . rev . *
d45 * , 1806 ( 1992 ) ; r. garisto and g. l. kane , arxiv : hep - ph/9302282 ; c. hamzaoui and m.e .
pospelov , phys . lett .
* b357 * , 616 ( 1995 ) ; t. inui , y. mimura , n. sakai and t. sasaki , nucl . phys . * b449 * , 49 ( 1995 ) ; t. falk and k.a .
olive , phys .
* 375 * , 196 ( 1996 ) ; t. falk , k.a .
olive and m. pospelov , nucl .
phys . * b560 * , 3 ( 1999 ) ; d. chang , w. y. keung and a. pilaftsis , phys . rev .
lett . * 82 * , 900 ( 1999 ) [ erratum - ibid .
* 83 * , 3972 ( 1999 ) ] ; d. a. demir , o. lebedev , k. a. olive , m. pospelov and a. ritz , nucl .
b * 680 * , 339 ( 2004 ) .
t. ibrahim and p. nath , phys .
d57 * , 478 ( 1997 ) ; erratum-_ibid .
_ * d58 * , 019901 ( 1998 ) ; erratum-_ibid . _ * d60 * 079903 ( 1999 ) ; erratum-_ibid . _ * d60 * 119901 ( 1999 ) ; phys . rev . *
d58 * , 111301 ( 1998 ) ; erratum-_ibid . _ * d60 * , 099902 ( 1999 ) ; phys . lett . * 418 * , 98 ( 1998 ) ; phys . rev . *
d61 * , 093004 ( 2000 ) .
r. arnowitt , j.l .
lopez and d.v .
nanopoulos , phys . rev . *
d42 * , 2423 ( 1990 ) .
barger , t. falk , tao han , j. jiang , t. li and t. plehn , phys .
rev . * d64 * , 056007 ( 2001 ) ; a. pilaftsis , nucl . phys . * b644 * , 263 ( 2002 ) ; a. bartl , w. majerotto , w. porod and d. wyler , phys . rev . *
d68 * , 053005 ( 2003 ) .
babu , b. dutta and r.n .
mohapatra , phys .
lett . * 85 * 5064 ( 2000 ) ; e. accomando , r. arnowitt and b. dutta , phys . rev . *
d61 * , 115003 ( 2000 ) ; k.s .
babu , s.m .
barr and i. dorsner , phys . rev . * d64 * , 053009 ( 2001 ) ; s. abel , s. khalil and o. lebedev , nucl . phys .
* b606 * , 151 ( 2001 ) ; y. farzan , phys .
d * 69 * , 073009 ( 2004 ) ; t. appelquist , m. piai and r. shrock , arxiv : hep - ph/0401114 . j. ellis , j. hisano , m. raidal and y. shimizu , phys . lett . *
b528 * , 86 ( 2002 ) .
regan , e.d .
commins , c.j .
schmidt and d. demille , phys .
* 88 * , 071805 ( 2002 ) .
d. kawall , f. bay , s. bickman , y. jiang and d. demille , phys .
lett . * 92 * , 133007 ( 2004 ) .
lamoreaux , talk at lepton moments " , cape cod , 912 june 2003 .
semertzidis et al , int . j. mod
* a16s1b * , 690 ( 2001 ) .
harris et al . , phys .
lett . * 82 * , 904 ( 1999 ) .
see for example , t. soldner , arxiv : hep - ex/0405062 . y. k. semertzidis _ et al . _
[ edm collaboration ] , aip conf .
proc . * 698 * , 200 ( 2004 ) [ arxiv : hep - ex/0308063 ] .
i. masina , nucl .
b * 671 * , 432 ( 2003 ) .
y. farzan and m.e .
peskin , arxiv : hep - ph/0405214 .
f. borzumati and a. masiero , phys .
lett . * 57 * , 961 ( 1986 ) ; l. j. hall , v. a. kostelecky and s. raby , nucl .
b * 267 * , 415 ( 1986 ) ; s. dimopoulos and l. j. hall , phys .
b * 344 * , 185 ( 1995 ) .
m. dine , n. seiberg and e. witten , nucl .
b289 * , 589 ( 1987 ) ; j. atick , l. dixon and a. sen , nucl . phys . *
b292 * , 109 ( 1987 ) .
k. s. babu and s. m. barr , phys .
b * 381 * , 202 ( 1996 ) ; c. h. albright , k. s. babu and s. m. barr , phys .
lett . * 81 * , 1167 ( 1998 ) ; j. sato and t. yanagida , phys .
* b430 * , 127 ( 1998 ) ; n. irges , s. lavignac and p. ramond , phys .
d * 58 * , 035003 ( 1998 ) ; p. ginsparg , phys .
b197 * , 139 ( 1987 ) ; v. s. kaplunovsky , nucl . phys . *
b307 * , 145 ( 1988 ) , erratum-_ibid .
_ * b382 * , 436 ( 1992 ) ; h. k. dreiner , h. murayama and m. thormeier , arxiv : hep - ph/0312012 ; m. cvetic , l. l. everett and j. wang , phys . rev .
* d59 * , 107901 ( 1999 ) .
e. witten , nucl .
b471 * , 135 ( 1996 ) .
see for example , j. ellis , t. falk , k. olive and m. srednicki , astropart .
* 13 * , 181 ( 2000 ) ; r. arnowitt , b. dutta and y. santoso , nucl . phys . * b606 * , 59 ( 2001 ) .
see for example , h. baer , c. balazs , a. belyaev and j. ofarrill , jcap * 0309 * , 007 ( 2003 ) .
romalis , w.c . griffith and e.n .
fortson , phys .
lett . * 86 * , 2505 ( 2001 ) .
j. hisano and y. shimizu , phys . lett .
* b581 * , 224 ( 2004 ) ; m. endo , m. kakizaki and m. yamaguchi , arxiv : hep - ph/0311072 .
see for example j. gasser and h. leutwyler , phys .
rept . * 87 * , 77 ( 1982 ) .
martin , phys . rev . *
d61 * , 035004 ( 2000 ) .
s. weinberg , phys .
lett . * 63 * , 2333 ( 1989 ) ; e. braaten , c.s .
li and t.c .
yuan , _ ibid .
_ * 64 * , 1709 ( 1990 ) ; j. dai , h. dykstra , r.g .
leigh , s. paban and d.a .
dicus , phys . lett . *
b237 * , 216 ( 1991 ) .
o. lebedev , k. a. olive , m. pospelov and a. ritz , arxiv : hep - ph/0402023 . | we show that in supersymmetric models with a gauged flavor symmetry explaining the hierarchy of fermion masses and mixings , the electron , muon , neutron and the deuteron acquire sizable electric dipole moments ( edm ) through loop diagrams involving the flavor gaugino / gauge boson near the planck scale .
these edms are proportional to the phases in the fermion yukawa couplings and are typically much larger than the neutrino seesaw induced edm for the leptons . in a popular class of models based on anomalous @xmath0 flavor symmetry of string origin , we find @xmath1 cm , @xmath2 cm , @xmath3 cm , and @xmath4 cm , which are within reach of next generation experiments .
osu - hep-04 - 5 + may 2004 + |
nitio@xmath4 ( nto ) and cotio@xmath4 ( cto ) are both antiferromagnetic below nel temperatures 23 k@xcite and 38 k@xcite , respectively .
they crystallize in ilmenite - type structure with the space group @xmath5 .
planes formed of double sheets of triangular network of ni ( or co ) and ti cations are altered along the hexagonal @xmath6-axis and each cation is coordinated by six oxygen atoms which form a distorted octahedron .
the magnetic moments of nto and cto order in the same way .
ni@xmath7 ( co@xmath7 ) moments , parallel to the planes ( the easy axis being perpendicular to the @xmath6-axis ) , are ferromagnetically coupled within each plane while the inter - plane coupling is antiferromagnetic @xcite .
when ilmenite oxides form a solid - solution , complex and interesting magnetic properties emerge .
an example is ( fe@xmath8co@xmath9)tio@xmath2 ( fct ) , where the spin arrangement of fetio@xmath4 ( fto ) is the same as that of nto except for the spin direction@xcite .
fct is known to be a mixed ising-_xy _
antiferromagnet @xcite .
the phase diagram of the concentration ( @xmath10 ) versus temperature ( @xmath11 ) given in ref .
@xcite shows that there are three magnetically ordered phases with respect to the composition : two antiferromagnetic phases with different spin directions for co- and fe - rich areas and a mixed ordering , or oblique antiferromagnetic , phase in the vicinity of the fe@xmath12co@xmath12tio@xmath4 composition .
the motivation for the present study is to look a solid - solution system closely related to fct but in which the two constituents have identical magnetic ordering .
the study is dedicated on polycrystalline ( ni@xmath0co@xmath1)tio@xmath2 ( nct ) solid solutions .
our experiments reveal that , despite identical spin structures of the constituents , magnetic properties of nct behave in a complicated manner .
polycrystalline stoichiometric ( ni@xmath0co@xmath1)tio@xmath2 ( @xmath13 ) were prepared through the conventional solid - state technique .
samples are referred to as nc@xmath14/@xmath15 indicating @xmath14% of ni and @xmath15% of co. nio ( 99 % , aldrich ) , coo ( @xmath16 99.99 % , aldrich ) and tio@xmath17 ( 99 - 100.5 % , riedel - de han ) powders were mixed in the desired molar ratio , then pressed into a pellet and sintered in air at 1373 k. phase quality was inspected by x - ray diffraction using cu k@xmath18 radiation . composition analysis were carried out by energy dispersive spectroscopy of x - rays ( eds ) employing jeol jsm-7500fa analytical field - emission scanning electron microscope ( sem ) .
magnetization measurements were performed by a superconducting quantum interference device ( squid ) magnetometer ( quantum design xl7 ) .
dc magnetization in zero - field - cooled ( zfc ) and field - cooled ( fc ) runs were carried out over the temperature range down to 5 k under various magnetic fields between 1 and 1000 oe .
field - dependent magnetization data as a function of applied field were collected at various temperatures .
ac magnetization measurements were performed with the heating rate of 0.8 k / min with 1 oe ac field amplitude .
prior to measurements , sample was carefully centered with respect to pick - up coils and the centering was regularly monitored .
residual magnetic field was estimated using a paramagnetic material prior to and after each measurement so that the real applied field is given .
because of the similarity in the crystal structure and chemical character between nto and cto , one can expect solid solution with no miscibility gap for whole composition range .
in fact , all solid - solutions in the tested composition range were obtained as single phase .
figure [ xrd ] shows x - ray diffraction patterns of selected nct powders . the nct sample with lowest co content was light green in color and became dark with increasing co content .
@xmath19 ( @xmath20 is odd ) reflections , such as 003 and 101 , indicative to the ilmenite phase @xmath5 .
disappearance of these reflections implies a higher symmetry @xmath21 . for nto and cto , however , the @xmath19 peaks are usually rather weak and the degree of ordering could not be determined from the xrd data alone .
the room - temperature x - ray diffraction also show that lattice parameters were increasing with increasing co - content , suggesting that co@xmath7 and ni@xmath7 were in a high - spin state ( their ionic radii are @xmath22 and @xmath23 , respectively@xcite , in contrast , the low - spin ionic radii of co@xmath7 is @xmath24 @xcite )
. however , as the antiferromagnetic ordering temperatures of cotio@xmath4 and nitio@xmath4 demonstrate , the consideration of bond - lengths alone is insufficient for understanding the composition dependent magnetic transition temperatures .
homogeneity of the compounds were also inspected by sem - eds .
no spatial composition variation was evidenced .
compositions determined by sem - eds , listed in table [ table1 ] show that the starting molar ratio was preserved well .
[ [ overview - of - dc - magnetization - results . ] ] overview of dc magnetization results .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + figures [ mtl ] and [ mth ] show the temperature dependence of dc magnetization [ @xmath25 at various applied fields .
overall the magnetic properties of nct samples were quite similar : there are two magnetic transitions at around 23 and 67 k. the higher temperature transition is only characteristic to the solid - solutions . above this temperature , ncts were paramagnetic : the inverse susceptibility was linearly decreasing with decreasing temperature down to the transition temperature . in the paramagnetic phase , susceptibility obeys the curie - weiss law .
correspondingly , effective magnetic moments @xmath26 were estimated from the paramagnetic region data , as shown in table [ table1 ] . for comparison
, the magnetic moment values were estimated from the equation based on the assumption of quenched orbital moments : @xmath27 , where @xmath28 and @xmath29 are total electron spin moment values of ni and co ions in their high - spin states , respectively .
the difference between the magnetic moments estimated from the susceptibility data and the quenched orbital moment values became larger with increasing co content , suggesting that the orbital moments of co - ions were not entirely quenched .
interestingly , the magnetic moment values of nitio@xmath4 @xcite and notably of cotio@xmath4 @xcite were larger than the values expected for quenched orbital moments . for comparison , also values based on the hund s rules
are given in table [ table1 ] .
also the co@xmath7 ions in lamn@xmath12co@xmath12o@xmath4 were found to possess a large orbital moment , which was predicted to result in a nontrivial temperature dependence of the magnetic susceptibility @xcite .
the low temperature transition was close to the @xmath30 of nto .
with increasing field , the cusp at around 27 k became sharper .
the cusp is characteristic signal to a transition to an antiferromagnetic phase .
the peak temperature of the cusp is referred to as @xmath31 .
antiferromagnetic order is consistent with the negative values of asymptotic curie point @xmath32/k and the results of field dependent magnetization [ @xmath33 , as given in figs .
[ mhl ] and [ mhh ] .
magnetization was almost linearly dependent of the applied field , except for the small hysteresis loop which diminished significantly at 40 k and vanished above the high temperature transition .
the opening was largest at around @xmath34 composition , and decreased for small and large values of @xmath10 . with increasing co content up to 0.8
, @xmath31 almost linearly increased to 32 k. the linear increase of @xmath31 with increasing @xmath10 parallels with the linear increase of magnetic moment @xmath26 , see table [ table1 ] . despite the fact that the solid solution has larger number of magnetic interactions ( ni - ni , ni - co and co - co interactions in simplest terms ) than the parent compounds , it seems that the exchange integrals @xmath35 of different magnetic atom pairs are roughly the same , in contrast to ( fe@xmath8co@xmath9)tio@xmath2 or other spin glass compounds . for a typical spin glass material ,
glass transition temperature declines when the composition approaches the intermediate region@xcite .
this behavior is explained by random distribution of @xmath35 over magnetic ion pairs .
we estimated the temperature at which the inverse dc susceptibility versus temperature curve starts to depart from the extrapolation of curie - weiss relation , the onset of a transition .
hereafter this temperature is referred to as irreversible point @xmath36 .
@xmath36 exhibited composition dependence : increasing co content pushed the transition temperature down .
the decrease was not linear as a function of co content but was faster at highest co content so that @xmath36 decreased to 57 k for nc20/80 sample .
@xmath36 was sensitive to an applied field : with increasing applied field @xmath36 was shifted to lower temperature .
[ [ dc - and - ac - magnetization - measurements - of - nc5050-and - nc6040 . ] ] dc and ac magnetization measurements of nc50/50 and nc60/40 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + to look closer to the nature of the transitions , frequency dependence of ac magnetization with various applied fields were measured for nc50/50 and nc60/40 .
in both cases , real and imaginary ac magnetization showed sharp peaks at the point corresponding to the high temperature transitions , see figs .
[ nc55 ] and [ nc64 ] .
@xmath31 also corresponds to the low temperature cusp of real part of ac magnetization curves .
next we focus on the result of nc50/50 , although the discussion is valid to all nct solid - solutions .
with increasing field , the difference between zfc and fc curves [ panels ( a ) , ( b ) , and ( c ) in fig . [ nc55 ] ] became smaller and the higher temperature transition slightly shifted to lower temperature , as was commonly observed in all other ncts .
the higher temperature transition at which both the real and imaginary parts of ac magnetization curves show a peak is labeled as @xmath37 .
we note that @xmath37 is slightly lower than @xmath36 , though they indicate the same magnetic transition . except for the peak at @xmath37 ,
the imaginary part of magnetization was close to zero .
a sharp peak in imaginary part generally indicates that magnetic domain structure with switchable magnetization is formed . on the contrary ,
clear yet small hysteresis loop was observed only at low temperature : the net magnetization was largest at lowest temperature .
thus , the dissipation peak at @xmath37 and the ferromagnetic hysteresis seem to be unrelated phenomena suggesting the different nature of the two phases . the small hysteresis loop can be explained by canting the antiferromagnetically ordered spins so that there is a small net magnetization .
both dc and ac magnetization results of the low - temperature phase are consistent with the canted antiferromagnetic order .
the difference in magnetic ordering was also reflected by the different shape of the magnetization versus applied field : whereas the lowest temperature curve had an s - shape , above @xmath38 the curve was pratically a straight line .
the dissipation peak at @xmath37 suggests a formation of domains which are sufficiently large to show no frequency dependent response .
consistently with the dc magnetization results , the @xmath37 , determined through the ac - measurements , was slightly shifted with increasing applied field .
this suggests that the origin of the phenomena is on the collective spin ordering on the layers perpendicular to the hexagonal @xmath6-axis rather than magnetic nanodomains .
specifically the high @xmath39 and the sensitivity to an applied field further implies that the magnetic phase transitions can be related to the in - plane spin order .
the magnetization of the nc50/50 samples in zfc curve measured with weak field was negative up to @xmath37 , see panel ( a ) in fig .
negative magnetization was also observed in the nc80/20 and nc60/40 .
the temperature at which the sign of magnetization changes varied by the heating rate . to verify the time dependence of the magnetization , data were collected keeping the temperature constant at 12 and 40 k for 100 min in zfc run with weak external field , as shown in fig .
magnetization curve in fig .
[ td](a ) crosses zero at much lower temperature comparing to fig . [
nc55](a ) , still the peak temperatures matched to @xmath38 and @xmath37 . with increasing time ,
magnetization at 12 and 40 k gradually increased though the change at 40 k was larger than 12 k. similar time - dependent magnetization can be seen in a spin glass .
however , ac magnetization data ( the right column of fig . [ nc55 ] ) contradict the characteristic feature of spin glass@xcite which displays a clear frequency dependence when a small field is applied : neither real or imaginary part of ac magnetization curves exhibited frequency dependence .
the construction of a detailed picture of the magnetic and possible cation ordering requires neutron and synchrotron diffraction studies .
inset shows the 2@xmath40 region where the characteristic peaks of the @xmath5 phase are seen.,width=302 ] .compositions , nel point @xmath31/k , irreversible point @xmath36/k , effective moment @xmath26 in bohr magneton unit @xmath41 and asymptotic curie point @xmath32/k estimated from curie - weiss law and effective moment @xmath42 in bohr magneton unit assuming orbital quenching of ( ni@xmath0co@xmath1)tio@xmath2 ( @xmath43 ) .
also moments @xmath44 , where @xmath45 is the total orbital angular moment and @xmath46 is the land g - value ( based on hund s rules expectations ) are given .
compositions indicate cation ratio ( oxygen content was not determined ) . [ cols= " < , < , <
, < , < , < , < , < " , ] .
( a)nc95/5 , ( b)nc90/10 , ( c)nc80/20 , ( d)nc70/30.,width=311 ] .
( e)nc60/40 , ( f)nc50/50 , ( g)nc40/60 , ( h)nc20/80.,width=311 ]
in conclusion , solid - solution of nitio@xmath4 and cotio@xmath4 , ( ni@xmath0co@xmath47)tio@xmath4 ( @xmath48 ) , were successfully prepared through the solid - state reaction at ambient atmosphere .
dc and ac magnetization revealed that below the room temperature nct solid solution has two magnetic transitions : to an antiferromagnetic phase , which took place @xmath38 = 27 k and another transition which occurred around @xmath39 = 63 k , being significantly higher than @xmath38 .
small net magnetization was observed at the lowest temperatures , which was probably due to the canted spins .
negative magnetization was observed at weak fields for samples with @xmath49 0.2 , 0.4 and 0.5 . unlike in most @xmath50 transition metal oxides ,
the orbital moments of ni and co in nct were not entirely quenched .
though the time dependence of magnetization was observed , there was no evidence for a spin glass transition . in summary
, we have shown that peculiar magnetic properties emerge in nct system .
we thank a. savin for his help in operating the magnetometer .
10 y. ishikawa , s. akimoto , j. phys .
japan 13 ( 1958 ) 1298 - 1310 .
newnham , j.h .
fang , r.p .
santoro , acta crystallogr .
17 ( 1964 ) 240 - 242 .
g. shirane , s.j .
pickart , y. ishikawa , j. phys .
japan 14 ( 1959 ) 1352 - 1360 .
a. ito , s. morimoto , y. someya , y. syono , and h. takei , j. phys .
japan 51 ( 1982 ) 3173 - 3182 .
a. ito , c. uchikoshi , a. fukaya , s. morimoto , k. nakajima , k. kakurai , j. phys .
solids 60 ( 1999 ) 1233 - 1235 .
landau and k. binder : _ a guide to monte carlo simulations in statistical physics _ cambridge university press , cambridge , 2000 .
g. s. rohrer : _ structure and bonding in crystalline materials _ cambridge university press , cambridge , 2001 .
t. burnus , z. hu , h. h. hsieh , v. l. j. joly , p. a. joy , m. w. haverkort , h. wu , a. tanaka , h. -j .
lin , c. t. chen and l. h. tjeng .
b 77 ( 2008 ) 125124 . | ilmenite solid - solutions , ( ni@xmath0co@xmath1)tio@xmath2 ( @xmath3 ) , were synthesized at ambient atmosphere through solid - state reaction and were studied by energy dispersive spectroscopy of x - rays , x - ray diffraction , and dc and ac magnetometry .
temperature dependent dc magnetic measurements revealed two transitions .
the first took place at around 27 k and the second at 63 k. the low - temperature phase was antiferromagnetic .
the phase observed between 27 and 63 k is characteristic to the solid - solution and is not found in either of the constituent members .
a fit of the data to the curie - weiss law gave magnetic moment values which were significantly larger than the values based on the quenched orbital momentum assumption .
zero - field - cooled magnetization measurements with weak magnetic field revealed unexpected negative magnetization at low temperatures .
below 63 k the dc magnetization exhibited time dependent behavior .
frequency and magnetic field dependent ac magnetization is also addressed . |
the recently discovered resonance with mass @xmath3 gev @xcite strongly resembles the standard model ( sm ) higgs boson .
there is a hint that the loop - induced @xmath0 event rate @xcite deviates , modulo qcd uncertainties @xcite , by a factor 1.5 - 2 @xcite from its sm value , @xmath4_{\rm lhc}\over \left[\sigma(gg\to h)\times { \rm br } ( h\to \gamma \gamma)\right]_{\rm sm } } = 1.71 \pm 0.33 \ .\end{aligned}\ ] ] this indicates the existence of additional charged particle(s ) on which the tree - level higgs decay modes are much less sensitive .
the history of the `` prediscovery '' of charmed and top quarks through the loop amplitudes may be repeated in the scalar sector .
on the other hand , if this enhancement in @xmath5 disappears with a larger integrated luminosity , it will still constrain the parameter space of various extensions of the sm containing new charged states which should affect this loop amplitude . in this spirit
there is a number of theoretical attempts to match the indicated discrepancy by the effects of an extended higgs sector @xcite .
notable , the @xmath0 decay rate is sensitive on the doubly charged scalar fields such as those existing in higgs triplet model of neutrino mass generation .
these states have been in focus of both the direct searches at the lhc @xcite and of theoretical investigations @xcite .
here we go a step further by studying the effects of singly and doubly charged components of a scalar quadruplet contained in our recent tev - scale seesaw model for neutrino masses @xcite .
the letter is organized as follows . in section 2
we briefly review the tev - scale seesaw model at hand .
a detailed study of the scalar potential is presented in section 3 .
we investigate the effects of exotic scalars on the loop - level higgs decays in sections 4 and 5 . possible direct bounds on exotic scalars are discussed in section 6 .
the conclusion is presented in section 7 .
a simple and predictive tev - scale seesaw model @xcite under consideration belongs to a class of beyond dimension - five seesaw models @xcite which , on account of introducing higher iso - multiplets , lower the originally high seesaw scale to the tev scale accessible at the lhc . in order to produce the seesaw diagram in the simple model at hand ,
the fermionic hypercharge - zero quintuplets @xmath6 transforming as @xmath7 under the sm gauge group have to be accompanied by a scalar quadruplet @xmath8 transforming as @xmath9 .
the gauge invariant and renormalizable lagrangian involving these new fields reads @xmath10 where the scalar potential contains the mass and the quartic terms for the doublet @xmath11 and the quadruplet @xmath12 fields @xmath13 in the tensor notation the terms in eqs .
( [ lagrangian ] ) and ( [ scalarpot ] ) read @xcite @xmath14 the light neutrino masses arise through tree - level dimension - nine operator and one - loop dimension - five operator .
the diagrams displayed in fig . [ diagrams ] lead to two contributions to light neutrino masses @xmath15 \ , \nonumber\\\end{aligned}\ ] ] determined by quartic couplings @xmath16 and @xmath17 , respectively .
let us remind @xcite that in the regime of comparable and light ( @xmath18 200 gev ) masses for exotic fermions and scalars the tree contribution prevails , while the loop contribution dominates for heavy masses ( @xmath18 500 gev ) and remains a sole contribution if additional discrete @xmath19 symmetry forbids the @xmath16 term in eq .
( [ scalarpot ] ) .
after the electroweak symmetry breaking ( ewsb ) the neutral components of the scalar fields acquire a vacuum expectation value and read @xmath20 if all couplings in the scalar potential are real the conditions for the minimum of the potential @xmath21 give @xmath22 the electroweak @xmath23 parameter is changed by @xmath24 from the unit value to @xmath25 constraining the ratio @xmath26 to be smaller then 0.015 .
the mixing between singly charged and between neutral components of @xmath11 and @xmath12 multiplets will occur after the ewsb .
let us illustrate this on the @xmath27 mass terms @xmath28 the mass term that mixes @xmath29 and @xmath30 is small because it is proportional either to @xmath24 from @xmath31 terms in eq .
( [ scalarpot ] ) , or to small lepton number violating couplings @xmath32 dictated to be small from considerations of neutrino masses .
therefore we neglect the mixing between the components of @xmath11 and @xmath12 multiplets and we neglect the terms of higher order in @xmath26 whenever possible . in the approximations given above , the masses of charged components of the quadruplet @xmath12 are @xmath33 their couplings to the higgs boson relevant for the @xmath0 decay are @xmath34 where the newly introduced couplings @xmath35 are expressed in terms of the quartic couplings @xmath36 and @xmath37 which , for simplicity , we assume to be equal in the following .
in our model in addition to the dominant sm contributions from the @xmath38 boson and top quark loops to the @xmath0 decay rate only the charged scalars contribute substantially .
our exotic scalars are colorless so that the higgs boson production through gluon fusion at the lhc is unaffected . also , since the higgs boson total decay width is only marginally affected , a dominant change in @xmath0 event rate comes from the change in higgs boson diphoton partial decay width .
the analytic expression for the diphoton @xmath0 partial width reads @xcite @xmath39 where the three contributions corresponding to @xmath40 ( @xmath41 ) refer to spin-1 ( @xmath38 boson ) , spin-1/2 ( top quark ) and charged spin-0 particles in the loop .
the electric charges for fermions and scalars , @xmath42 and @xmath43 , are given in units of @xmath44 , and their respective number of colors are @xmath45 and @xmath46 for color singlet scalars under consideration .
the loop functions for spin-1 , spin-1/2 , and spin-0 particles are given as @xmath47\ , \\
\label{t - loop } a_{1/2}(x ) & = & 2 \
, x^2 \left[x^{-1}+ ( x^{-1}-1)f(x^{-1})\right ] \ , \\ \label{s - loop } a_0(x ) & = & -x^2 \left[x^{-1}-f(x^{-1})\right ] \ , \end{aligned}\ ] ] where , for a higgs mass below the kinematic threshold of the loop particle @xmath48 , we have @xmath49 = ( 2.5 , 2 , 1.75 , 1.5 , 1.25 ) for the @xmath0 branching ratio in dependence on scalar coupling @xmath50 and the mass of the lightest charged component of the @xmath12 multiplet , @xmath51 . ]
we now look at the possible enhancement of the @xmath0 decay rate through extra contributions with charged components of scalar quadruplet @xmath12 running in the loops . following @xcite we define the enhancement factor with respect to the sm decay width @xmath52 in order to get an enhancement of the @xmath0 decay rate , the contribution of the new charged scalars has to interfere constructively with the dominant sm contribution of the @xmath38 boson . since this requires negative couplings @xmath53 in eq .
( [ w - t - s - loop ] ) , we assume @xmath54 .
the biggest effect on the @xmath0 decay rate comes from @xmath1 for two reasons .
first , it has a charge @xmath55 and second , for @xmath56 it is the lightest of the charged @xmath12 components .
we will not consider @xmath57 case for which to get a significant enhancement in @xmath58 the charged scalars should be too light . in fig .
[ rgamma ] we plot the enhancement factor @xmath58 as a function of the scalar couplings @xmath59 and the mass of the lightest charged scalar @xmath51 . for simplicity
we assume @xmath60 , and in order to keep the stability of the scalar potential , we restrict ourselves to the interval @xmath61 $ ] . in this interval the value @xmath62 can be achieved up to @xmath63 gev and the value @xmath64 up to @xmath65 gev .
another loop - mediated higgs decay sensitive to new charged particles is @xmath2 .
it is sensitive both on the charge and weak isospin of the particles in the loop . after adding a contribution of a new scalar to the sm contributions from the @xmath38 boson and top quark the decay rate
is given by @xcite @xmath66 where @xmath67 @xmath68 here @xmath69 and @xmath70 and @xmath71 correspond to the weak isospin of the top quark and the new scalar .
the loop functions are given by @xmath72 i_1(x , y ) \
, \nonumber\\ a_{1/2}(x , y ) & = & i_1(x , y)-i_2(x , y ) \ , \nonumber\\ a_{0}(x , y ) & = & i_1(x , y ) \ , \end{aligned}\ ] ] where @xmath73 + \frac{x^2 y}{(x - y)^2}[g(x^{-1})-g(y^{-1 } ) ] \ , \nonumber\\ i_2(x , y ) & = & - \frac{x y}{2(x - y ) } [ f(x^{-1})-f(y^{-1 } ) ] \ .\end{aligned}\ ] ] for the higgs mass below the kinematic threshold of the loop particle , @xmath48
, we have @xmath74 the additional contributions in a model with scalar quadruplet lead to the modification factor @xmath75 for the @xmath2 decay rate with respect to the sm decay width @xmath76 in fig .
[ rzg ] we plot the modification factor @xmath75 as a function of the scalar couplings @xmath59 ( assuming @xmath60 ) and the mass of the lightest charged scalar @xmath51 .
we obtain a moderate suppression of the @xmath2 decay rate in the region of the parameter space where the @xmath0 decay rate is enhanced . in the chosen interval @xmath61 $ ] the factor @xmath77
is achieved up to @xmath78 gev and @xmath79 up to @xmath80 gev .
= ( 0.6 , 0.7 , 0.8 , 0.9 ) for the @xmath2 branching ratio in dependence on the scalar coupling @xmath50 and the mass of the lightest charged component of the @xmath12 multiplet , @xmath51 . ]
the measurement of the @xmath0 rare decay mode may mark a spectacular onset of a virtual physics at the lhc .
the interaction with virtual particles is essential both for the production and the decay factor in eq .
( [ expt ] ) .
it is in order to discuss the features of additional charged scalars , which determine their possible appearance both in the higgs boson loop decays under consideration and in their direct searches .
now , an essential point is a lack of yukawa couplings of the scalar quadruplet to a pair of sm fermions .
the lack of these yukawa couplings means that a doubly charged @xmath1 does not decay to a pair of charged sm leptons , which has decisive repercussions .
the direct searches for doubly charged scalars through resonance in the invariant mass of a pair of charged leptons at the lhc @xcite leave the quadruplet scalar states unconstrained .
the strongest bound in the range ( 383 gev , 408 gev ) set in the inclusive search for the cms benchmark points depends on @xmath1 production analysis , and counts all possible final state lepton pairs in br(@xmath81 ) , @xmath82 .
of course , @xmath1 can decay also to @xmath83 final state , but this channel has not been measured yet and does nt constrain the quadruplet scalar mass . in comparison ,
the studies of @xmath0 enhancement by the doubly charged scalar from the higgs triplet model of neutrino mass generation @xcite also lead to their rather small masses , which already seem to be excluded by the above mentioned direct searches at the lhc @xcite . in case that the decay of the doubly charged component of the triplet scalar to @xmath83 were the dominant one , it would invalidate the discussed cms search and no limit on the triplet mass would be placed either .
it is conceivable that the first sign of new physics observed at the lhc may stem from loop effects of new particles additional to the sm content .
we take under scrutiny particular extensions of the sm higgs sector aimed to explain the small masses of neutrinos , which simultaneously provide doubly charged scalars that significantly affect the loop amplitude for the @xmath0 decay . in view of the upper mentioned exclusion of the doubly charged scalars from the higgs triplet model of neutrino masses ,
we find it timely to explore the effect of doubly charged component of the scalar quadruplet existing in our recent simple tev - scale seesaw model for neutrino masses @xcite .
the results which we obtain show that it is easier to obtain an enhancement of the @xmath0 decay rate in our model than in the higgs triplet model .
the higgs quadruplet is devoid of the yukawa couplings to a pair of leptons which play a decisive role in existing lhc exclusion bounds for doubly charged scalars .
therefore our model in the sub - tev mass region , where a significant enhancement in @xmath58 can be obtained , is not excluded .
also , there is a remarkable feature of achieving the biggest effect on @xmath0 decay rate from the doubly charged @xmath1 state at hand .
the essential @xmath56 coupling ensures that @xmath1 is the lightest among the charged @xmath12 components and that it interferes constructively with the dominant sm contribution of the @xmath38 boson . concerning the monitoring @xmath2 decay ,
the effect is smaller and opposite .
there is a moderate suppression of the @xmath2 decay rate in the region of the parameter space where the @xmath0 decay rate is enhanced .
finally , the imposed role on the scalar quadruplet in explaining the smallness of neutrino masses is decisive in keeping the relevant couplings from the scalar potential small so that the mixing between components of @xmath11 and @xmath12 multiplets is negligible . accordingly , the state announced at the lhc @xcite corresponds in our setup to the sm higgs boson .
g. aad _ et al . _ [ atlas collaboration ] , _ observation of a new particle in the search for the standard model higgs boson with the atlas detector at the lhc _ ,
b * 716 * , 1 ( 2012 ) [ 1207.7214 [ hep - ex ] ] .
s. chatrchyan _ et al . _
[ cms collaboration ] _
search for the standard model higgs boson decaying into two photons in pp collisions at sqrt(s)=7 tev _ , phys .
b * 710 * , 403 ( 2012 ) [ 1202.1487 [ hep - ex ] ] .
g. aad _ et al . _
[ atlas collaboration ] , _ search for the standard model higgs boson in the diphoton decay channel with 4.9 fb-1 of pp collisions at sqrt(s)=7 tev with atlas _ , phys .
lett . * 108 * , 111803 ( 2012 ) [ 1202.1414 [ hep - ex ] ] .
g. aad _ et al . _ [ atlas collaboration ] , _ search for anomalous production of prompt like - sign muon pairs and constraints on physics beyond the standard model with the atlas detector _
d * 85 * , 032004 ( 2012 ) [ 1201.1091 [ hep - ex ] ] ; g. aad
_ et al . _ [ atlas collaboration ] , _ search for doubly - charged higgs bosons in like - sign dilepton final states at sqrt(s ) = 7 tev with the atlas detector _ , [ 1210.5070 [ hep - ex ] ] .
s. kanemura and k. yagyu , _ radiative corrections to electroweak parameters in the higgs triplet model and implication with the recent higgs boson searches _ , phys .
d * 85 * , 115009 ( 2012 ) [ 1201.6287 [ hep - ph ] ] . | we examine the role of seesaw - motivated exotic scalars in loop - mediated higgs decays .
we consider a simple tev - scale seesaw model built upon the fermionic quintuplet mediator in conjunction with the scalar quadruplet , where we examine portions of the model parameter space for which the contributions of charged components of the scalar quadruplet significantly increase the @xmath0 decay rate .
the most significant change in the diphoton width comes from a doubly charged scalar @xmath1 which should be the lightest component in the scalar quadruplet . in the part of the parameter space where the @xmath0 decay width is enhanced by a factor 1.25
2 there is a mild suppression of the @xmath2 decay width by a factor 0.9 0.7 .
= 2 * * * * + = 0.5 ivica picek and branimir radovi + _ department of physics , faculty of science , university of zagreb , p.o.b .
331 , hr-10002 zagreb , croatia + _ + _ pacs _ : 14.80.fd ; 12.60.fr ; 14.60.pq + _ keywords _ : charged higgs bosons ; extensions of higgs sector ; neutrino mass |
a major open topic in stellar cluster research is whether clusters can host multiple star - formation events or extended star - formation epochs .
young massive clusters ( ymcs ) have stellar populations that are well reproduced as single burst events , with upper limits on potential age spreads of @xmath2 myr ( cf .
longmore et al .
globular clusters , on the other hand , host chemical abundance anomalies within their stellar populations ( e.g. , gratton et al . 2012 ) , and while age spreads can not be discerned at these old ages , some scenarios to explain the anomalies invoke multiple star - forming epochs , spanning tens to hundreds of myr ( e.g. , dercole et al .
2008 ; conroy & spergel 2011 ) , although other scenarios have little or no age spreads ( e.g. , krause et al .
2013 ; bastian et al .
2013b ) .
mackey & broby nielsen ( 2007 ) constructed a colour - magnitude diagram ( cmd ) for ngc 1846 , an intermediate age ( @xmath0 gyr ) massive ( @xmath3 ) cluster in the lmc , and found that the main sequence turn - off region was not well described by a simple stellar population ( ssp ) , but rather was extended .
the authors interpreted this spread as a spread in age of @xmath4 myr and postulated that ngc 1846 was the result of a rare cluster merger , where the two clusters had a @xmath4 myr age difference .
however , subsequent work has shown that the majority of intermediate age ( @xmath0 gyr ) clusters in the lmc and smc display this extended main - sequence turn - off ( emsto - e.g. , mackey et al .
2008 ; milone et al .
2009 ; goudfrooij et al . 2011a , b ; keller et al . 2012 ; girardi et al . 2013 ) , ruling out rare events like unequal age cluster mergers . if interpreted as being caused by age spreads , these results suggest , rather , that star - formation can proceed for @xmath5 myr within massive clusters , and that age spreads of this magnitude should be a common feature in massive clusters .
such age spreads have been searched for in younger massive clusters , where such spreads should be readily apparent .
larsen et al . ( 2011 ) studied the resolved cmds of six massive ( @xmath6 ) clusters in nearby galaxies , and did not find evidence for age spreads of more than @xmath7 myr ( and being consistent with an ssp in most cases ) .
bastian & silva - villa ( 2013 ) looked at the cmds of ngc 1866 and 1856 , two clusters with masses of @xmath3 , and ages of 180 and 280 myr , respectively .
no age spreads were found , with upper limits of any potential spreads of @xmath8 myr .
niederhofer et al .
( 2015 ) extended the bastian & silva - villa survey to an additional eight lmc young massive clusters , including ngc 1850 , a @xmath9 , 100 myr cluster , and also did not find any evidence for age spreads .
it is important to note that the ymcs studied so far have current masses ( and escape velocities ) well in excess of the intermediate age clusters that show emstos .
hence , unless extreme cluster mass loss is invoked , the ymcs have the same properties ( or even higher mass and escape velocities ) as the intermediate age clusters that display the emsto features did when they were at a similar age .
finally , bastian et al . ( 2013a ) examined the integrated spectra ( or resolved cmds ) of @xmath10 ymcs with ages between @xmath11 myr and masses of @xmath12 , looking for evidence of ongoing star - formation , and none were found .
these studies have called into question the interpretation of the emsto feature being due to age spreads . since age spreads are not seen in ymcs , alternative explanations for the emsto
have been put forward .
bastian & de mink ( 2009 ) suggested that stellar rotation may cause the emsto feature , as stellar rotation effects the structure of the star and the inclination angle of the star relative to the observer will change the effective temperature , hence observed colour .
however , girardi et al . ( 2011 ) calculated isochrones for a modestly rotating star and found that the extended lifetime of the star , due to rotation , may cancel the stellar structure effects , resulting in a non - extended msto .
platais et al .
( 2012 ) found that in the intermediate age ( @xmath13 gyr ) galactic cluster trumpler 20 , that modestly rotating stars were blue shifted from the nominal main sequence , in agreement with the girardi et al .
( 2011 ) predictions , suggesting that stellar rotation may not be the cause of the emsto phenomenon .
however , no rapid rotating stars were present in the cluster , and further studies are needed to confirm this as a general result .
yang et al . (
2013 ) used the dartmouth stellar evolutionary code to investigate these contradictory results , and found that the mixing efficiency caused by rotation is a key parameter , and that in fact stellar rotation may cause emstos , depending on the specific choice of the rotational mixing parameter .
milone et al .
( 2013 ) found that while the young ( @xmath14 myr ) cluster ngc 1844 in the lmc did not have an age spread , its cmd does display features that are not well described by an ssp model , namely a broadened main sequence at similar magnitudes as the emsto in the intermediate age clusters .
hence , as cmds are studied in more detail , new features are still being discovered . hence , the cause of the emsto is still under debate , with some studies still advocating age spreads as the best explanation ( e.g. , correnti et al .
one way to test the above scenarios is to look for other features in the cmd that may also reflect age spreads .
girardi et al .
( 2009 ) introduced the red clump as such a test , showing that the red clump in ngc 419 is more elongated than would be expected in a simple stellar population .
interestingly , trumpler 20 also displays an extended or dual red clump " although it does not show any evidence for an age spread within it ( platais et al .
li et al .
( 2014a ) found , on the other hand , that the younger emsto cluster ngc 1831 , has a compact red - clump which is inconsistent with the presence of a significant age spread within the cluster .
it is currently unclear what the role of rotation will be on the red clump morphology .
extra - mixing from rotation would likely give larger core masses , which may result in an extended red clump , however this effect may be compensated by a longer main sequence lifetime .
detailed modelling is necessary to address this .
another feature of the cmd that is sensitive to age spreads is the width of the sub - giant branch ( e.g. , marino et al .
li et al .
( 2014b ) have shown that the sgb width in ngc 1651 , an intermediate age lmc cluster that displays the emsto phenomenon , is consistent with a simple stellar population ( i.e. , with no age spread ) .
the authors place an upper limit of 80 myr on any age spread within the cluster , significantly less than the age spread of 450 myr inferred from the emsto . while the sgb width does suffer from lower numbers of stars than the msto or red clump regions of the cmds , a number of clusters contain enough stars to place strict limits on
any age spreads present within the cluster .
here we use the same technique as li et al .
( 2014b ) , studying the width of the sgb in two intermediate age lmc clusters that display the emsto phenomenon , namely , ngc 1806 & ngc 1846 . in
[ sec : obs ] we present the observations , while in [ sec : results ] we analyse the cmds of the two clusters , testing the hypothesis of significant age spreads within the clusters . finally , in
[ sec : conclusions ] we discuss our results and present our conclusions .
for the present work we use optical hubble space telescope ( hst ) imaging taken with the advanced camera for surveys ( acs ) wide field camera ( wfc ) from programmes go-9891 ( pi gilmore ) and go-10595 ( pi goudfrooij ) .
the observations , reduction , and photometry are all described in detail in milone et al .
we use the field - star subtracted ( based on a suitable nearby reference field ) catalogues for our analysis , and refer the interested reader to milone et al . ( 2009 ) for details on this process .
we have corrected the observed photometry for extinction , using values of @xmath15 and @xmath16 and distance moduli of 18.5 and 18.42 for ngc 1806 and 1846 ( goudfrooij et al .
2011b ) , respectively .
the values were taken from goudfrooij et al .
( 2011b ) , although independent analysis using the present datasets found good agreement . for our analysis
we adopt the stellar isochrones of marigo et al .
( 2008 ) , with z=0.008 and y=0.25 . however , we note that the conclusions of this paper are unchanged if we would have adopted the basti ( pietrinferni et al .
2004 ) or the dartmouth ( dotter et al .
2008 ) isochrones with similar parameters .
while the absolute ages of the clusters do depend on the model adopted , the relations between the msto , sgb and rc positions remain unchanged , i.e. if an age spread is present it will show up in the same way in each of the different portions of the cmd .
[ fig : ngc1806_cmd ] shows the cmd of ngc 1806 , centred on the msto , sgb , and red clump portions of the distribution .
( blue ) lines show isochrones of age 1.41 ( top ) , 1.58 ( middle ) and 1.86 gyr ( bottom ) , i.e. the range required to explain the observed spread in the msto region .
some of the spread observed in the msto region is due to binaries and photometric errors , hence the isochrones used to bracket the emsto are upper limits to any actual spread within the clusters .
it should be noted that different authors have derived somewhat conflicting age spreads , with goudfrooij et al .
( 2011b ) suggesting age spreads of @xmath1 myr for ngc 1806 and milone et al .
( 2009 ) suggesting a lower value of @xmath17 myr .
we identify candidate sgb branch stars between @xmath18 as ( red ) diamonds .
as can already be seen from the figure , the sgb stars cluster around the youngest isochrone needed to explain the emsto feature , and do not span the full range expected if large age spreads were present in the cluster .
the average photometric errors were derived from the observed main sequence distribution below the msto . to do this we verticalised " the main sequence between @xmath19 .
we then made histograms of the stars in @xmath20 in 0.25 magnitude bins , and fit gaussians to the distribution .
while there was clearly a tail of stars to the red ( due to binaries ) the core of each distribution was well described by a gaussian with dispersion of @xmath21 mag .
we took this to be the typical error in colour and correspondingly an error of @xmath22 mag in brightness .
we begin by estimating the age spread present as inferred by the extended msto . to do this
, we take a slice through the data ( shown as the solid red box in fig .
[ fig : ngc1806_cmd ] ) , and find the best fitting isochrone for each observed star .
this is similar to what was done in goudfrooij et al .
( 2011b ; 2014 ) , although we note that we have not attempted to include the effects of binaries in this simple calculation . the resulting age distribution is shown in fig .
[ fig : ngc1806_age_dist_msto ] .
we then fit a gaussian to the distribution ( shown as a blue dash - dotted line ) to the data , and the best fit parameters are given in the panel .
we will use this age distribution in constructing synthetic cmds to test if other regions of the cmd are consistent with the estimated sfh .
we then generated two synthetic cluster cmds .
the first is the expected distribution if the emsto was due to an age spread ( i.e. adopting the age spread in fig .
[ fig : ngc1806_age_dist_msto ] ) .
the second was under the assumption of a single age population ( i.e. , an ssp ) with an age of 1.44 gyr .
synthetic stellar populations were then made adopting a salpeter ( 1955 ) stellar imf and the marigo et al .
( 2008 ) isochrones , applying representative errors taken from the observations .
stars along the same colour / magnitude region used to define the sgb and the rc in the observations were selected and analysed in the same way as the observations . in appendix
[ sec : appendix_test ] we test all the used techniques against synthetic clusters with known ( input ) age spreads .
additionally , we test our relatively simplistic age spread estimate based on the cut across the emsto ( note that this is the same technique adopted by goudfrooij et al .
( 2011b ; 2014 ) ) against a full sfh based on the starfish package . for the sgb , we can quantify the offset between the observed distribution and that expected from a population with an extended sfh or a single burst .
this is done by measuring the magnitude offset of each observed star ( or a star from our synthetic cmd ) identified as being part of the sgb from a nominal isochrone .
we adopt an isochrone at 1.44 gyr , as this best describes the observed sgb in this cluster , and looked at the difference between each observed sgb star and the nominal isochrone .
the results are shown in fig .
[ fig : ngc1806_age_dist_sgb ] .
we also show the expected magnitude difference between other isochrones and the nominal isochrone as vertical dashed lines ( labelled in the panel ) . as can be seen in fig .
[ fig : ngc1806_age_dist_sgb ] , the extended sfh simulation ( the average of 1,000 monte carlo realisations ) fails to reproduce both the peak of the observed distribution as well as the width .
the expected and observed peaks in the distribution are offset by @xmath23 mag .
the ssp simulation well describes the width of the core of the observed histogram , but the observations display extended tails on both sides , suggesting that photometric errors alone can not explain the full distribution . however , we note that the observed spread in the distribution ( corrected for photometric uncertainties ) is an upper limit to the actual spread , as we have not corrected for the effects of differential extinction , nor binarity , in our sample
. additionally , if stellar rotation is affecting the msto then we may expect to see some influence on the sgb , which causes some spread in the observed sgb stars ( e.g. , girardi et al .
2011 ) .
finally , we note that the base of the red - giant branch also appears narrower than would be expected if large age spreads were present .
again , the stars cluster along the youngest isochrone that fits the msto region . in fig .
[ fig : ngc1806_cmd_rc ] we show a zoom in of the cmd of ngc 1806 centred on the base of the red giant branch and the red clump ( rc ) .
additionally , we show four isochrones with ages between 1.31 and 1.77 gyr . as was the case for the sgb
, we note that the position of the observed stars do not conform to the expectations if there was a significant age spread present within the cluster . while a significant spread in colour is expected , there is little spread in observations .
we quantify this in a similar way as was done for the sgb .
we simply assign an age to each of the observed ( and synthetic ) stars based on the closest isochrone .
the results are shown in fig .
[ fig : ngc1806_age_dist_rc ] . note that the rc width is consistent with the expectations of an ssp , while an extended sfh results in a distribution that is clearly at odds with the observations .
the reason why this pseudo - age distribution " is not symmetric ( the input age distribution is ) is that the isochrones bunch up at older ages at the rc , so small photometric errors may lead to large age differences .
we have explored the effect of binaries on the sgb morphology , by creating cmds of synthetic clusters with and without binaries . for the binary populations we assume a flat mass ratio and differing binary fractions , ranging from zero ( only single stars ) to 1 ( all stars have binary companions ) . only in the cases of high binary fraction
was the sgb significantly affected , shifting the stars to brighter magnitudes .
for realistic binary fractious we found that the sgb is only marginally broadened by binaries , consistent with the more detailed simulations of milone et al .
( 2009 - their figs .
@xmath24 ) .
the reason for this is the relatively low fraction of near - equal mass binaries in the clusters ( @xmath25% - milone et al .
2009 ) . due to the intrinsic brightness of sgb stars
, low mass companions do not significantly contribute to the combined flux of a binary pair .
due to the relatively short duration of the rc phase in stellar evolution , and the fact that rc stars are much brighter than their ms counterparts , binaries are not expected to have a large effect on the rc position or morphology . only in the rare circumstance that a rc star has an rc star binary companion
would the magnitude be significantly affected .
vertical dashed lines show the expected magnitude difference of different isochrones , labelled in the panel .
the dash - dotted ( red ) line shows the expected distribution of stars if the underlying distribution was an ssp convolved with the observational errors .
the dash - dot - dot ( blue ) line shows the expected distribution adopting the parameters that best fit the emsto region ( assuming that an age spread is the cause ) .
all distributions shown are for the same number of stars in the sgb region ( @xmath26 in this case).,width=302 ] we have carried out a similar analysis of the colour - magnitude diagram of ngc 1846 , again focussing on the sgb and rc regions .
the cmd of ngc 1846 is shown in fig .
[ fig : ngc1846_cmd ] . as was the case for ngc 1806 ,
different authors have suggested different age spreads for this cluster , with milone et al .
( 2009 ) suggesting a spread of @xmath27 myr , rubele et al .
( 2013 ) finding @xmath28 myr and goudfrooij et al .
( 2011b ) suggesting @xmath29 myr .
we adopt a nominal isochrone of @xmath30 gyr for our analysis .
we carried out the same analysis as was done for ngc 1806 , and the results are shown in figs .
[ fig : ngc1846_age_dist_msto ] , [ fig : ngc1846_age_dist_sgb ] , [ fig : ngc1846_cmd_rc ] & [ fig : ngc1846_age_dist_rc ] . as was found for ngc 1806 ,
the expected sgb morphology for an extended sfh ( as inferred from the emsto region ) is clearly offset from the observed distribution ( fig .
[ fig : ngc1846_age_dist_sgb ] ) .
the sgb morphology suggests a significantly lower age ( @xmath31 gyr ) than that inferred from the emsto region ( @xmath32 gyr ) , and also a narrower age spread .
the inferred age from the sgb is that of the youngest isochrone needed to match the emsto , which is @xmath30 gyr , similar to that found for ngc 1806 . from this we conclude that the sgb morphology is inconsistent with the presence of a significant age spread within the cluster
additionally , from figs .
[ fig : ngc1846_cmd_rc ] & [ fig : ngc1846_age_dist_rc ] it is clear that the rc position and morphology is also inconsistent with the inferred age spread from the msto , being best described by a single isochrone ( an ssp ) . as was the case for ngc 1806 , we again note that we have not included the effects of binarity nor differential extinction ( and also for the potential spread caused by stellar rotation ) , hence any observed spread is an upper limit to the intrinsic spread in the distribution .
additionally , there is likely some amount of stellar field star contamination in the used cmd .
we have also tested our results using different filter combinations ( e.g. , including @xmath33 ) based on new photometry , milone et al .
( in prep ) , and find consistent results .
our results are not in agreement with those reported by rubele et al .
( 2013 ) who fit the sfh of ngc 1846 with the starfish code ( harris & zaritsky 2001 ) and found that the cmd was best reproduced by an extended sfh .
their fit included the sgb and rc portions of the cmd , as well as the msto , so it could have found an inconsistency between the different cmd regions .
the code , however , is weighted to regions with the most stars , hence was heavily weighted to the msto region of the cmd .
figure 7 of rubele et al .
( 2013 ) clearly shows high @xmath34 values for this cluster at the position of the rc , inferring that the best fitting extended sfh did not reproduce this region , consistent with what we have reported here .
mackey et al . (
2013 ) have found evidence that ngc 1846 displays systematic rotation , similar in magnitude to the velocity dispersion of the cluster .
one explanation for this phenomenon is that a dynamically colder 2nd generation " formed with a net rotation within an existing 1st generation " ( e.g. , bekki 2010 ) .
however , since the cluster does not display any abundance variations ( mackey et al . in prep . )
it is not possible to differentiate between the 1st " and 2nd generations " ( only the
2nd generation " stars would be expected to show rotation in this model ) .
hence , it may simply be that the cluster as a whole rotates , with no relation to the emsto phenomenon or any potential age spreads within the cluster .
indeed , rotation may be a common property of massive clusters , and has been observed in young massive clusters , e.g. , r136 and glimpse - c01 , hnault - brunet et al .
( 2012 ) and davies et al .
( 2011 ) , respectively . but now for ngc 1846.,width=302 ] but now for ngc 1846.,width=302 ] ) .
, width=302 ] , width=302 ]
the two clusters analysed in the present work ( ngc 1806 & ngc 1846 ) display sub - giant branches ( sgbs ) and red clumps ( rcs ) that are significantly narrower and offset from what would be expected if age spreads of @xmath35 myr would be present within the clusters .
the observed spreads are in fact upper limits to the actual spreads , as we have not accounted for the affects of differential extinction ( platais et al . 2012 ) or binarity in our sample ( or for the potential effects of stellar rotation ) .
li et al .
( 2014b ) have found a similar behaviour in the cmd of ngc 1651 , confirming that the lack of spreads in the sgb is common in clusters that display an emsto .
hence , our results are not consistent with the interpretation of age spreads being the cause of the emsto phenomenon , in contradiction with previous claims ( e.g. , goudfrooij et al . 2011b ; rubele et al .
2013 ; correnti et al .
2014 ) .
rubele et al . (
2013 ) fit the star - formation history of two intermediate age clusters , ngc 1846 and ngc 1783 , under the assumption that the emsto was due to an age spread .
the authors used the starfish star - formation history code ( harris & zaritsky 2001 ) which compares synthetic observations based on linear combinations of theoretical isochrones ( plus errors ) to the observations , performing a @xmath34 minimisation .
the authors report age spreads of @xmath36 myr for the two clusters .
much of the weight of the analysis comes from the msto region , due to the large number of stars in that phase , and a relatively low weight is applied to portions of the cmd with fewer stars ( e.g. , sgb , rgb , rc ) .
however , we note that for both clusters , the best fit extended sfh fails to reproduce the morphology of the red clump ( in particular in the @xmath37 filters ) , in the sense that the models predict too many red rc stars .
we also note that the position of the rc was not matched in the analyses of ngc 1806 , 1846 , 411 , 1718 and 2203 by goudfrooij et al .
( 2011a ; 2014 ) , again calling into question the interpretation of large spreads within these clusters .
this is similar to the results found in the present work .
we will analyse the cmd of ngc 1783 in detail in a future work ( niederhofer et al . in preparation ) . based on the sgb and rc locations ( and also on the base of the red - giant branch ) , we infer that ngc 1806 and 1846 have younger ages than has been reported in the past , with the best fitting age from the sgb agreeing with the youngest age inferred from the msto region ( see [ sec : appendix_metallicity ] for a discussion of the degeneracies between age and metallicity in the cmd fitting of these clusters ) .
hence , it appears that some effect is causing stars to move red - ward in colour space at the msto .
ngc 1651 displays the same behaviour , with an inferred age of @xmath38 gyr , suggesting that it is a general property of the emsto intermediate age clusters , and not an age effect ( li et al .
this is the behaviour predicted if stellar rotation was the cause of the emsto ( bastian & de mink 2009 ; yang et al .
2013 ) . at present , it is unclear how rotation will affect the sgb and rc , which we will leave to a future work .
goudfrooij et al . (
2014 ) found a correlation between the fraction of blue / bright stars on the msto ( i.e. , young stars if interpreted as due to age effects ) and the fraction of secondary rc " stars to the main body of rc stars , i.e. stars that are bluer and fainter than the main red clump .
the authors use this to argue for the presence of age spreads within their cluster sample , although , as shown here , the main body of the rc does not show evidence for age spreads .
goudfrooij et al .
( 2011 , 2014 ) have suggested that a correlation exists between the width of the emsto and the escape velocity of the cluster .
however , this result depends heavily on large correction factors " to change the current observed escape velocities to the initial ones " . without these correction factors many of the clusters that display the emsto feature would have lower escape velocities than young clusters that have been shown not to have age spreads within them ( e.g. , bastian & silva - villa 2013 ; niederhofer et al .
2015 ) . the applied
correction factors " are based on a number of extreme assumptions .
first , the models of cluster dissolution that are adopted are not appropriate to clusters in the lmc ( e.g. , the tidal field experienced by the model clusters is a milky way like galaxy at @xmath39 kpc from the galactic centre , unlike the much lower tidal field of the lmc - e.g. , lamers et al .
secondly , it is assumed that the clusters begin their lives tidally limited in order to maximise cluster mass - loss , whereas most clusters in the lmc / smc that have been studied to date are not tidally limited ( e.g. , glatt et al .
this means that cluster mass loss is in fact minimal , regardless of the initial state of mass segregation present in the cluster .
hence , realistic correction factors would be much smaller than those adopted in goudfrooij et al .
( 2011b ; 2014 ) and correnti et al .
( 2014 ) .
we can also use the cluster population properties of the lmc to test the goudfrooij et al .
( 2011b ; 2014 ) scenario . while the current masses of the emsto clusters are similar to the most massive ymcs ( e.g. , ngc 1850 , 1866 and 1856 - niederhofer et al .
2015 ) , goudrooij et al .
( 2011 ; 2014 ) posit that the intermediate age clusters were much more massive ( by a factor of @xmath40 ) at birth ( and also significantly denser ) . based on size - of - sample effects ( e.g. , gieles & bastian 2008 ) we would expect that the sfr of the lmc was @xmath40 times higher while the emsto clusters were forming ( c.f . , maschberger & kroupa 2011 ) .
however , no such increase in the sfh , based on resolved stars , is seen from @xmath0 gyr ago relative to today in the lmc ( e.g. , harris & zaritsky 2009 ; weisz et al .
the lack of increase in the sfh from @xmath0 gyr is consistent with the fact that the clusters in this age range would have been born with similar masses ( at the upper end of the mass function ) as that observed in the ymcs in the lmc today .
as discussed above , the clusters are unlikely to be tidally limited so their current masses should be good approximations of the initial masses .
this can be quantified by using the lmc cluster sample of baumgardt et al .
there are nine clusters more massive than @xmath41 ( a high value was taken to avoid any completeness limit issues ) with ages less than @xmath42 myr , or 0.018 clusters / myr . taking the same mass limit but looking at the @xmath0 gyr age range , we find 15 such clusters , or 0.015 clusters / myr , very similar to that found for the past @xmath42 myr .
hence , the cluster population suggests that the cluster formation rate , like the sfr derived from cmd analyses , was roughly the same for the past @xmath42 myr as it was from @xmath0 gyr in the lmc . neither the sfh , based on resolved stars , nor the cluster population suggest an increase of a factor of @xmath40 in the lmc from @xmath0 gyr relative to now , that would be expected in the goudfrooij et al .
( 2011b ; 2014 ) scenario .
we conclude that it is unlikely that the emsto cluster had masses , hence escape velocities , significantly higher in the past .
girardi et al .
( 2009 ) studied at smc intermediate age clusters , ngc 419 , focussing on the rc morphology .
they find that the msto is broader than would be expected from a nominal isochrone , and suggest an age spread of @xmath1 myr within the cluster .
we note that the sgb appears to show the same morphology , i.e. clusters towards the youngest isochrones , as ngc 1806 and ngc 1846 studied here .
however , the rc position is consistent with that expected for an older age , unlike that found for the clusters studied here .
our results corroborate a number of other results from the literature , that have shown that age spreads of hundreds of myr are not present within clusters . as discussed in the introduction
, a number of studies have searched for evidence of ongoing star - formation , or large age spreads , within older ( @xmath43 myr ) clusters , with none found ( e.g. , perina et al .
2009 ; larsen et al . 2011 ; bastian & silva - villa 2013 , bastian et al .
2013a ; niederhofer et al .
additionally , cabrera - ziri et al .
( 2014 ) used an integrated optical spectrum of cluster 1 in ngc 34 , a @xmath44 , 100 myr cluster , to estimate its star - formation history .
the cluster was found to be well described by an ssp ( i.e. no age spread was found ) , with upper limits of @xmath45% of the cluster mass being made up of younger stars .
the authors have extended this study by looking at a number of other massive @xmath46 clusters , including uv and optical photometry in the analysis , and have not found evidence for extended sfhs ( cabrera - ziri et al . in preparation ) .
additionally , bastian & strader ( 2014 ) searched for gas and dust in 12 lmc / smc ymcs , the presence of which is required if new stars are to form , and found none in any cluster , down to limits of @xmath47% of the stellar mass .
cabrera - ziri et al .
( 2015 ) extended this analysis to study seven massive clusters ( @xmath48 ) in the antennae merging galaxies , using the atacama large millimetre array ( alma ) to search for gas within the clusters ( with ages of @xmath49 myr ) .
no gas was detected within the clusters , with an upper limit of @xmath50% of the stellar mass present in gas .
mucciarelli et al . (
2008 ) searched for chemical spreads ( expected if large age spreads exist , in particular if self - enrichment scenarios are invoked ) in three lmc clusters that display the emsto phenomenon , and no abundance spreads were found .
mucciarelli et al .
( 2014 ) have extended this analysis to ngc 1806 ( also studied in the present work ) and again the authors have not found any abundance spreads .
finally , mackey et al .
( in preparation ) have also searched for abundance spreads , targeting ngc 1846 , and like for the other young and intermediate age clusters studied to date , no chemical spreads were found . while a lack of abundance spreads does not necessarily mean that age spreads can not exist within the clusters , it does however require either 1 ) cluster / cluster mergers with different ages ( although this would not lead to an extended msto , but rather a dual msto ) or 2 ) that the cluster was able to remove stellar ejecta ( e.g. , the ejecta of agb stars ) at the same time as it was accreting primordial gas from its surroundings , and not have the two gas flows interact and mix .
hence , significant fine - tuning would be required .
taken together , we conclude that age spreads are not the cause of the emsto feature found in many intermediate age ( @xmath0 gyr ) clusters .
hence , alternative models , such as stellar rotation or interacting binaries should be investigated further .
new alternative scenarios should also be encouraged and developed .
we are grateful to antonino milone for providing his photometry and catalogues .
we would like to thank dougal mackey , gary da costa selma de mink , and richard de grijs for helpful discussions and comments on the manuscript .
the anonymous referee is thanked for suggestions that helped improve the paper .
nb is partially funded by a royal society university research fellowship .
the results presented here are partially based on observations made with the nasa / esa hubble space telescope , and obtained from the hubble legacy archive , which is a collaboration between the space telescope science institute ( stsci / nasa ) , the space telescope european coordinating facility ( st - ecf / esa ) and the canadian astronomy data centre ( cadc / nrc / csa ) .
99 bastian , n. , & de mink , s. e. 2009 , mnras , 398 , l11 bastian , n. , & silva - villa , e. 2013 , mnras , 431 , l122 bastian , n. , cabrera - ziri , i. , davies , b. , & larsen , s. s. 2013a , mnras , 436 , 2852 bastian , n. , lamers , h. j. g. l. m. , de mink , s. e. , et al . 2013b , mnras , 436 , 2398 bastian , n. , & strader , j. 2014 , mnras , 443 , 3594 baumgardt , h. , parmentier , g. , anders , p. , & grebel , e. k. 2013 , mnras , 430 , 676 bekki , k. 2010 , apjl , 724 , l99 cabrera - ziri , i. , bastian , n. , davies , b. , et al .
2014 , mnras , 441 , 2754 cabrera - ziri , i. , et al .
2015 , mnras , submitted conroy , c. , & spergel , d. n. 2011 , apj , 726 , 36 correnti , m. , goudfrooij , p. , kalirai , j. s. , et al .
2014 , apj , 793 , 121 davies , b. , bastian , n. , gieles , m. , et al .
2011 , mnras , 411 , 1386 dercole , a. , vesperini , e. , dantona , f. , mcmillan , s. l. w. , & recchi , s. 2008 , mnras , 391 , 825 dotter , a. , chaboyer , b. , jevremovi , d. , et al .
2008 , apjs , 178 , 89 gieles , m. , & bastian , n. 2008 , a&a , 482 , 165 girardi , l. , rubele , s. , & kerber , l. 2009 , mnras , 394 , l74 girardi , l. , eggenberger , p. , & miglio , a. 2011 , mnras , 412 , l103 girardi , l. , goudfrooij , p. , kalirai , j. s. , et al .
2013 , mnras , 431 , 3501 glatt , k. , grebel , e. k. , jordi , k. , et al .
2011 , aj , 142 , 36 goudfrooij , p. , puzia , t. h. , kozhurina - platais , v. , & chandar , r. 2011a , apj , 737 , 3 goudfrooij , p. , puzia , t. h. , chandar , r. , & kozhurina - platais , v. 2011b , apj , 737 , 4 goudfrooij , p. , girardi , l. , kozhurina - platais , v. , et al .
2014 , apj , 797 , 35 gratton , r. g. , carretta , e. , & bragaglia , a. 2012 , a&arv , 20 , 50 harris , j. , & zaritsky , d. 2001 , apjs , 136 , 25 harris , j. , & zaritsky , d. 2009 , aj , 138 , 1243 hnault - brunet , v. , gieles , m. , evans , c. j. , et al .
2012 , a&a , 545 , l1 keller , s. c. , mackey , a. d. , & da costa , g. s. 2011 , apj , 731 , 22 keller , s. c. , mackey , a. d. , & da costa , g. s. 2012 , apjl , 761 , l5 krause , m. , charbonnel , c. , decressin , t. , meynet , g. , & prantzos , n. 2013 , a&a , 552 , a121 lamers , h. j. g. l. m. , gieles , m. , & portegies zwart , s. f. 2005 , a&a , 429 , 173 larsen , s. s. , de mink , s. e. , eldridge , j. j. , et al .
2011 , a&a , 532 , a147 li , c. , de grijs , r. , & deng , l. 2014a , apj , 784 , 157 li , c. , de grijs , r. , & deng , l. 2014b , nature , 516 , 367 longmore , s. n. , kruijssen , j. m. d. , bastian , n. , et al .
2014 , ppvi , in press ( arxiv:1401.4175 ) mackey , a. d. , & broby nielsen , p. 2007
, mnras , 379 , 151 mackey , a. d. , broby nielsen , p. , ferguson , a. m. n. , & richardson , j. c. 2008 , apjl , 681 , l17 mackey , a. d. , da costa , g. s. , ferguson , a. m. n. , & yong , d. 2013 , apj , 762 , 65 marigo , p. , girardi , l. , bressan , a. , et al .
2008 , a&a , 482 , 883 marino , a. f. , milone , a. p. , sneden , c. , et al .
2012 , a&a , 541 , a15 maschberger , t. , & kroupa , p. 2011 , mnras , 411 , 1495 milone , a. p. , bedin , l. r. , piotto , g. , & anderson , j. 2009 , a&a , 497 , 755 milone , a. p. , bedin , l. r. , cassisi , s. , et al .
2013 , a&a , 555 , a143 mucciarelli , a. , carretta , e. , origlia , l. , & ferraro , f. r. 2008 , aj , 136 , 375 mucciarelli , a. , dalessandro , e. , ferraro , f. r. , origlia , l. , & lanzoni , b. 2014 , apjl , 793 , l6 niederhofer , f. , hilker , m. , bastian , n. , silva - villa , s. , 2015 , a&a , in press ( arxiv:1501.02275 ) perina , s. , barmby , p. , beasley , m. a. , et al .
2009 , a&a , 494 , 933 pietrinferni , a. , cassisi , s. , salaris , m. , & castelli , f. 2004 , apj , 612 , 168 platais , i. , melo , c. , quinn , s. n. , et al .
2012 , apjl , 751 , l8 rubele , s. , girardi , l. , kozhurina - platais , v. , et al . 2013 ,
mnras , 430 , 2774 salpeter , e. e. 1955 , apj , 121 , 161 weisz , d. r. , dolphin , a. e. , skillman , e. d. , et al .
2013 , mnras , 431 , 364 yang , w. , bi , s. , meng , x. , & liu , z. 2013 , apj , 776 , 112
in order to test whether the age spreads derived from the emsto accurately reproduce the observations ( i.e. whether the cut perpendicular to the msto is not affected by differing lifetimes of stars in that phase in the narrow mass range considered ) we have performed a series of tests , which are discussed here .
we note , again , that this technique has been used previously , namely in goudfrooij et al .
( 2011a , b ; 2014 ) .
first we created a synthetic population of stars with an age spread similar to the ones inferred from the perpendicular cut across the emsto for ngc 1806 and ngc 1846 .
for this we adopted the same isochrones used to analyse the observations ( same metallicity ) , assumed a salpeter ( 1955 ) stellar initial mass function , and created a population with a gaussian age spread with a peak at 1.59 gyr and a dispersion of 136 myr .
no extinction was added to the data .
additionally , we adopted the same distance modulus used for ngc 1806 .
we then ran this synthetic distribution through the same procedures used in the present work .
the results are shown in figs .
[ fig : syn_cmd]@xmath51[fig : syn_age_dist_rc ] .
we find that the peak of the age distribution is well recovered , and that the dispersion found in the analysis techniques results in a slightly ( @xmath52% ) smaller dispersion than is input .
hence , we are slightly underestimating the actual age spread within the cluster using the analyses technique of cutting across the emsto .
however , as can be seen in figs .
[ fig : syn_cmd]@xmath51[fig : syn_age_dist_rc ] , such an age spread would easy be detected in the morphologies and positions of the sgb and rcs .
we have run the above tests for a number of ages between 1 and 2 gyr and age dispersion of @xmath53 myr and found consistent results .
however , for ages of @xmath54 gyr and above it is difficult to infer short age dispersions using the techniques used here .
age spreads can still be recovered , however their actual extent becomes highly uncertain .
this is discussed in detail in keller , mackey & da costa ( 2011 ) .
in addition to the synthetic cluster experiments discussed above , we have also fit the observed cmds of ngc 1806 and ngc 1846 with the starfish ( harris & zaritsky 2001 ) star - formation history code .
this is the same technique which was used by rubele et al .
we adopted the same isochrones as used in the other analyses performed in the current work ; adopting the same metallicity , distance modulus , and extinction as before .
this starfish code finds the best - fitting model by creating linearly combined synthetic cmds of simple stellar populations ( i.e. single isochrones ) and comparing them to the data .
we used isochrones between 8.70 @xmath55 log(t / yr ) @xmath55 9.48 , logarithmically spaced in steps of 0.03 . as with the previous experiments ,
we adopt a salpeter ( 1955 ) stellar imf.we created an analytical model for the observational errors , assuming an error for the brightest stars in the clusters of about 0.01 which increases exponentially to 0.04 at a magnitude of about 24 in the _ f435w _ band .
furthermore , we assumed a binary fraction of 0.3 for both clusters . in order to compare directly with the adopted sfh based on the cut along the msto , we restricted the starfish fitting to a region around the msto .
we chose boxes with the following limits : 0.5 @xmath56 1.5 and 19.5 @xmath57 22.00 .
the results for ngc 1806 and ngc 1846 are shown in figs .
[ fig : starfish ] .
the histogram shows the derived sfh from starfish while the dashed ( blue ) lines show the adopted distribution based on the cut across the emsto .
we find good agreement between the two methods , hence the analysis performed in the current work is able to provide a reasonable estimate of the potential age spread along the msto .
it is this potential age spread that has been tested against the morphology and position of the sgb and rc . in both regions
we do not find evidence for such age spreads . ) .
the histogram shows the results of fitting the msto region of each cluster with the star - formation history fitting code starfish .
note that both methods result in very similar age distributions.,title="fig:",width=302 ] ) .
the histogram shows the results of fitting the msto region of each cluster with the star - formation history fitting code starfish .
note that both methods result in very similar age distributions.,title="fig:",width=302 ] in order to test whether other combinations of age and metallicity can successfully reproduce the observed cluster cmds , in fig .
[ fig : metallicity ] we show isochrones ( marigo et al .
2008 ) of various ages ( 1.32 , 1.44 , 1.58 and 1.9 gyr ) and metallicities ( @xmath58 ) focussing on the position of the red clump . additionally , we show the data of ngc 1806 .
we adopt the same distance modulus and extinction for ngc 1806 as used in the text ( [ sec : obs ] & [ sec : ngc1806 ] ) .
the metallicity range shown is the same as the uncertainties for this cluster ( and all clusters ) in the sample of goudfrooij et al .
( 2011a , b , 2014 ) based on their analysis of the cmds . as can be seen , no combination of age and metallicity can fit the position of the rc for metallicities of @xmath59 and @xmath60 ( for the adopted values of extinction and distance modulus ) .
we have also investigated whether such metallicities can be consistent with the data if we also allow the distance modulus and extinction to vary .
based on the compact morphology of the rc , the age of the population is constrained to @xmath61 gyr for the metallicities considered here . in each case
we adjusted the extinction and distance modulus to best match the position of the rc and the main sequence , for ages between @xmath62 gyr and @xmath63 gyr .
the results are shown in figs .
[ fig : metallicity2 ] and [ fig : metallicity3 ] , for isochrones with @xmath59 and @xmath60 , respectively . for the case of @xmath59 ,
the best fitting parameters are @xmath64 and distance modulus@xmath65 .
this distance modulus would put ngc 1806 @xmath66 kpc in front of the main body of the lmc . for this choice of metallicity
, we see that the rc is best fit with an age of @xmath31 gyr , which is the same as the blue side of the emsto .
this is the same as found ( in [ sec : conclusions ] ) for the case of @xmath67 .
however , in the @xmath59 case the sgb is best described by an isochrone of age of @xmath68 gyr , slightly older than that found for the blue side of the emsto and position of the rc . for the @xmath60 isochrones ,
the best fit is obtained for an @xmath69 and distance modulus of 0.03 and 18.38 , however we note a mis - match between the isochrones and the observed main sequence at faint magnitudes , which increases for fainter stars . however ,
if the @xmath60 isochrones are adopted , the position of the rc is still best reproduced by the youngest isochrone needed to explain the emsto ( @xmath70 gyr ) . in this case , the sgb is also consistent with a single isochrone , but one that slightly older than that needed to explain the rc ( @xmath71 gyr for the sgb compared to @xmath72 gyr for the rc and bluest part of the emsto ) .
we conclude that the ages derived for the clusters in this work based on the position and morphology of the sgb and rc do not suffer from a metallicity degeneracy , and that the sgb and rc are consistent with a single isochrone for all metallicities considered .
for metallicities slightly higher or lower than that adopted , the position of the rc still is best fit by the isochrone that best reproduces the blue side of the emsto .
however , in both cases ( @xmath59 and @xmath60 ) the derived age based on the sgb is slightly older than that of the blue side of the emsto and the rc .
regardless of the adopted metallicity , the observed positions and morphologies of the sgb and rc are inconsistent with the large age spreads inferred from the emsto
. shown as blue squares , @xmath67 shown as green circles , @xmath60 shown as red triangles ) and age from the marigo et al . ( 2008 ) isochrones .
the ages shown are 1.2 , 1.32 , 1.44 , 1.58 and 1.90 gyr from left to right .
we have adopted the same distance modulus and extinction for ngc 1806 as in the other analyses of this cluster ( [ sec : obs ] & sec : ngc1806 ) .
the filled points are the data for ngc 1806.,width=302 ] , and adjusting the distance modulus and extinction in order to best reproduce the position of the rc .
the ages shown are 1.2 , 1.32 , 1.44 , 1.58 and 1.90 gyr from left to right ( or top to bottom on the sgb ) .
the best fit corresponds to an @xmath73 and distance moduls of @xmath74 and @xmath75 , respectively .
, width=302 ] but now showing the results for isochrones of @xmath60 .
the best fitting parameters are a distance modulus of 18.38 mag and an extinction of @xmath76 mag .
the ages shown are 1.2 , 1.32 , 1.44 , 1.58 and 1.90 gyrfrom left to right ( or top to bottom on the sgb ) . however ,
if this metallicity is adopted the rc is still best reproduced by the youngest isochrone needed to explain the emsto.,width=302 ] | a recent surprise in stellar cluster research , made possible through the precision of hubble space telescope photometry , was that some intermediate age ( @xmath0 gyr ) clusters in the large and small magellanic clouds have main sequence turn - off ( msto ) widths that are significantly broader than would be expected for a simple stellar population ( ssp ) .
one interpretation of these extended mstos ( emstos ) is that age spreads of the order of @xmath1 myr exist within the clusters , radically redefining our view of stellar clusters , which are traditionally thought of as single age , single metallicity stellar populations .
here we test this interpretation by studying other regions of the cmd that should also be affected by such large age spreads , namely the width of the sub - giant branch ( sgb ) and the red clump ( rc ) .
we study two massive clusters in the lmc that display the emsto phenomenon ( ngc 1806 & ngc 1846 ) and show that both have sgb and rc morphologies that are in conflict with expectations if large age spreads exist within the clusters .
we conclude that the sgb and rc widths are inconsistent with extended star - formation histories within these clusters , hence age spreads are not likely to be the cause of the emsto phenomenon .
our results are in agreement with recent studies that also have cast doubt on whether large age spreads can exist in massive clusters ; namely the failure to find age spreads in young massive clusters , a lack of gas / dust detected within massive clusters , and homogeneous abundances within clusters that exhibit the emsto phenomenon .
[ firstpage ] galaxies - star clusters , galaxy - globular clusters |
we have developed@xcite two simple models for the @xmath0 gluon plasma equation of state , both based on the use of the polyakov loop @xmath10 as the natural order parameter of the deconfinement transition in pure gauge theories@xcite . in these models ,
the eigenvalues of the polyakov loop are the essential degrees of freedom rather than the polyakov loop trace alone , a possibility also recently explored by pisarski @xcite .
since @xmath10 is unitary , we can parametrize its diagonal form by @xmath11 phases @xmath12 which we refer to as eigenvalues . in our models
, confinement is obtained from a set of spatially constant eigenvalues which make @xmath13 .
this is naturally obtained from a uniform distribution of eigenvalues around the unit circle , constrained by the unitary of @xmath10 . in a pure gauge theory below the deconfinement temperature @xmath14 ,
the eigenvalues are frozen in this uniform distribution . as @xmath15 moves upward from @xmath16 ,
the eigenvalues of the polyakov loop rotate towards @xmath17 or one of its @xmath18equivalents , and @xmath10 moves towards an element of the center . in the case of a first order transition , the eigenvalues jump at @xmath14 .
this motion of the eigenvalues is responsible for the approach to the blackbody limit over the range @xmath1 . as a consequence of asymptotic freedom ,
the perturbative expression for the free energy as a function of the polyakov loop eigenvalues will be valid at sufficiently high temperatures . at one - loop order , the result has the form @xmath19 where @xmath20 has a simple expression in terms of the fourth bernoulli polynomial ; @xmath21 gives the blackbody behavior expected at high temperature .
note that @xmath21 is not a function of @xmath22 , instead depending directly on the eigenvalues .
this distinction begins to matter with @xmath23 , because for @xmath24 , the eigenvalues can not be determined solely from @xmath22 .
the first of our two models is obtained by adding , by hand , a mass @xmath25 to the gauge bosons , and working with the high temperature expansion of the resultant free energy to order @xmath26 .
the result is @xmath27 where @xmath28 is given as sums of second bernoulli polynomials @xcite .
we stress that this derivation merely indicates the type of additional terms that might appear in the free energy .
the bernoulli polynomials appear naturally as class functions which are almost everywhere polynomials in the @xmath29 s , and are well - suited for the construction of a landau theory in the eigenvalues .
our second model is obtained by supposing that there is a natural scale @xmath30 in position space over which color neutrality is enforced . in other words ,
net color is allowed in volumes of less than @xmath31 , but the net color on larger scales is zero .
this leads to @xmath32\ ] ] where @xmath33 is the jacobian associated with haar measure on @xmath0 . in both models , the @xmath34 term , which favors the deconfined phase , dominates for large @xmath15 .
the other term , which favors the confined phase , dominates for small t. the deconfinement phase transition results from a conflict between the two terms .
in the case of @xmath36 and @xmath8 , model a is analytically tractable .
the deconfining transition is second order for @xmath36 and first order for @xmath8 , in accord with universality arguments @xcite .
numerical analysis for the cases of @xmath23 and @xmath37 show that the deconfinement transition is first order ; recent lattice results for @xmath23 also indicate a first order transition@xcite .
identical results on the order of the transition are obtained numerically for model b. in the case of @xmath8 , both models show good agreement with lattice simulations@xcite upon comparing the pressure @xmath38 , the energy density @xmath39 , and the dimensionless interaction measure @xmath40 as a function of the dimensionless variable @xmath41 .
the range of temperatures over which metastable behavior occurs can be easily determined , and confirms the weakly first order character of the transition . versus @xmath42 for model a.,width=288 ] versus @xmath42 for model a.,width=288 ] versus @xmath42 for model a.,width=288 ] it is enlightening to plot @xmath38 , @xmath39 , and @xmath43 , each divided by @xmath44 , versus @xmath41 for @xmath2 , @xmath45 , @xmath4 , and @xmath5 .
figures 1 - 3 show the results for model a ; model b is similar .
both models appear to quickly approach a large-@xmath11 limit .
both models show power law behavior in @xmath43 for sufficiently large @xmath15 , consistent with @xmath46 .
this is the asymptotic behavior found analytically for @xmath36 in model a , and is compatible with the @xmath8 lattice data . in camparison , the bag model@xcite ,
predicts a @xmath47 behavior for @xmath43 , which is ruled out by lattice results .
quark effects can be included as the free energy of quarks propagating in a constant polyakov loop background .
we limit ourselves here to a discussion of very heavy quarks @xcite and the leading order effect of light quarks , which do not depend on chiral symmetry effects @xcite .
the expected effect of very heavy quarks is to lower the critical temperature @xcite .
this line of first - order critical points in the @xmath48 plane terminates in a second - order end point at some finite value of @xmath49 . for light quarks
, we expect that the deconfinement transition is replaced by a smooth crossover with a rapid rise in all thermodynamic quantities .
we have confirmed both sets of behaviors in the case of @xmath8 .
we have developed two simple phenomenological equations of state for the quark - gluon plasma , which reproduce much of the thermodynamic behavior seen in lattice simulations .
both models correctly predict the order of the deconfining phase transitions for @xmath36 ( second order ) and @xmath8 ( weakly first order ) , and predict first order transitions in @xmath23 and @xmath37 as well .
both models appear to have smooth large-@xmath11 limits .
the numerical value of the parameters introduced are reasonable for both models in the case of @xmath8 .
the deconfinement temperature in a pure @xmath8 gauge theory is about @xmath50 , giving a value for @xmath25 of @xmath51 , a plausible value for a constituent gluon mass . in model b , we find that @xmath52is @xmath53fermi .
it is clear that by allowing the parameters @xmath25 and @xmath54 to depend on the temperature , a better fit to lattice data can be obtained at the cost of introducing additional phenomenological parameters .
the success of these phenomenological models suggests a point of view on the nature of the deconfinement mechanism . in confining theories ,
the eigenvalue distributions of the polyakov loop are peaked at low temperature around values evenly spaced about the unit circle in such a way that the expectation value of the polyakov loop is zero . at the deconfining transition temperature , the peaks of the eigenvalue distributions move towards @xmath7 , which is the asymptotic limit as @xmath15 goes to infinity .
we believe that this picture of deconfinement coupled with a field - theory inspired model of chiral symmetry breaking has the potential to fully model the equation of state of the quark - gluon plasma @xcite .
n. meisinger , t. r. miller and m. c. ogilvie , hep - ph/0108009 .
l. g. yaffe and b. svetitsky , phys .
d * 26 * , 963 ( 1982 ) .
r. d. pisarski , phys .
d * 62 * , 111501 ( 2000 ) [ hep - ph/0006205 ] .
p. n. meisinger and m. c. ogilvie , hep - ph/0108026 .
m. wingate and s. ohta , phys .
d * 63 * , 094502 ( 2001 ) [ hep - lat/0006016 ] .
g. boyd , j. engels , f. karsch , e. laermann , c. legeland , m. lutgemeier and b. petersson , nucl .
b * 469 * , 419 ( 1996 ) [ hep - lat/9602007 ] .
j. cleymans , r. v. gavai and e. suhonen , phys .
rept . * 130 * , 217 ( 1986 ) .
p. n. meisinger and m. c. ogilvie , phys .
d * 52 * , 3024 ( 1995 ) [ hep - lat/9502003 ] .
d. j. gross , r. d. pisarski and l. g. yaffe , rev .
phys . * 53 * , 43 ( 1981 ) .
n. weiss , phys .
d * 24 * , 475 ( 1981 ) .
n. weiss , phys .
d * 25 * , 2667 ( 1982 ) .
f. green and f. karsch , nucl .
b * 238 * , 297 ( 1984 ) .
m. ogilvie , phys .
lett . * 52 * , 1369 ( 1984 ) .
p. n. meisinger and m. c. ogilvie , nucl .
suppl . * 47 * , 519 ( 1996 ) [ hep - lat/9509050 ] .
a. dumitru and r. d. pisarski , phys .
b * 504 * , 282 ( 2001 ) [ hep - ph/0010083 ] . | two phenomenological models describing an @xmath0 quark - gluon plasma are presented .
the first is obtained from high temperature expansions of the free energy of a massive gluon , while the second is derived by demanding color neutrality over a certain length scale .
each model has a single free parameter , exhibits behavior similar to lattice simulations over the range @xmath1 , and has the correct blackbody behavior for large temperatures .
the @xmath2 deconfinement transition is second order in both models , while @xmath3,@xmath4 , and @xmath5 are first order .
both models appear to have a smooth large-@xmath6limit . in both models ,
the confined phase is characterized by a mutual repulsion of polyakov loop eigenvalues that makes the polyakov loop expectation value zero . in the deconfined phase ,
the rotation of the eigenvalues in the complex plane towards @xmath7 is responsible for the approach to the blackbody limit over the range @xmath1 .
the addition of quarks in @xmath8 breaks @xmath9symmetry weakly and eliminates the deconfining phase transition for sufficiently light quarks . |
in quantum mechanics , a multiparticle production process is described in terms of a density matrix @xmath0 which characterizes the final state of the system . from the density matrix , all n - particle distributions can be determined , and conversely , a measurement of these distributions yields information about the density matrix of the multiparticle system .
the n - particle inclusive distribution is defined through the creation and annihilation operator @xmath11 and @xmath12 of a particle of momentum @xmath13 ( @xmath14 labels internal degrees of freedom ) : @xmath15 where , @xmath16 is the invariant volume in momentum space .
the general n - particle correlation function is defined as @xmath17 in order to determine the density matrix for a given reaction from first principles , one would have to specify the initial state of the projectile and target and then apply the @xmath18 matrix to this state . in general
this is not possible .
one way to proceed is to parametrize @xmath0 according to a reasonable phenomenological description of the system . for this
, one uses the external source ( current ) formalism . in this
approach particle sources are treated as external classical currents , and their fluctuations are described by a gaussian distribution .
this last choice can be justified by the fact that , if one has a superposition of @xmath19 independent sources , the gaussian form follows from the central limit theorem in the limit of large @xmath19 . in the following
we will also consider correlation functions as functions of @xmath20 : @xmath21 ; @xmath22 ; @xmath23 and @xmath24 . ]
@xmath25 with ( @xmath26 ) , @xmath27\ ] ] @xmath28\ ] ] here we give a brief summary of the derivation of bose - einstein correlation functions in the current formalism @xcite .
the current can in general be written as the sum of a chaotic and a coherent component , @xmath29 . the gaussian current distribution is completely specified by its first two moments : @xmath30=@xmath31 and the two - current correlator + @xmath32 .
@xmath33 and @xmath34 can be parametrized as @xmath35 and @xmath36 , where @xmath37 and @xmath38 are the space - time distributions of the coherent and the chaotic components of the source .
the primordial correlator @xmath39 reflects intrinsic dynamical properties of the source .
it contains some characteristic length ( or time ) scales @xmath40 , so - called correlation lengths ( for a system in thermal equilibrium the correlation length can be related to the inverse of the temperature ) . in the general case of a partially coherent source , the single inclusive distributions of pions
can be expressed also as a sum of a chaotic and coherent component @xmath41 denotes the charge ) : @xmath42 where , @xmath43 @xmath44 @xmath45 and @xmath46 are the on - shell fourier transforms of @xmath33 and @xmath34 , respectively . in general , the chaoticity parameter will be momentum - dependent : @xmath47 to write down the correlations functions in a concise form one introduces the normalized current correlator : @xmath48^{1/2}}\ ] ] where the indices @xmath49 label the particles . since @xmath50 is in general a complex number , one may prefer to express the correlation functions in terms of the magnitudes and the phases of @xmath51 , @xmath52 and the phase of the coherent component @xmath53 with this notation we write the chaoticity parameter : @xmath54 .
the normalized cumulant correlation functions @xmath55 for identical charged particles are : @xmath56 @xmath57 @xmath58 the correlation functions @xmath1 ( 3 ) can be expressed in terms of @xmath55 by using the relation between the corresponding generating functionals @xcite :
@xmath59 @xmath60 @xmath61 as one can see , all correlation functions depend only on the functions @xmath62 , @xmath63 and @xmath64 which will be specified in the next section for an expanding source . for the space - time description of an expanding source
is useful to define variables @xmath65 , @xmath66 and @xmath67 : @xmath68 where @xmath65 is the proper - time , @xmath67 the coordinate in the direction of the collision axis and @xmath66 the space - time rapidity . here
we will consider invariance under boosts of the coordinate frame in the longitudinal direction , which corresponds to the assumption that the single inclusive distribution in rapidity is flat .
the space - time distribution of the chaotic and coherent source , as well as the primordial correlator , and @xmath69 respectively . ] are expressed in terms of 10 parameters : @xmath70 and @xmath71 are the proper time coordinates of the chaotic and the coherent source ; @xmath72 and @xmath73 their widths in proper - time ; @xmath74 and @xmath75 are the transverse radii ; @xmath76 and @xmath77 are the correlation lengths and @xmath78 is the correlation time .
the tenth one is the chaoticity parameter . to reduce the number of parameters we assume now that the widths in proper - time of both sources are vanishing ( @xmath79 ) ;
with this choice the results do not depend on the correlation time @xmath78 .
the space - time distributions of the chaotic and coherent sources are then parametrized as : @xmath80 note that the @xmath66 dependence of @xmath81 was neglected in equations ( 22 ) and ( 23 ) .
this corresponds to a boost - invariant ansatz of the source expansion .
the model contains now @xmath82 independent parameters : @xmath70 , @xmath71 , @xmath74 , @xmath75 , @xmath76 , @xmath77 and the chaoticity parameter @xmath83 . at same arbitrary scale ,
e.g. , @xmath84 ( cf .
( 32 ) ) . ]
we will define , for convenience : @xmath85 the single inclusive distribution will be a sum of a chaotic and a coherent term : @xmath86 e\left|\frac{1}{\sigma}\frac{d^3\sigma}{d^3k}\right|_{k=0}\\ s_{ch}(k)&=&\frac{m_{\pi}}{m_{\bot}}exp\left(-\frac{k_{\bot}^2r_{l}^2}{2}\right)\\ s_{co}(k)&=&\frac{m_{\pi}}{m_{\bot}}exp\left(-\frac{k_{\bot}^2r_{co}^2}{2}\right)\end{aligned}\ ] ] where @xmath87 is the transverse mass of the pions emitted and the momentum dependence of chaoticity parameter takes the form @xmath88 : @xmath89\end{aligned}\ ] ] the magnitudes and phases of @xmath90 , @xmath91 and @xmath64 are : @xmath92\end{aligned}\ ] ] @xmath93 we proceed now to the the calculation of the correlation functions .
the procedure for calculations consists in : + a ) the phase - space is generated by monte carlo routines which take into account the experimental detection conditions mentioned in @xcite ( defined by @xmath94 ) and references quoted there .
, @xmath95 were performed .
we found that in the range of @xmath96@xmath97 the third and fourth order correlation functions are more sensitive to the detections conditions than the second order correlation function . ]
+ b)the simulated produced pions are registered with a determined @xmath98 and azimuth angle @xmath99 and from them one calculates the correspondent @xmath20 value .
the simulated particles are then selected in @xmath20 bins , as described in the experimental papers quoted above .
+ c)the calculation of the functions @xmath90 , @xmath91 and @xmath100 defined for an expanding source model is performed . from these results
one is able to calculate the correlation function ( and its integral for each defined bin ) , as determined by equations ( 4)-(6 ) and ( 18)-(20 ) .
+ d)we use this procedure for second , third and fourth order correlation calculations , @xmath101 and @xmath5 . to simplify the calculations we reduce the number of free parameters using the criteria presented in @xcite .
we assume at first that the chaotic and coherent components have the same transverse momentum spectrum , i.e. , we choose @xmath102 ( if @xmath103 , @xmath75 is obviously not present anymore in the formalism . ) .
@xmath104 is constrained by the relation is taken from experiment . ]
we also take @xmath107 .
we are thus left with only four parameters : @xmath108 and @xmath83 .
these parameters have been investigated in the intervals . ] : @xmath109 ; @xmath110 ; @xmath111 and @xmath112 .
we will use the @xmath113 ( @xmath114 : number of degrees of freedom ) method to evaluate the quality of the fits .
we are looking for a common fit for all @xmath115 using one set of parameters ( @xmath83,@xmath74,@xmath116,@xmath77 ) .
the procedure to select this one set for an overall fit was : + 1 ) calculation of @xmath117 for each @xmath115 ( @xmath118 ) function for all parameter sets to delimit sectors in the parameters phase - space to be investigated .
the groups of parameters which gave @xmath113@xmath119 for any calculated function were eliminated .
+ 2 ) search for regions of intersection in the parameters space for an overall description of the data .
these regions were defined by minimizing @xmath113 for all three functions simultaneously of _ each _ function , since we look for regions in the parameter phase - space which minimize the @xmath113 for _ all _ three functions . ] and imposing @xmath120@xmath121 .
from 1 ) we concluded that there is a large number of parameter combinations which permit acceptable fits for @xmath3 .
the situation is quite different when one extends the search to @xmath4 and @xmath5 .
table 1 shows the groups of parameters we would like to comment on primarily . in all these groups
this table also contains the @xmath113 for each calculation .
figure 1 shows the results for the best overall fit , for @xmath123@xmath97 @xmath124 .
the normalization factor is very close to one . ]
corresponding to the parameters set @xmath125 .
besides @xmath126 .
the @xmath127 functions did not lead to new conclusions about the choice of parameters , they were in general agreement with the behaviour of @xmath128 . ] some results for @xmath129 should be commented .
calculations using @xmath130 and @xmath131@xmath132 can produce good fits for @xmath3 ( @xmath133 ) in @xmath134@xmath135@xmath136 and in the entire @xmath116 interval , but the minimum @xmath137 values for @xmath138 and @xmath139 are around 2.0 in the ( very small ) region of intersection of parameters .
the behaviour of the @xmath140 function does not change even if the radius varies up to @xmath141@xmath132 , but then there does not exist an intersection region of parameters ( with @xmath137@xmath121 ) for @xmath138 and @xmath139 .
the calculations using @xmath142 presented almost the same behaviour .
we found a clear tendency towards a common value of @xmath143@xmath136 .
the prefered range of the radius is @xmath144@xmath145@xmath132 .
the range of @xmath116 seems to be connected with that of the radius .
one gets the best fits for @xmath131@xmath132 with @xmath146@xmath136 and for @xmath147@xmath132 with @xmath148@xmath136 ( see table 1 ) .
value by one unity was obtained .
the results for the set @xmath125 ( @xmath83,@xmath74,@xmath77,@xmath116 ) are : @xmath149 ; @xmath150 ; @xmath151 and @xmath152 . ]
as already mentioned , the number of parameter sets which provide an acceptable fit for @xmath3 is much bigger than the number of parameter sets which provide an acceptable general fit .
this is exemplified in figure 2 .
this figure should also also clarify the method used in this work , which differs from the one used in @xcite , where the parameters obtained from a fit of @xmath140 data were used to predict " @xmath138 .
for the results in figure 2 we used : @xmath142 , @xmath153@xmath132 , @xmath154@xmath136 and @xmath155@xmath136 .
the corresponding @xmath113 values for @xmath140 , @xmath138 and @xmath139 functions are : @xmath156 , @xmath157 , @xmath158 .
our main conclusion is that in the frame of a model based on quantum statistical principles and a space - time picture of an expanding source , a general description of the higher order correlations data with the same parameters , as those appearing in the first two correlation orders , is possible .
the assumption of a gaussian form for the density matrix is consistent with the data . on the other hand in @xcite and @xcite
the momentum - space approach was used , the currents were assumed to be real and no simultaneous fit of all correlation functions has been performed .
similar caveats apply to @xcite where an analysis of the same na22 data as those investigated in the present paper was presented and where only marginal " agreement with the simplified quantum optical model of @xcite was found .
the fact that in the present work we could fit the same data supports the space - time approach and the necessity of a simultaneous fit . besides the fact mentioned already that correlators in momentum - space are associated with a four dimensional radius " which has no clear physical interpretation , the space - time approach and the momentum - space approach differ also from a purely mathematical point of view . in @xcite
the @xmath10 dependence is postulated from the beginning while in the present paper , as in @xcite , it results from a complex process of integration . in @xcite
the same momentum - space model ( @xcite ) as that used in @xcite was applied and compared with ua1-collaboration data .
this time a new technique for estimating the correlation data was used but only second and third order correlations were considered . at first the best fit parameters set for the second order correlation data was established and then used to _ predict _ the third order correlation function , which was found to disagree with the measured one . given the insensitivity of of the fit parameters found by us on the @xmath140 function ,
this result is not surprising and the fitting procedures used in @xcite ( as well as in in @xcite ) have to be qualified . furthermore the general reservations expressed above about the momentum - space approach apply here again .
one may put the question why it is necessary to invoke higher order correlations to constrain the source parameters whereas the presence of a gaussian density matrix implies that the first and second moment of the current distribution , @xmath45 and @xmath159 , determine all higher order correlations .
of eq.([eq : phase ] ) enter in different combinations in the various correlation functions .
this is true both in the momentum - space approach @xcite , as well as in the space - time approach . in the application to data
@xcite and @xcite of the quantum optical approach this complication was circumvented " by assuming that the fields are real . ] in the applications of the space - time approach @xcite one has to take into account that due to the limited statistics , most experimental bec data including those analyzed in the present paper give the correlations in one single variable , @xmath10 . on the other hand , the correlation function @xmath140 , for instance , depends on 6 variables : the three momenta of the pions of the pair , @xmath160 and @xmath161 .
thus a large amount of information is lost due to the integration that projects out the @xmath10 dependence ( cf .
( 5 ) and ( 6 ) ) .
this is a fortiori true for higher order correlations .
therefore phenomenologically higher order correlation data can play an important role in constraining the source parameters . on the other hand
the fact that we were able to account for the data even within this simplified approach , proves that there are still many degrees - of - freedom not used at the present level of theory / data comparison so that the challenge of disproving the gaussian density matrix will remain a hard task for a long time ahead .
we are indebted to w. kittel for an instructive correspondence and to b. schlei , u. ornik and o.a.m .
helene for important discussions .
was partially supported by cnpq ( brazil ) and by fapesp ( 98/01446 - 0 ) .
* figure 1 : * the two- , three- and four - particle correlation functions in @xmath20 calculated using the space - time model defined in the text ( continous lines ) , for parameters @xmath125 ( see table 1 ) compared with experimental data from @xcite .
* figure 2 : * the same calculation as in figure 1 using the other set of parameters as defined in the text .
the figure shows that parameters which produce acceptable fits for @xmath3 do not necessarily give good overall fits .
99 v. p. kenney et al . , nucl .
b144 ( 1978 ) 312 . m. althof et al , tasso coll .
, z. phys .
c30 ( 1986 ) 355 .
i. juricic et al , mark ii coll . , phys .
rev . d39 ( 1989 ) 1 .
j. l. baily et al , na23 coll . , z. phys .
c43 ( 1989 ) 341 .
a. breakstone et al . , mod .
a6 ( 1991 ) 2785 . v. v. aivazyan et al .
, ehs / na22 coll .
, z. phys .
c51 ( 1991 ) 2785 .
t. akesson et al . , afs coll . , z. phys .
c36 ( 1987 ) 517 .
n. neumeister et al . , ua1 coll . , phys .
lett b275 ( 1992 ) 186 .
i. v. ajinenko et al . , ehs / na22 coll . , z. phys .
c61 ( 1994 ) 567 . n. m. agababyan et al . ,
ehs / na22 coll .
b332 ( 1994 ) 458 . n. m. agababyan et al .
, ehs / na22 coll .
, z. phys .
c68 ( 1995 ) 229 .
r.m.weiner , phys .
lett b232 ( 1989 ) 278 ; b242 ( 1990 ) 547 .
m. biyajima et al . , prog .
84 ( 1990 ) 931 .
p. carruthers , h. c. eggers and ina sarcevic , phys .
b254 ( 1991 ) 258 .
r. plmer , i. v. razumov and r. m. weiner , phys .
lett b286 ( 1992 ) 335 .
i. v. andreev , m. plmer , b. r. schlei and r. m. weiner , phys.lett .
b316 ( 1993 ) 583 ; phys .
d49 ( 1994 ) 1217 . i. v. andreev , m. plmer and r. m. weiner , int .
phys . a , vol . 8 , n. 26 , ( 1993)4577 .
m. adamus et al . , ehs / na22 collaboration , z. phys .
c39 ( 1988 ) 311
. n. m. agababyan et al . , ehs / na22 collaboration , z. phys .
c59 ( 1993 ) 195 .
h. c. eggers , b. buschbeck and p. lipa , phys rev .
79 ( 1997 ) 197 .
r.m.weiner , bose einstein correlations in particle and nuclear physics , john wiley & sons , 1997 . n. m. agababyan et al .
, ehs / na22 collaboration , z. phys .
c71 ( 1996 ) 405 . | a multiparticle system produced by a large number of independent sources is described by a gaussian density matrix @xmath0 . all theoretical approaches to bose - einstein correlations @xmath1 in high energy physics use this form for @xmath0 .
one of the most salient consequences of this form is the fact that all higher order ( @xmath2 ) moments of the current distribution can be expressed in terms of the first two .
we test this property by comparing the data on @xmath3 , @xmath4 and @xmath5 from @xmath6p and @xmath7p reactions at @xmath8 @xmath9 with the predictions of a general quantum statistical space - time approach .
even a simplified version of such an approach involving only 4 ( instead of the total of 10 ) independent parameters ( proper - time , correlation - length , transverse radius of chaotic source , chaoticity ) can account for the data .
previous attempts along these lines , which did not use the space - time approach , met with difficulties .
0.5 cm 0.3 cm 0.3 cm -1 cm 22.5 cm = 10000 to appear in physics letters b ( plb 13397 ) bose - einstein correlations ( bec ) are the basis of an experimental method for the determination of sizes and lifetimes of sources in particle and nuclear physics .
this knowledge is essential for an understanding of the dynamics of strong interactions .
a particularly important aspect of bec is represented by higher order correlations because of the prediction , which follows from the gaussian density matrix assumption , that all higher order moments of the current distribution can be described in terms of the first two .
the gaussian form ( in the coherent state representation ) of the density matrix is a fundamental assumption of bec and follows from the central limit theorem for a large number of independent sources , which are expected to act in a high energy reaction .
furthermore , higher order correlations provide important constraints on the space - time form of the sources , their dynamics ( expansion ) and chaoticity
. experimentally correlations of three and more particles have been studied in the last years in @xcite-@xcite and more recently attempts have been made to analyze these correlations in terms of simplified models @xcite-@xcite without clear space - time implications for the emitting source ( sources ) .
usually gaussian or exponential forms for the correlation functions in momentum space are postulated .
in contrast to this , the approach used in this work starts with the space - time characteristics of an expanding source and the space - time form of the correlators within the classical current formalism .
the dependence of @xmath1 on the four - momentum difference ( @xmath10 ) follows after explicit integration over space - time variables @xcite .
the aim of this investigation is to use the higher order correlations na22 data to test the validity of the gaussian density matrix assumption , within the space - time approach to bec . |
the important notion of entwining structures has been introduced by t. brzeziski and s. majid in @xcite . an entwining structure ( over a commutative ring @xmath0 ) consists of a @xmath0-algebra @xmath1 , a @xmath0-coalgebra @xmath2 and a certain @xmath0-homomorphism @xmath3 satisfying some axioms .
associated to @xmath4 there is the category @xmath5 of entwining modules whose objects are at the same time @xmath1-modules and @xmath2-comodules , with compatibility relation given by @xmath4 . the algebra @xmath1 can be identified with the monad @xmath6 whose eilenberg - moore category of algebras , @xmath7 , is ( isomorphic to ) the category of right @xmath1-modules .
similarly , @xmath2 can be identified with the comonad @xmath8 , and the corresponding eilenberg - moore category of coalgebras with the category of @xmath2-comodules .
it turns out that to give an entwining structure @xmath9 is to give a mixed distributive law @xmath10 from the monad @xmath11 to the comonad g in the sense of j. beck @xcite , which are in bijective corresondence with liftings ( or extensions ) @xmath12 of the comonad @xmath13 to the category @xmath7 ; or , equivalently , liftings @xmath14 of the monad @xmath11 to the category @xmath15 .
moreover , the categories @xmath5 , @xmath16 and @xmath17 are isomorphic .
thus , the ( mixed ) distributive law formalism can be used to study entwining structures and the corresponding category of modules . in this article -based on this formalism- we extend in the context of monoidal categories some of basic results on entwining structures that appear in the literature ( see , for example , @xcite , @xcite , @xcite ) .
the paper is organized as follows . after recalling the notion of beck s mixed distributive law and the basic facts about it ,
we define in section 3 an entwining structure in any monoidal category . in section 4 , we prove some categorical results that are needed in the next section , but may also be of independent interest . finally , in the last section we present our main results .
we refer to m. barr and c. wells @xcite , s. maclane @xcite and f. borceux @xcite for terminology and general results on ( co)monads , and to t. brzeziski and r. wisbauer @xcite for coring and comodule theory .
let @xmath18 be a monad and @xmath19 a comonad on a category @xmath20 .
a _ mixed distributive law _ from @xmath18 to @xmath19 is a natural transformation @xmath21 for which the diagrams @xmath22_{\eta g } \ar[rd]^{g \eta } & & & t g \ar[dl]_{\lambda } \ar[rd]^{t \varepsilon}&\\ t g \ar[rr]_{\lambda } & & g t \ , , & g t \ar[rr]_{\varepsilon t } & & t,}\ ] ] @xmath23_{\lambda } \ar[r]^{t\delta } & t g^2 \ar[r]^{\lambda g } & g t g \ar[d]^{g \lambda } & \text{and } & t^2 g \ar[d]_{\mu g } \ar[r]^{t \lambda } & t g t \ar[r]^{\lambda t } & g t t \ar[d]^{g \mu}\\ g t \ar[rr]_{\delta t } & & g g t & & tg \ar[rr]_{\lambda } & & g t}\ ] ] commute . given a monad @xmath18 on @xmath20 ,
write @xmath24 for the eilenberg - moore category of @xmath25-algebras , and write @xmath26 for the corresponding forgetful - free adjunction . dually , if @xmath19 is comonad on @xmath20 , then write @xmath27 for the category of @xmath28-coalgebras , and write @xmath29 for the corresponding forgetful - cofree adjunction . _
( _ see @xcite _ ) _ let @xmath18 be a monad and @xmath19 a comonad on a category @xmath20 .
then the following structures are in bijective correspondences : * mixed distributive laws @xmath30 ; * comonads @xmath31 on @xmath32 that extend @xmath28 in the sense that @xmath33 , @xmath34 and @xmath35 ; * monads @xmath36 on @xmath37 that extend @xmath25 in the sense that @xmath38 , @xmath39 and @xmath40 .
these correspondences are constructed as follows : * given a mixed distributive law @xmath41 then @xmath42 , @xmath43 , @xmath44 , for any @xmath45 ; and @xmath46 , @xmath47 , @xmath48 for any @xmath49 .
* if @xmath31 is a comonad on @xmath24 extending the comonad @xmath19 , then the corresponding distributive law @xmath50 is given by @xmath51 ^ -{tg \eta } & tgt = u^{t } f^{t } g u^{t } f^{t}=u^t f^t u^t \bar{g}f^t \ar[rr]^-{u^t \varepsilon ^t \bar{g}f^t } & & u^t \bar{g } f^t = gu^tf^t = gt,}\ ] ] where @xmath52 is the counit of the adjunction @xmath53 . * if @xmath36 is a monad on @xmath27 extending @xmath18 , then the corresponding mixed distributive law is given by @xmath54 ^ -{u_g \eta_g \bar{t } f_g } & & u_gf_gu_g \bar{t } f_g = u_g f_g t u_g f_g = gtg \ar[r]^-{gt \varepsilon } & gt},\ ] ] where @xmath55 is the unit of the adjunction @xmath56 it follows from this theorem that if @xmath57 is a mixed distributive law , then @xmath58 .
we write @xmath59 for this category .
an object of this category is a three - tuple @xmath60 , where @xmath61 , @xmath62 , for which @xmath63 . a morphism @xmath64 in @xmath65 is a morphism @xmath66 in @xmath20 such that @xmath67 and @xmath68 .
let @xmath69 be a monoidal category with coequalizers such that the tensor product preserves the coequalizer in both variables . then for all algebras @xmath70 and @xmath71 and all @xmath72 , @xmath73 and @xmath74 , the tensor product @xmath75 exists and the canonical morphism @xmath76 is an isomorphism . using maclane s coherence theorem ( see , @xcite , xi.5 )
, we may assume without loss of generality that @xmath77 is strict .
it is well known that every algebra @xmath78 in @xmath77 defines a monad @xmath79 on @xmath77 by * @xmath80 , * @xmath81 , * @xmath82 , and that @xmath83 is ( isomorphic to ) the category @xmath84 of right @xmath1-modules .
dually , if @xmath85 is a coalgebra ( = comonoid ) in @xmath77 , then one defines a comonad @xmath86 on @xmath77 by * @xmath87 , * @xmath88 * @xmath89 , and @xmath90 is ( isomorphic to ) the category @xmath91 of right @xmath2-comodules . quite obviously , if @xmath4 is a mixed distributive law from @xmath79 to @xmath86 , then the morphism @xmath92 makes the following diagrams commutative : @xmath93_{c \otimes e_a } \ar[dr]^{e_a \otimes c } & & & c \otimes a \ar[d]_{\lambda ' } \ar[rd]^-{\varepsilon_{\mathbb{c } } \otimes a}&\\ c \otimes a \ar[r]_{\lambda ' } & a \otimes c \ , , & & a \otimes c \ar[r]_{a \otimes \varepsilon_{\mathbb{c } } } & a , } \ ] ] @xmath94_-{\lambda ' } \ar[r]^-{\delta_c \otimes a } & c\otimes c \otimes a \ar[r]^{c \otimes \lambda ' } & c\otimes a \otimes c \ar[d]^ { \lambda ' \otimes c } & c\otimes a \otimes a \ar[d]_{c \otimes m_a } \ar[r]^{\lambda ' \otimes a } & a \otimes c \otimes a \ar[r]^{a \otimes \lambda ' } & a \otimes a \otimes c \ar[d]^{m_a \otimes c}\\ a \otimes c \ar[rr]_{a \otimes \delta_c } & & a \otimes c \otimes c , & c \otimes a \ar[rr]_{\lambda ' } & & a \otimes c \,\,.}\ ] ] conversely , if @xmath95 is a morphism for which the above diagrams commute , then the natural transformation @xmath96 is a mixed distributive law from the monad @xmath79 to the comonad @xmath86 .
it is easy to see that @xmath97 .
when @xmath98 is a regular generator in @xmath77 and the tensor product preserves all colimits in both variables , it is not hard to show that @xmath99 . when this is the case , then the correspondences @xmath100 and @xmath101 are inverses of each other .
an entwining structure @xmath102 consists of an algebra @xmath78 and a coalgebra @xmath103 in @xmath77 and a morphism @xmath104 such that the natural transformation @xmath105 is a mixed distributive law from the monad @xmath79 to the comonad @xmath86 .
let be @xmath106 be an entwining structure and let @xmath31 be the comonad on @xmath84 that extends @xmath107 .
then we know that , for any @xmath108 , @xmath109 ^ -{v \otimes \lambda } & v \otimes a \otimes c \ar[r]^-{\xi_v \otimes c } & v \otimes c ) .}\ ] ] in particular , since @xmath110 , @xmath111 is a right @xmath1-module with right action @xmath112
^ -{a \otimes \lambda } & a \otimes a \otimes c \ar[r]^-{m_a \otimes c } & a \otimes c.}\ ] ] view @xmath111 as a left @xmath1-module through @xmath113 .
then @xmath114 is an @xmath1-@xmath1-bimodule .
clearly @xmath115 moreover , since @xmath116 , it follows from the associativity of @xmath117 that the diagram @xmath118 ^ -{a \otimes a \otimes \lambda } \ar[d]_{m_a \otimes c \otimes a } & a \otimes a \otimes a \otimes c \ar[d]^{a \otimes m_a \otimes c } \\ a \otimes c \otimes a \ar[d]_{a \otimes \lambda } & a \otimes a \otimes c \ar[d]^{m_a \otimes c } \\ a \otimes a \otimes c \ar[r]_{m_a \otimes c } & a \otimes c } \ ] ] is commutative , which just means that @xmath119 is an @xmath1-@xmath1-bimodule .
since @xmath120 and @xmath121 are morphisms of right @xmath1-modules , and since @xmath122 @xmath123 and @xmath124@xmath125 , it follows that @xmath126 and @xmath127 are both morphisms of right @xmath1-modules .
clearly they are also morphisms of left @xmath1-modules with the obvious left @xmath1-module structures arising from the multiplication @xmath128 , and hence morphisms of @xmath1-@xmath1-bimodules .
since @xmath129 is a coalgebra in @xmath77 , it follows that the triple @xmath130 , where @xmath131 @xmath132 and @xmath133 @xmath127 , is an @xmath1-coring . since , for any @xmath134 @xmath135 the comonad @xmath136 is isomorphic to the comonad @xmath137 .
thus , any entwining structure @xmath106 defines a right @xmath1-module structure @xmath138 on @xmath111 such that @xmath139 is an @xmath1-@xmath1-bimodule and the triple @xmath140 is an @xmath1-coring .
moreover , when this is the case , the comonad @xmath137 on @xmath84 extends the comonad @xmath86 .
it follows that @xmath141 conversely , let @xmath78 be an algebra and @xmath142 a coalgebra in @xmath77 , and suppose that @xmath143 has the structure @xmath144 of a right @xmath1-module such that the triple @xmath145 ^ -{a \otimes \varepsilon_{\mathbb{c } } } & a , \,\ , a \otimes c \ar[r]^-{a \otimes \delta_{\mathbb{c } } } & a \otimes c \otimes c)}\ ] ] is an @xmath1-coring .
then it is easy to see that the comonad @xmath146 on @xmath84 extends the comonad @xmath86 on @xmath77 , and thus defines an entwining structure @xmath147 .
summarising , we have let @xmath78 be an algebra and @xmath142 a coalgebra in @xmath77 .
then there exists a bijection between right @xmath1-module structures @xmath144 making @xmath148 an @xmath1-bimodule for which the triple ( 1 ) is an @xmath1-coring and entwining structures @xmath149 , given by : @xmath150 & ( \lambda_{\underline{a \otimes c } } : c \otimes a \ar[rr]^-{e_a \otimes c \otimes a } & & a \otimes c \otimes a \ar[r]^-{\xi_{a \otimes c } } & a \otimes c)}\ ] ] with inverse given by @xmath151 & ( \xi_{a \otimes c } : a \otimes c \otimes a \ar[r]^-{a \otimes \lambda } & a \otimes a \otimes c \ar[rr]^-{m_a \otimes c } & & a \otimes c)}\ ] ] under this equivalence @xmath152 .
let @xmath19 be a comonad on a category @xmath20 , and let @xmath153 be the forgetful functor . fix a functor @xmath154 , and consider a functor @xmath155 making the diagram @xmath156 ^ -{\bar{f } } \ar[dr]_{f } & & a_{{\textbf{g } } } \ar[dl]^{u_{\textbf{g}}}\\ & { { \mathcal { a } } } & } \ ] ] commutative .
then @xmath157 for some @xmath158 .
consider the natural transformation @xmath159 whose @xmath160-component is @xmath161 .
it is proved in @xcite that : suppose that @xmath162 has a right adjoint @xmath163 with unit @xmath164 and counit @xmath165 .
then the composite @xmath166 ^ -{\bar{\alpha}_{_f } u } & gfu \ar[r]^{g \varepsilon } & g.}\ ] ] is a morphism from the comonad @xmath167 generated by the adjunction @xmath168 to the comonad @xmath28 .
moreover , the assignment @xmath169 yields a one to one correspondence between functors @xmath170 making the diagram ( 2 ) commutative and morphisms of comonads @xmath171 .
write @xmath172 for the composite @xmath173 ^ -{\eta u } & ufu \ar[r]^-{u t_{\bar{f } } } & ug}.$ ] the equalizer @xmath174 , if it exists , of the following diagram @xmath175 ^ -{uu_g \eta_g } \ar@ { ->}@<-0.5ex > [ rr]_-{\beta_u u_g } & & ugu_g = uu_gf_gu_g,}\ ] ] where @xmath55 is the unit of the adjunction @xmath176 is right adjoint to @xmath177 .
see @xcite or @xcite .
let @xmath178 be a functor making ( 2 ) commutative and let @xmath179 be the corresponding morphism of comonads . consider the following composition @xmath180^{k_{\textbf{g } ' } } & { { \mathcal { a}}}_{\textbf{g } ' } \ar[r]^{{{\mathcal { a}}}_{t_{\bar{f } } } } & { { \mathcal { a}}}_{\textbf{g}},}\ ] ] where * @xmath181 is the eilenberg - moore comparison functor for the comonad @xmath182 .
* @xmath183 is the functor @xmath184 induced by the morphism of comonads @xmath185 the diagram @xmath186^{k_{g ' } } \ar[dr]_{{\bar{f } } } & { { \mathcal { a}}}_{g ' } \ar[d]^{{{\mathcal { a}}}_{{t_{\bar{f}}}}}\\ & { { \mathcal { a}}}_{g}}\ ] ] is commutative .
let @xmath187 .
then @xmath188 and @xmath189 since @xmath190 is the composite @xmath191 ^ -{(\bar{\alpha}_f)_{uf(b ) } } & & gfuf(b ) \ar[r]^-{g \varepsilon _ { f(b ) } } & gf(b),}\ ] ] and since by naturality of @xmath192 , the diagram @xmath193^{(\bar{\alpha})_b } \ar[d]_{f(\eta_b ) } & gf(b ) \ar[d]^{gf(\eta_b)}\\ fuf(b ) \ar[r]_{(\bar{\alpha})_{uf(b ) } } & gfuf(b)}\ ] ] commutes , we have @xmath194 thus @xmath195 which just means that @xmath196 .
we are now ready to prove the following let @xmath28 be a comonad on a category @xmath20 , @xmath197 an adjunction and @xmath198 a functor with @xmath199 .
then the following are equivalent : * the functor @xmath200 is an equivalence .
* the functor @xmath162 is comonadic and the morphism of comonads @xmath201 is an isomorphism .
suppose that @xmath200 is an equivalence of categories .
then @xmath162 is isomorphic to the comonadic functor @xmath202 and thus is comonadic .
hence the comparison functor @xmath203 is an equivalence and it follows from the commutative diagram ( 4 ) that @xmath204 is also an equivalence , and since the diagram @xmath205^{{{\mathcal { a}}}_{t_{\bar{f } } } } \ar[rd]_{u_{g ' } } & & { { \mathcal { a}}}_g \ar[dl]^{u_g}\\ & { { \mathcal { a}}}&}\ ] ] is commutative , @xmath206 is an isomorphism of comonads .
so @xmath207 .
suppose now that @xmath208 is an isomorphism of comonads and @xmath162 is comonadic .
then * @xmath209 is an equivalence , since @xmath162 is comonadic .
* @xmath210 is an equivalence , since @xmath206 is an isomorphism . and
it now follows from the commutative diagram ( 4 ) that @xmath200 is also an equivalence .
thus @xmath211 .
this completes the proof of the theorem . in @xcite ,
j. gmez - torrecillas has proved that @xmath200 is an equivalence of categories iff @xmath206 is an isomorphism of comonads , @xmath162 is conservative , and for any @xmath212 , @xmath162 preserves the equalizer of the pair of parallel morphisms @xmath213^{u(x ) } \ar[r]_-{\eta_{u(x)}}&ug'(x ) \ar[rr]_{u((t_{\bar{f}})_x ) } & & ug(x ) \,.}\ ] ] when @xmath206 is an isomorphism of comonads , to say that @xmath162 preserves the equalizer of the pair of morphisms @xmath214 is to say that @xmath162 preserves the equalizer of the pair of morphisms @xmath215 ^ -{\eta_{u(x ) } } \ar@ { ->}@<-0.5ex > [ rr]_-{u((t^{-1}_{\bar{f}})_x ) \cdot u(x ) } & & ug'(x),}\ ] ] which we can rewrite as @xmath216 ^ -{\eta_{u(x ) } } \ar@ { ->}@<-0.5ex > [ rr]_-{u((t^{-1}_{\bar{f}})_x \,\cdot \,x ) } & & ug'(x)=ufu(x).}\ ] ] since @xmath206 is an isomorphism of comonds , @xmath204 is an equivalence of categories , and thus each object @xmath217 is isomorphic to the @xmath182-coalgebra @xmath218 , where @xmath219 .
it follows that when @xmath206 is an isomorphism of comonds , to say that @xmath162 preserves the equalizer of @xmath214 for each @xmath212 is to say that @xmath162 preserves the equalizer of @xmath220 for each @xmath221 .
thus , when @xmath206 is an isomorphism of comonds , @xmath200 is an equivalence of categories iff @xmath162 is conservative and preserves the equalizer of @xmath220 for each @xmath222 , which according to ( the dual of ) beck s theorem ( see @xcite ) , is to say that the functor @xmath162 is comonadic .
hence our theorem 4.4 is equivalent to theorem 1.7 of @xcite .
let @xmath223 be an entwining structure in a monoidal category @xmath69 , and let @xmath224 be a group - like element of @xmath225 .
( recall that a morphism @xmath224 is said to be a group - like element of @xmath225 if the following diagrams @xmath226^{g } \ar@{=}[dr]^{(1 ) } & c \ar[d]^{\varepsilon_{\mathbb{c } } } & & i \ar@{}[dr]^{(2)}\ar[r]^{g } \ar[dr]_{g\otimes g } & c \ar[d]^{\delta_{\mathbb{c}}}\\ & i & & & c \otimes c}\ ] ] are commutative . )
consider the diagram @xmath229 \ar[r]^-{g \otimes a } & c \otimes a \ar[r]^{\lambda } \ar[d]^{\varepsilon_{\mathbb{c } } \otimes a } & a \otimes c \ar[d]^{a \otimes \varepsilon_{\mathbb{c}}}\\ & a \ar@{=}[r ] & a\ , .}\ ] ] the triangle is commutative by ( 1 ) of the definition of @xmath230 and the square is commutative by the definition of @xmath4 ( see the second commutative diagram in the definition of entwining structures ) .
@xmath231_{g \otimes a } \ar[r]^{g \otimes a } & c \otimes a \ar[r]^{\lambda } & a \otimes c \ar[dd]^-{a \otimes \delta_{\mathbb{c}}}\\ c \otimes a \ar[d]_{\lambda}\\ a \otimes c \ar[r]_-{g \otimes a \otimes c } & c \otimes a \otimes c \ar[r]_{\lambda\otimes c } & a \otimes c \otimes c } \ ] ] is also commutative , which it is since @xmath232 by the definition of @xmath4 and since the diagram ( 2 ) of definition of group - like elements is commutative .
consider the diagram @xmath238^{i_a } & a\ar@/^/@<+1ex>[rr]^{a \otimes g } \ar[r]_{g \otimes a } & c \otimes a \ar[r]_{\lambda } & a \otimes c \\
i \ar@{.>}[u]^{e_{a^{\mathbb{c}}}}\ar[ru]_{e_a}&&}\ ] ] since @xmath239 is a natural transformation , the diagram @xmath240_{e_a}\ar[r]^{g } & c \ar[d]^{c \otimes e_a}\\ a \ar[r]_{g \otimes a } & c \otimes a}\ ] ] is commutative . similarly , since @xmath241 is a natural transformation , the following diagram is also commutative : @xmath242 ^ -{e_a } \ar[d]_{g } & a \ar[d]^{a \otimes g}\\ c
\ar[r]_-{e_a \otimes c } & a \otimes c\,.}\ ] ] now we have : @xmath243@xmath244 thus there exists a unique morphism @xmath245 for which @xmath246 * the diagram @xmath247_{m_a}\ar[r]^-{g \otimes a \otimes a } & c \otimes a \otimes a \ar[d]^{c \otimes m_a}\\ a \ar[r]_{g \otimes a } & c \otimes a}\ ] ] is commutative by naturality of @xmath248 ; * @xmath249 by the definition of @xmath4 ; * @xmath250 , since @xmath251 is an equalizer of @xmath252 and @xmath253 ; * the diagram @xmath247_{m_a}\ar[r]^-{a \otimes a \otimes g } & a \otimes a \otimes c \ar[d]^{m_a
\otimes c}\\ a \ar[r]_{a \otimes g } & a \otimes c}\ ] ] is commutative by naturality of @xmath254 , we have @xmath255@xmath256@xmath257@xmath258 thus the morphism @xmath259 equalizes the morphisms @xmath260 and @xmath253 , and hence there is a unique morphism @xmath261 such that the diagram @xmath262_{m_{a^{\mathbb{c } } } } \ar[r]^-{i_a \otimes i_a } & a \otimes a \ar[d]^{m_a } \\ a^{\mathbb{c } } \ar[r]_{i_a } & a}\ ] ] commutes .
it is now straightforward to show that the triple @xmath263 is an algebra in @xmath77 ; moreover , the triangle of the diagram @xmath264 and the diagram @xmath265 show that @xmath251 is an algebra morphism .
@xmath271_{m_a } \ar[r]^-{g \otimes a \otimes a } & c \otimes a \otimes a \ar[r]^-{\lambda \otimes a } \ar@{.>}[d]^{c \otimes m_a } & a \otimes c \otimes a \ar[r]^{a \otimes \lambda } & a \otimes a \otimes c \ar[d]^{m_a \otimes c}\\ a \ar[r]_{g \otimes a } & c \otimes a \ar[rr]_{\lambda } & & a \otimes c.}\ ] ] but this diagram is commutative , since the algebra morphism @xmath237 makes @xmath1 an @xmath273-@xmath273-bimodule and thus induces the _ extension - of - scalars _
functor @xmath274@xmath275 and the forgetful functor @xmath276@xmath277 which is right adjoint to @xmath278 .
the corresponding comonad on @xmath279 makes @xmath280 into an @xmath1-coring with the following counit and comultiplication : @xmath281 ^ -{q } & a \otimes a \ar[r]^-{m_a } & a,}\ ] ] ( where @xmath282 is the canonical morphism ) and @xmath283 ^ -{a \otimes_{a^\mathbb{c } } i_a \otimes_{a^\mathbb{c } } a } & & a \otimes_{a^\mathbb{c } } a \otimes_{a^\mathbb{c } } a = ( a \otimes_{a^\mathbb{c } } a)_a \otimes ( a \otimes_{a^\mathbb{c } } a).}\ ] ] we write @xmath284 for this @xmath1-coring .
clearly @xmath288 and @xmath289 .
moreover , by ( 9 ) , the following diagram @xmath290 ^ -{x \otimes_{a^\mathbb{c } } g_a \otimes a } \ar[d]_{x \otimes_{a^\mathbb{c } } m_a } & & x \otimes_{a^\mathbb{c } } a \otimes c \otimes a \ar[rr]^-{x \otimes_{a^\mathbb{c } } a \otimes \lambda } & & x \otimes_{a^\mathbb{c } } a \otimes a \otimes c \ar[d]^{x \otimes_{a^\mathbb{c}}m_a \otimes
c}\\ x \otimes_{a^\mathbb{c } } a \ar[rrrr]_{x \otimes_{a^\mathbb{c } } g_a } & & & & x \otimes_{a^\mathbb{c } } a \otimes c}\ ] ] is commutative .
thus , @xmath291 the lemma shows that the assignment @xmath292 yields a functor @xmath293 it is clear that @xmath294 , where @xmath295 is the underlying functor .
it now follows from theorem 3.1 that the composite @xmath296^{a \otimes g_a } & & a \otimes a \otimes c \ar[rr]^-{m_a \otimes c } & & a \otimes c}\ ] ] is a morphism of @xmath1-corings @xmath297 we write @xmath298 for this morphism .
we say that @xmath1 is @xmath299-galois if @xmath298 is an isomorphism of @xmath1-corings . applying theorem 4.4 the commutative diagram @xmath300 ^ -{\bar{f } } \ar[dr]_{f_{i_a}= - \otimes_{a ^{\mathbb{c } } } a } & {
{ { \mathcal { v}}}_{\mathbb{a}}}^{(\underline{a \otimes c})_{\lambda } } \ar[d]^{u_{(\underline{a \otimes c})_{\lambda}}}\\ & { { \mathcal { v}}}_{\mathbb{a}}}\ ] ] we get : let @xmath301 be an entwining structure , and let @xmath224 be a group - like element of @xmath225 . then the functor @xmath302 is an equivalence if and only if @xmath1 is @xmath303-galois and the functor @xmath162 is comonadic .
* _ flat _ , if the functor @xmath308 ( resp .
@xmath309 ) preserves equalizers ; * _ faithfully flat _ , if the functor @xmath310 ( resp .
@xmath311 ) is conservative and flat ( equivalently , preserves and reflects equalizers ) ; since any left adjoint functor that is conservative and preserves equalizers is comonadic by a simple and well - known application ( of the dual of ) beck s theorem , one direction is clear from therem 5.5 ; so suppose that @xmath200 is an equivalence of categories .
then , by theorem 4.5 , @xmath1 is @xmath303-galois and the functor @xmath278 is comonadic . since any comonadic functor is conservative , @xmath278 is also conservative .
thus , it only remains to show that @xmath313 is flat .
since @xmath2 is flat by our assumption , @xmath314 is also flat .
it follows that the underlying functor of the comonad @xmath137 on @xmath279 preserves equalizers .
we recall ( for example , from @xcite ) that if @xmath315 is a comonad on a category @xmath316 , and if @xmath316 has some type of limits preserved by @xmath13 , then the category @xmath317 has the same type of limits and these are preserved by the underlying functor @xmath318 .
thus the functor @xmath319 preserves equalizers , and since @xmath200 is an equivalence of categories , the functor @xmath320 also preserves equalizers , which just means that @xmath313 is flat .
this completes the proof . from now on we suppose at all times that our @xmath77 is a strict braided monoidal category with braiding @xmath321 .
then the tensor product of two ( co)algebras in @xmath77 is again a ( co)algebra ; the multiplication @xmath322 and the unit @xmath323 of the tensor product of two algebras @xmath70 and @xmath324 are given through @xmath325 and @xmath326 a bialgebra @xmath327 in @xmath77 is an algebra @xmath328 and a coalgebra @xmath329 , where @xmath330 and @xmath331 are algebra morphisms , or , equivalently , @xmath332 and @xmath333 are coalgebra morphisms .
recall that for any bialgebra @xmath335 , the category @xmath338 is monoidal : the tensor product @xmath339 of two right @xmath335-comodules @xmath340 and @xmath341 is their tensor product @xmath342 in @xmath77 with the coaction @xmath343
^ -{\delta_x \otimes \delta_y } & x \otimes h \otimes y \otimes h \ar[rr]^-{x \otimes \sigma_{x , y } \otimes y } & & x \otimes y \otimes h \otimes h \ar[rr]^-{x \otimes y \otimes m_h } & & x \otimes y \otimes h}.\ ] ] the unit object for this tensor product is @xmath98 with trivial @xmath344-comodule structure @xmath345 * @xmath347 is an algebra in the monoidal category @xmath348 ; * @xmath347 is an @xmath349-comodule algebra ; that is , @xmath1 is a right @xmath349-comodule and the @xmath349-comodule coaction @xmath350 is a morphism of algebras in @xmath77 from the algebra @xmath70 to the algebra @xmath351 .
suppose now that @xmath347 is a right @xmath349-comodule algebra with @xmath349-coaction @xmath352 . by the previous proposition
, @xmath1 is an algebra in the monoidal category @xmath348 , and thus defines a monad @xmath353 on @xmath348 as follows : it is easy to see that the monad @xmath357 extends the monad @xmath358 ; and it follows from theorem 2.1 that there exists a distributive law @xmath359 from the monad @xmath360 to the comonad @xmath361 , and hence an entwining structure @xmath362 , where @xmath363 .
let @xmath364 be a right @xmath335-comodule algebra .
then the entwining structure @xmath366 is given by the composite : @xmath367 ^ -{h \otimes \alpha_a } & h \otimes a \otimes h \ar[r]^-{\sigma_{h , a}\otimes h } & a \otimes h \otimes h \ar[r]^-{a \otimes m_h } & a \otimes h}.\ ] ] since @xmath368 the pair @xmath369 , where @xmath370 is the composite @xmath371 ^
-{\delta_h \otimes \alpha_a } & h \otimes h \otimes h \otimes a \ar[rr]^-{h \otimes \sigma_{h , a}\otimes h } & & h \otimes a \otimes h \otimes h \ar[rr]^-{h \otimes a \otimes m_h } & & h \otimes a \otimes h\ , , } \ ] ] is also an object of @xmath338 , and it follows from theorem 1.1 that @xmath372 is the composite @xmath373 ^ -{\delta_{a \otimes h } } & h \otimes a \otimes h \ar[rr]^-{\varepsilon_h \otimes a \otimes h } & & a \otimes h .}\ ] ] consider now the following diagram @xmath374 ^ -{\delta_h \otimes a \otimes h } \ar@{=}[ddrr ] & & { h \otimes h } { \otimes a \otimes h } \ar[rr]^-{h \otimes \sigma_{h , a } \otimes h } \ar[dd]_{\varepsilon_h \otimes h \otimes a \otimes h } & & h \otimes a \otimes h \otimes h \ar[rr]^-{h \otimes a \otimes m_h } \ar[dd]_{\varepsilon_h \otimes a \otimes h \otimes h } & & h \otimes a \otimes h \ar[dd]_{\varepsilon_h \otimes a \otimes h}\\\\ h \otimes a \ar[uu]^-{h \otimes \alpha_a } & & h \otimes a \otimes h \ar[rr]_-{\sigma_{h , a}\otimes h } & & a \otimes h \otimes h \ar[rr]_{a \otimes m_h } & & a \otimes h \,.}\ ] ] since in this diagram let @xmath346 be a bialgebra in @xmath77 , and let @xmath380 be a right @xmath335-comodule algebra . then the right @xmath344-comodule structure on @xmath1 corresponding to the group - like element @xmath381 as in proposition 4.1 coincides with @xmath382 .
let @xmath346 be a bialgebra in @xmath77 , let @xmath380 be a right @xmath335-comodule algebra , and let @xmath396 be the corresponding entwining structure .
then the functor @xmath397 @xmath398 is an equivalence of categories iff the extension - of - scalars functor @xmath399 @xmath400 is comonadic and @xmath1 is @xmath344-galois ( in the sense that the canonical morphism @xmath401 is an isomorphism ) .
let @xmath346 be a bialgebra in @xmath77 , let @xmath380 be a right @xmath335-comodule algebra , and let @xmath396 be the corresponding entwining structure .
suppose that @xmath349 is flat .
then the following are equivalent : let @xmath327 be a bialgebra in @xmath77 , and let @xmath380 be a right @xmath335-comodule algebra .
a right @xmath404-module is a right @xmath1-module which is a right @xmath344-comodule such that the @xmath344-comodule structure morphism is a morphism of right @xmath1-modules .
morphisms of right @xmath405-modules are right @xmath1-module right @xmath344-comodule morphisms .
we write @xmath406 for this category .
note that the category @xmath406 is the category @xmath407 of right @xmath408-modules in the monoidal category @xmath338 , and it follows from theorem 2.1 that let @xmath410 be a bialgebra in @xmath77 , and let @xmath380 be a right @xmath335-comodule algebra .
then the functor @xmath411 is an equivalence of categories iff the extension - of - scalars functor @xmath399 is comonadic and @xmath1 is @xmath344-galois .
we will show that the composite @xmath415 ^
-{h \otimes \delta_h } & h \otimes h \otimes h \ar[r]^-{h \otimes s \otimes h } & h \otimes h \otimes h \ar[r]^-{m_h \otimes h } & h \otimes h } \ ] ] is the inverse for @xmath416 . indeed , consider the diagram @xmath417^{(1 ) } \ar[dd]_-{h \otimes \delta_h } \ar[rr]^-{h \otimes \delta_h } & & h \otimes h \otimes h \ar@{}[rrdd]^{(2 ) } \ar[rr]^-{m_h \otimes h } \ar[dd]_-{h \otimes h \otimes \delta_h } & & h \otimes h \ar[dd]^{h \otimes \delta_h}\\\\ h \otimes h \otimes h \ar[rr]^-{h \otimes \delta_h \otimes h } & & h \otimes h \otimes h \otimes h \ar@{}[rrdd]^{(3)}\ar[dd]_{h \otimes h \otimes s \otimes h } \ar[rr]^{m_h \otimes h \otimes h } & & h \otimes h \otimes h \ar[dd]^{h \otimes s \otimes h}\\\\ & & h \otimes h \otimes h \otimes h \ar@{}[rrdd]^{(4 ) } \ar[dd]_{h \otimes m_h \otimes h } \ar[rr]^-{m_h \otimes h \otimes h } & & h \otimes h \otimes h \ar[dd]^{m_h \otimes h}\\\\ & & h \otimes h \otimes h \ar[rr]_{m_h \otimes h } & & h \otimes h\ , .}\ ] ] we have : * square ( 1 ) commutes because of coassociativity of @xmath331 ; * square ( 2 ) commutes because of naturality of @xmath418 ; * square ( 3 ) commutes because @xmath376 is a bifunctor ; * square ( 4 ) commutes because of associativity of @xmath333 . we will first show that the diagram @xmath428\ar@{->}@<0.5ex>[rr]^-{h \otimes e_h } \ar@ { ->}@<-0.5ex > [ rr]_-{e_h \otimes h } & & h \otimes h \ar[d]^{x}\\ h \ar@{->}@<0.5ex>[rr]^-{\delta_h } \ar@
{ ->}@<-0.5ex > [ rr]_-{e_h \otimes h } & & h \otimes h}\ ] ] is serially commutative .
indeed , we have : @xmath429@xmath430@xmath431 @xmath432@xmath433 thus , @xmath434 is isomorphic to the equalizer of the pair @xmath435 .
but since @xmath436 is a split monomorphism in @xmath77 , the diagram @xmath226 ^ -{e_h } & h \ar@{->}@<0.5ex>[rr]^-{h \otimes e_h } \ar@ { ->}@<-0.5ex > [ rr]_-{e_h \otimes h } & & h \otimes h}\ ] ] is an equalizer diagram . hence @xmath437 it follows from propositions 5.16 and 5.17 that @xmath349 is @xmath344-galois , and according to theorem 5.12 , the functor @xmath441 is an equivalence iff the functor @xmath442 is comonadic .
but since the morphism @xmath378 is a split monomorphism in @xmath77 , the unit of the adjunction @xmath443 is a split monomorphism , and it follows from 3.16 of @xcite that @xmath444 is comonadic .
this completes the proof . | interpreting entwining structures as special instances of j. beck s distributive law , the concept of entwining module can be generalized for the setting of arbitrary monoidal category . in this paper
, we use the distributive law formalism to extend in this setting basic properties of entwining modules . |
the minimum spanning tree ( mst ) problem is one of the first and foundational problems in the field of combinatorial optimisation . in its initial formulation by borvka @xcite ,
one is given distinct , positive edge weights ( or _ lengths _ ) for @xmath7 , the complete graph on vertices labelled by the elements of @xmath8 .
writing @xmath9 for this collection of edge weights , one then seeks the unique connected subgraph @xmath10 of @xmath7 with vertex set @xmath11 that minimizes the _ total length _ @xmath12 algorithmically , the mst problem is among the easiest in combinatorial optimisation : procedures for building the mst are both easily described and provably efficient .
the most widely known mst algorithms are commonly called _
kruskal s algorithm _ and _ prim s algorithm_. both procedures are important in this work ; as their descriptions are short , we provide them immediately .
kruskal s algorithm : start from a forest of @xmath0 isolated vertices @xmath8 . at each step , add the unique edge of smallest weight joining two distinct components of the current forest .
stop when all vertices are connected .
prim s algorithm : fix a starting vertex @xmath13 . at each step , consider all edges joining the component currently containing @xmath13 with its complement , and from among these add the unique edge of smallest weight . stop when all vertices are connected .
unfortunately , efficient procedures for constructing mst s do not automatically yield efficient methods for understanding the typical structure of the resulting objects . to address this ,
a common approach in combinatorial optimisation is to study a procedure by examining how it behaves when given random input ; this is often called _ average case _ or _ probabilistic analysis .
_ the probabilistic analysis of mst s dates back at least as far as beardwood , halton and hammersley @xcite who studied the _
euclidean mst _ of @xmath0 points in @xmath14 .
suppose that @xmath15 is an absolutely continuous measure on @xmath14 with bounded support , and let @xmath16 be i.i.d .
samples from @xmath15 . for edge @xmath17 , take @xmath18 to be the euclidean distance between @xmath19 and @xmath20 .
then there exists a constant @xmath21 such that if @xmath22 is the total length of the minimum spanning tree , then @xmath23 this _ law of large numbers for euclidean mst s _ is the jumping - off point for a massive amount of research : on more general laws of large numbers @xcite , on central limit theorems ( @xcite , and on the large-@xmath0 scaling of various other `` localizable '' functionals of random euclidean mst s ( @xcite . ( the above references are representative , rather than exhaustive .
the books of penrose @xcite and of yukich @xcite are comprehensive compendia of the known results and techniques for such problems . ) from the perspective of borvka s original formulation , the most natural probabilistic model for the mst problem may be the following .
weight the edges of the complete graph @xmath7 with independent and identically distributed ( i.i.d . ) random edge weights @xmath24 whose common distribution @xmath15 is atomless and has support contained in @xmath25 , and let @xmath26 be the resulting random mst .
the conditions on @xmath15 ensure that all edge weights are positive and distinct .
frieze @xcite showed that if the common distribution function @xmath27 is differentiable at @xmath28 and @xmath29 , then the total weight @xmath22 satisfies @xmath30 } } \to \zeta(3),\ ] ] whenever the edge weights have finite mean .
it is also known that @xmath31 without any moment assumptions for the edge weights @xcite .
results analogous to ( [ eq : knlln ] ) have been established for other graphs , including the hypercube @xcite , high - degree expanders and graphs of large girth @xcite , and others @xcite .
returning to the complete graph @xmath7 , aldous @xcite proved a distributional convergence result corresponding to ( [ eq : knlln ] ) in a very general setting where the edge weight distribution is allowed to vary with @xmath0 , extending earlier , related results @xcite .
janson @xcite showed that for i.i.d .
uniform@xmath32 $ ] edge weights on the complete graph , @xmath33 is asymptotically normally distributed , and gave an expression for the variance that was later shown @xcite to equal @xmath34 .
if one is interested in the _ graph theoretic _ structure of the tree @xmath26 rather than in information about its edge weights , the choice of distribution @xmath15 is irrelevant . to see this ,
observe that the behaviour of both kruskal s algorithm and prim s algorithm is fully determined once we order the edges in increasing order of weight , and for any distribution @xmath15 as above , ordering the edges by weight yields a uniformly random permutation of the edges .
we are thus free to choose whichever distribution @xmath15 is most convenient , or simply to choose a uniformly random permutation of the edges .
taking @xmath15 to be uniform on @xmath32 $ ] yields a particularly fruitful connection to the now - classical _ erds rnyi random graph process_. this connection has proved fundamental to the detailed understanding of the global structure of @xmath26 and is at the heart of the present paper , so we now explain it .
let the edge weights @xmath24 be i.i.d .
uniform@xmath32 $ ] random variables .
the erds rnyi graph process @xmath35 is the increasing graph process obtained by letting @xmath36 have vertices @xmath8 and edges @xmath37 . , and we note that the definition of @xmath36 still makes sense in this case . ] for fixed @xmath38 , each edge of @xmath7 is independently present with probability @xmath38 . observing the process
as @xmath38 increases from zero to one , the edges of @xmath7 are added one at a time in exchangeable random order .
this provides a natural coupling with the behaviour of kruskal s algorithm for the same weights , in which edges are _ considered _ one at a time in exchangeable random order , and added precisely if they join two distinct components .
more precisely , for @xmath39 write @xmath40 for the subgraph of the mst @xmath26 with edge set @xmath41 .
then for every @xmath42 , the connected components of @xmath40 and of @xmath36 have the same vertex sets . in their foundational paper on the subject @xcite , erds and rnyi
described the _ percolation phase transition _ for their eponymous graph process .
they showed that for @xmath43 with @xmath44 fixed , if @xmath45 ( the _ subcritical _ case ) then @xmath36 has largest component of size @xmath46 in probability , whereas if @xmath47 ( the _ supercritical _ case ) then the largest component of @xmath36 has size @xmath48 , where @xmath49 is the survival probability of a poisson@xmath50 branching process .
they also showed that for @xmath51 , all components aside from the largest have size @xmath46 in probability . in view of the above coupling between the graph process and kruskal s algorithm
, the results of the preceding paragraph strongly suggest that `` most of '' the global structure of the mst @xmath26 should already be present in the largest component of @xmath52 , for any @xmath47 . in order to understand @xmath26 , then
, a natural approach is to delve into the structure of the forest @xmath40 for @xmath53 ( the _ near - critical _ regime ) and , additionally , to study how the components of this forest attach to one another as @xmath38 increases through the near - critical regime . in this paper
, we use such a strategy to show that after suitable rescaling of distances and of mass , the tree @xmath26 , viewed as a measured metric space , converges in distribution to a random compact measured metric space @xmath54 of total mass measure one , which is a _ random real tree _ in the sense of @xcite .
the space @xmath54 is the scaling limit of the minimum spanning tree on the complete graph .
it is binary and its mass measure is concentrated on the leaves of @xmath54 . the space @xmath54 shares all these features with the first and most famous random real tree , the brownian continuum random tree , or crt @xcite
. however , @xmath54 is not the crt ; we rule out this possibility by showing that @xmath54 almost surely has minkowski dimension 3 .
since the crt has both minkowski dimension 2 and hausdorff dimension 2 , this shows that the law of @xmath54 is mutually singular with that of the crt , or any rescaled version thereof .
the remainder of the introduction is structured as follows .
first , section [ sec : overview ] , below , we provide the precise statement of our results .
second , in section [ sec : proof_idea ] we provide an overview of our proof techniques .
finally , in section [ sec : history ] , we situate our results with respect to the large body of work by the probability and statistical physics communities on the convergence of minimum spanning trees , and briefly address the question of universality . before stating our results ,
a brief word on the spaces in which we work is necessary .
we formally introduce these spaces in section [ sec : metrics ] , and here only provide a brief summary .
first , let @xmath55 be the set of measured isometry - equivalence classes of compact measured metric spaces , and let @xmath56 denote the gromov hausdorff
prokhorov distance on @xmath57 ; the pair @xmath58 forms a polish space .
we wish to think of @xmath26 as an element of @xmath58 . in order to do this ,
we introduce a measured metric space @xmath59 obtained from @xmath26 by rescaling distances by @xmath60 and assigning mass @xmath61 to each vertex .
the main contribution of this paper is the following theorem .
[ thm : main-0 ] there exists a random , compact measured metric space @xmath54 such that , as @xmath62 , @xmath63 in the space @xmath58 .
the limit @xmath54 is a random @xmath3-tree .
it is almost surely binary , and its mass measure is concentrated on the leaves of @xmath54 .
furthermore , almost surely , the minkowski dimension of @xmath54 exists and is equal to @xmath4 .
a consequence of the last statement is that @xmath54 is not a rescaled version of the brownian crt @xmath64 , in the sense that for any non - negative random variable @xmath65 , the laws of @xmath54 and the space @xmath64 , in which all distances are multiplied by @xmath65 , are mutually singular . indeed ,
the brownian tree has minkowski dimension @xmath66 almost surely .
the assertions of theorem [ thm : main-0 ] are contained within the union of theorems [ thm : big_statement ] and [ thm : basic_properties ] and corollary [ cor : notcrt ] , below . in a preprint @xcite posted simultaneously with the current work ,
the first author of this paper shows that the _ unscaled _ tree @xmath67 , when rooted at vertex @xmath68 , converges in the local weak sense to a random infinite tree , and that this limit almost surely has cubic volume growth .
the results of @xcite form a natural complement to theorem [ thm : main-0 ] .
as mentioned earlier , we approach the study of @xmath59 and of its scaling limit @xmath54 via a detailed description of the graph @xmath36 and of the forest @xmath40 , for @xmath38 near @xmath61 .
as is by this point well - known , it turns out that the right scaling for the `` critical window '' is given by taking @xmath69 , for @xmath70 , and for such @xmath38 , the largest components of @xmath36 typically have size of order @xmath71 and possess a bounded number of cycles @xcite .
adopting this parametrisation , for @xmath70 write @xmath72 for the components of @xmath73 listed in decreasing order of size ( among components of equal size , list components in increasing order of smallest vertex label , say ) . for each @xmath74
, we then write @xmath75 for the measured metric space obtained from @xmath76 by rescaling distances by @xmath60 and giving each vertex mass @xmath77 , and let @xmath78 we likewise define a sequence @xmath79 of graphs , and a sequence @xmath80 of measured metric spaces , starting from @xmath81 instead of @xmath82 . in order to compare sequences @xmath83 of elements of @xmath57 ( i.e. , elements of @xmath84 ) , we let @xmath85 , for @xmath86 , be the set of sequences @xmath87 with @xmath88 and for two such sequences @xmath89 and @xmath90 , we let @xmath91 the resulting metric space @xmath92 is a polish space . the second main result of this paper is the following ( see theorems [ thm : msf - converge ] and [ thm : big_statement ] below ) .
[ thm : msf - converge-0 ] fix @xmath70 .
then there exists a random sequence @xmath93 of compact measured metric spaces , such that as @xmath62 , @xmath94 in the space @xmath95 . furthermore , let @xmath96 be the first term @xmath97 of the limit sequence @xmath98 , with its measure renormalized to be a probability
. then as @xmath99 , @xmath96 converges in distribution to @xmath54 in the space @xmath100 .
theorem 1 of @xcite states that for each @xmath70 , there is a random sequence @xmath101 of compact measured metric spaces , such that @xmath102 in the space @xmath103 .
( theorem 1 of @xcite is , in fact , slightly weaker than this because the metric spaces there are considered without their accompanying measures , but it is easily strengthened ; see section [ sec : proofs ] . )
the limiting spaces are similar to @xmath3-trees ; we call them _
@xmath3-graphs_. in section [ sec : rtreesrgraphs ] we define @xmath3-graphs and develop a decomposition of @xmath3-graphs analogous to the classical `` core and kernel '' decomposition for finite connected graphs ( see , e.g. , @xcite ) .
we expect this generalisation of the theory of @xmath3-trees to find further applications .
the main results of @xcite provide precise distributional descriptions of the cores and kernels of the components of @xmath104 .
it turns out that , having understood the distribution of @xmath105 , we can access the distribution of @xmath106 by using a minimum spanning tree algorithm called _
cycle breaking_. this algorithm finds the minimum weight spanning tree of a graph by listing edges in _ decreasing _ order of weight , then considering each edge in turn and removing it if its removal leaves the graph connected .
using the convergence in ( [ eq : abbr ] ) and an analysis of the cycle breaking algorithm , we will establish theorem [ thm : msf - converge-0 ] .
the sequence @xmath107 is constructed from @xmath104 by a continuum analogue of the cycle breaking procedure .
showing that the continuum analogue of cycle breaking is well - defined and commutes with the appropriate limits is somewhat involved ; this is the subject of section [ sec : ckcc ] . for fixed @xmath0 , the process
@xmath108 is eventually constant , and we note that @xmath109 in order to establish that @xmath59 converges in distribution in the space @xmath58 as @xmath62 , we rely on two ingredients .
first , the convergence in ( [ eq : abbgr1 ] ) is strong enough to imply that the first component @xmath110 converges in distribution as @xmath2 to a limit @xmath111 in the space @xmath58 .
second , the results in @xcite entail lemma [ lem : cauchy - property2 ] , which in particular implies that for any @xmath112 , @xmath113 this is enough to prove a version of our main result for the metric spaces without their measures . in lemma [ lem : mst - lim ] , below , we strengthen this statement .
let @xmath114 be the measured metric space obtained from @xmath115 by rescaling so that the total mass is one ( in @xmath115 we gave each vertex mass @xmath77 ; now we give each vertex mass @xmath116 ) .
we show that for any @xmath112 , @xmath117 since @xmath58 is a complete , separable space , the so - called principle of accompanying laws entails that @xmath118 in the space @xmath58 for some limiting random measured metric space @xmath54 which is thus the scaling limit of the minimum spanning tree on the complete graph . furthermore , still as a consequence of the principle of accompanying laws , @xmath54 is also the limit in distribution of @xmath111 as @xmath119 in the space @xmath58 . for fixed @xmath70
, we will see that each component of @xmath107 is almost surely binary . since @xmath54 is compact and
( if the measure is ignored ) is an increasing limit of @xmath120 as @xmath99 , it will follow that @xmath54 is almost surely binary . to prove that the mass measure is concentrated on the leaves of @xmath54
, we use a result of uczak @xcite on the size of the largest component in the barely supercritical regime . this result in particular
implies that for all @xmath121 , @xmath122 since @xmath123 has @xmath0 vertices , it follows that for any @xmath70 , the proportion of the mass of @xmath124 `` already present in @xmath115 '' is asymptotically negligible .
but ( [ eq : abbr09b ] ) tells us that for @xmath125 large , with high probability every point of @xmath124 not in @xmath115 has distance @xmath126 from a point of @xmath115 , so has distance @xmath127 from a leaf of @xmath124 . passing this argument to the limit
, it will follow that @xmath54 almost surely places all its mass on its leaves .
the statement on the minkowski dimension of @xmath54 depends crucially on an explicit description of the components of @xmath128 from @xcite , which allows us to estimate the number of balls needed to cover @xmath129 . along with a refined version of , which yields an estimate of the distance between @xmath111 and @xmath54
, we are able to obtain bounds on the covering number of @xmath54 .
this completes our overview , and we now proceed with a brief discussion of related work , before turning to details . in the majority of the work on convergence of mst s , inter - point distances are chosen so that the edges of the mst have constant average length ( in all the models discussed above , the average edge length was @xmath130 ) . for such weights ,
the limiting object is typically a non - compact infinite tree or forest .
as detailed above , the study bifurcates into the `` geometric '' case in which the points lie in a euclidean space @xmath14 , and the `` mean - field '' case where the underlying graph is @xmath7 with i.i.d edge weights . in both cases ,
a standard approach is to pass directly to an infinite underlying graph or point set , and define the minimum spanning tree ( or forest ) directly on such a point set .
it is not _ a priori _ obvious how to define the minimum spanning tree , or forest , of an infinite graph , as neither of the algorithms described above are necessarily well - defined ( there may be no smallest weight edge leaving a given vertex or component ) .
however , it is known @xcite that given an infinite locally finite graph @xmath131 and distinct edge weights @xmath132 , the following variant of prim s algorithm is well - defined and builds a forest , each component of which is an infinite tree .
invasion percolation : for each @xmath133 , run prim s algorithm starting from @xmath134 and call the resulting set of edges @xmath135 .
then let @xmath136 be the graph with vertices @xmath137 and edges @xmath138 .
the graph @xmath136 is also described by the following rule , which is conceptually based on the coupling between kruskal s algorithm and the percolation process , described above . for each @xmath139 , let @xmath140 be the subgraph with edges @xmath141 .
then an edge @xmath142 with @xmath143 is an edge of @xmath136 if and only if @xmath144 and @xmath134 are in distinct components of @xmath140 and one of these components is finite .
the latter characterisation again allows the msf to be studied by coupling with a percolation process .
this connection was exploited by alexander and molchanov @xcite in their proof that the msf almost surely consists of a single , one - ended tree for the square , triangular , and hexagonal lattices with i.i.d.uniform@xmath32 $ ] edge weights and , later , to prove the same result for the msf of the points of a homogeneous poisson process in @xmath145 @xcite .
newman @xcite has also shown that in lattice models in @xmath146 , the critical percolation probability @xmath147 is equal to @xmath148 if and only if the msf is a proper forest ( contains more than one tree ) .
lyons , peres and schramm @xcite developed the connection with critical percolation . among several other results , they showed that if @xmath149 is any cayley graph for which @xmath150 , then the component trees in the msf all have one end almost surely , and that almost surely every component tree of the msf itself has percolation threshold @xmath151 .
( see also @xcite for subsequent work on a similar model . ) for two - dimensional lattice models , more detailed results about the behaviour of the so - called `` invasion percolation tree '' , constructed by running prim s algorithm once from a fixed vertex , have also recently been obtained @xcite . in the mean - field case ,
one common approach is to study the mst or msf from the perspective of local weak convergence @xcite .
this leads one to investigate the minimum spanning forest of aldous _ poisson - weighted infinite tree _ ( pwit ) .
such an approach is used implicitly in @xcite in studying the first @xmath152 steps of prim s algorithm on @xmath7 , and explicitly in @xcite to relate the behaviour of prim s algorithm on @xmath7 and on the pwit .
aldous @xcite establishes a local weak limit for the tree obtained from the mst of @xmath7 as follows .
delete the ( typically unique ) edge whose removal minimizes the size of the component containing vertex @xmath68 in the resulting graph , then keep only the component containing @xmath68 .
almost nothing is known about _ compact _ scaling limits for whole mst s . in two dimensions , aizenman , burchard , newman and wilson @xcite
have shown tightness for the family of random sets given by considering the subtree of the mst connecting a finite set of points ( the family is obtained by varying the set of points ) , either in the square , triangular or hexagonal lattice , or in a poisson process .
they also studied the properties of subsequential limits for such families , showing , among other results , that any limiting `` tree '' has hausdorff dimension strictly between @xmath68 and @xmath66 , and that the curves connecting points in such a tree are almost surely hlder continuous of order @xmath153 for any @xmath154 .
recently , garban , pete , and schramm @xcite announced that they have proved the existence of a scaling limit for the mst in 2d lattice models .
the mst is expected to be invariant under scalings , rotations and translations , but not conformally invariant , and to have no points of degree greater than four . in the mean - field case , however , we are not aware of any previous work on scaling limits for the mst . in short , the scaling limit @xmath54 that we identify in this paper appears to be a novel mathematical object .
it is one of the first mst scaling limits to be identified , and is perhaps the first scaling limit to be identified for any problem from combinatorial optimisation .
we expect @xmath54 to be a universal object : the mst s of a wide range of `` high - dimensional '' graphs should also have @xmath54 as a scaling limit . by way of analogy , we mention two facts .
first , peres and revelle @xcite have shown the following universality result for _ uniform _ spanning trees ( here informally stated ) .
let @xmath155 be a sequence of vertex transitive graphs of size tending to infinity .
suppose that ( a ) the uniform mixing time of simple random walk on @xmath156 is @xmath157 , and ( b ) @xmath156 is sufficiently `` high - dimensional '' , in that the expected number of meetings between two random walks with the same starting point , in the first @xmath158 steps , is uniformly bounded .
then after a suitable rescaling of distances , the spanning tree of @xmath156 converges to the crt in the sense of finite - dimensional distributions .
second , under a related set of conditions , van der hofstad and nachmias @xcite have very recently proved that the largest component of critical percolation on @xmath156 in the barely supercritical phase has the same scaling as in the erds rnyi graph process ( we omit a precise statement of their result as it is rather technical , but mention that their conditions are general enough to address the notable case of percolation on the hypercube ) . however , a proof of an analogous result for the mst seems , at this time , quite distant . as will be seen below , our proof requires detailed control on the metric and mass structure of all components of the kruskal process in the critical window and , for the moment , this is not available for any other models .
the reader may wish to simply skim this section on a first reading , referring back to it as needed . given a metric space @xmath159 ,
we write @xmath160 $ ] for the isometry class of @xmath159 , and frequently use the notation @xmath161 for either @xmath159 or @xmath160 $ ] when there is no risk of ambiguity . for a metric space @xmath159
we write @xmath162 , which may be infinite .
let @xmath163 and @xmath164 be metric spaces .
if @xmath165 is a subset of @xmath166 , the _ distortion _ @xmath167 is defined by @xmath168 a _ correspondence _
@xmath165 between @xmath169 and @xmath170 is a measurable subset of @xmath171 such that for every @xmath172 , there exists @xmath173 with @xmath174 and vice versa .
write @xmath175 for the set of correspondences between @xmath169 and @xmath170 .
the gromov
hausdorff distance @xmath176 between the isometry classes of @xmath159 and @xmath177 is @xmath178 and there is a correspondence which achieves this infimum .
( in fact , since our metric spaces are assumed separable , the requirement that the correspondence be measurable is not strictly necessary . )
it can be verified that @xmath179 is indeed a distance and , writing @xmath180 for the set of isometry classes of compact metric spaces , that @xmath181 is itself a complete separable metric space .
let @xmath182 and @xmath183 be metric spaces , each with an ordered set of @xmath184 distinguished points ( we call such spaces _ @xmath184-pointed metric spaces _ ) , we simply refer to pointed ( rather than @xmath68-pointed ) metric spaces , and write @xmath185 rather than @xmath186 .
we say that these two @xmath184-pointed metric spaces are _ isometry - equivalent _ if there exists an isometry @xmath187 such that @xmath188 for every @xmath189 .
as before , we write @xmath190 $ ] for the isometry equivalence class of @xmath182 , and denote either by @xmath191 when there is little chance of ambiguity . the _ @xmath184-pointed gromov
hausdorff distance _ is defined as @xmath192 much as above , the space @xmath193 of isometry classes of @xmath184-pointed compact metric spaces is itself a complete separable metric space .
a _ compact measured metric space _ is a triple @xmath194 where @xmath159 is a compact metric space and @xmath15 is a ( non - negative ) finite measure on @xmath195 , where @xmath196 is the borel @xmath197-algebra on @xmath159 . given a measured metric space @xmath194 , a metric space @xmath177 and a measurable function @xmath198 , we write @xmath199 for the push - forward of the measure @xmath15 to the space @xmath177 .
two compact measured metric spaces @xmath194 and @xmath200 are called _ isometry - equivalent _ if there exists an isometry @xmath201 such that @xmath202 . the isometry - equivalence class of @xmath194 will be denoted by @xmath203 $ ] .
again , both will often be denoted by @xmath161 when there is little risk of ambiguity .
if @xmath204 then we write @xmath205 .
there are several natural distances on compact measured metric spaces that generalize the gromov
hausdorff distance , see for instance @xcite .
the presentation we adopt is still different from these references , but closest in spirit to @xcite since we are dealing with arbitrary finite measures rather than just probability measures . in particular
, it induces the same topology as the compact gromov hausdorff
prokhorov metric of @xcite .
if @xmath159 and @xmath177 are two metric spaces , let @xmath206 be the set of finite non - negative borel measures on @xmath166 .
we will denote by @xmath207 the canonical projections from @xmath208 to @xmath169 and @xmath170 .
let @xmath15 and @xmath209 be finite non - negative borel measures on @xmath169 and @xmath170 respectively .
the _ discrepancy _ of @xmath210 with respect to @xmath15 and @xmath209 is the quantity @xmath211 where @xmath212 is the total variation of the signed measure @xmath213 .
note in particular that @xmath214 , by the triangle inequality and the fact that @xmath215 , where @xmath216 is the total mass of @xmath213 . if @xmath15 and @xmath209 are probability distributions ( or have the same mass ) , a measure @xmath210 with @xmath217 is a coupling of @xmath15 and @xmath209 in the standard sense .
recall that the prokhorov distance between two finite non - negative borel measures @xmath15 and @xmath209 on the _ same _ metric space @xmath159 is given by @xmath218 an alternative distance , which generates the same topology but more easily extends to the setting where @xmath15 and @xmath209 are measures on different metric spaces , is given by @xmath219 to extend this , we replace the condition on @xmath220 by an analogous condition on the measure of the set of pairs lying outside the correspondence .
more precisely , let @xmath221 and @xmath222 be measured metric spaces .
the gromov hausdorff
prokhorov distance between @xmath191 and @xmath223 is defined as @xmath224 the infimum being taken over all @xmath225 and @xmath226 . here and elsewhere we write @xmath227 ( and , likewise , @xmath228 ) . just as for @xmath179
, it can be verified that @xmath56 is a distance and that writing @xmath57 for the set of measured isometry - equivalence classes of compact measured metric spaces , @xmath58 is a complete separable metric space ( see , e.g. , @xcite ) .
note that @xmath229 .
in other words , the mapping @xmath160\mapsto [ x , d,0]$ ] is an isometric embedding of @xmath230 into @xmath55 , and we will sometimes abuse notation by writing @xmath160\in \mathcal{m}$ ] .
note also that @xmath231 in particular , if @xmath232 is the `` zero '' metric space consisting of a single point with measure @xmath148 , then @xmath233\ , .\ ] ] finally , we can define an analogue of @xmath56 for measured isometry - equivalence class of spaces of the form @xmath234 where @xmath235 are points of @xmath169 and @xmath236 are finite borel measures on @xmath169 . if @xmath237 are such spaces , whose measured , pointed isometry classes are denoted by @xmath238 , we let @xmath239 where the infimum is over all @xmath240 such that @xmath241 and all @xmath242 . writing @xmath243 for
the set of measured isometry - equivalence classes of compact metric spaces equipped with @xmath184 marked points and @xmath244 finite borel measures , we again obtain a complete separable metric space @xmath245 .
we will need the following fact , which is in essence ( * ? ? ?
* proposition 10 ) , except that we have to take into account more measures and/or marks .
this is a minor modification of the setting of @xcite , and the proof is similar .
[ sec : grom - hausd - prokh ] let @xmath246 converge to @xmath247 in @xmath243 , and assume that the first measure @xmath248 of @xmath249 is a probability measure for every @xmath250 .
let @xmath251 be a random variable with distribution @xmath252 , and let @xmath253 .
then @xmath254 converges in distribution to @xmath255 in @xmath256 .
we now consider a natural metric on certain sequences of measured metric spaces . for @xmath86 and @xmath257 in @xmath258 ,
we let @xmath259 if @xmath260 for some @xmath261 , we consider @xmath262 as an element of @xmath84 by appending to @xmath262 an infinite sequence of copies of the `` zero '' metric space @xmath232 .
this allows us to use @xmath263 to compare sequences of metric spaces with different numbers of elements , and to compare finite sequences with infinite sequences . in particular ,
let @xmath264 , and @xmath265 so , by , @xmath266 if and only if the sequences @xmath267 and @xmath268 are in @xmath269 .
we leave the reader to check that @xmath92 is a complete separable metric space .
let @xmath159 be a metric space .
for @xmath270 and @xmath271 , we let @xmath272 and @xmath273 .
we say @xmath159 is _ degenerate _ if @xmath274 . as regards metric spaces , we mostly follow @xcite for our terminology .
let @xmath275,x)$ ] be the set of continuous functions from @xmath276 $ ] to @xmath169 , hereafter called _ paths with domain @xmath276 $ ] _ or _ paths from @xmath277 to @xmath278_. the image of a path is called an _ arc _ ; it is a _ simple arc _ if the path is injective .
if @xmath279,x)$ ] , the length of @xmath280 is defined by @xmath281 , t_0\leq t_1\leq \ldots\leq t_k\bigg\}\ , .\ ] ] if @xmath282 , then the function @xmath283\to [ 0,{\ensuremath{\mathrm{len}}}(f)]$ ] defined by @xmath284})$ ] is non - decreasing and surjective .
the function @xmath285 , where @xmath286 is the right - continuous inverse of @xmath287 , is easily seen to be continuous , and we call it the path @xmath280 _ parameterized by arc - length_. the _ intrinsic distance _ ( or _ intrinsic metric _ ) associated with @xmath159 is the function @xmath288 defined by @xmath289,x ) , f(0)=x , f(1)=y\}\ , .\ ] ] the function @xmath288 need not take finite values .
when it does , then it defines a new distance on @xmath169 such that @xmath290 .
the metric space @xmath159 is called _ intrinsic _ if @xmath291 .
similarly , if @xmath292 then the _ intrinsic metric on @xmath293 _ is given by @xmath289,y ) , f(0)=x , f(1)=y\}\ , .\ ] ] given @xmath294 , a _ geodesic between @xmath295 and @xmath296 _ ( also called a _
shortest path between @xmath295 and @xmath296 _ ) is an isometric embedding @xmath297 \to x$ ] such that @xmath298 and @xmath299 ( so that obviously @xmath300 ) . in this case
, we call the image @xmath301 a _ geodesic arc between @xmath295 and @xmath296_. a metric space @xmath159 is called a geodesic space if for any two points @xmath302 there exists a geodesic between @xmath295 and @xmath296 .
a geodesic space is obviously an intrinsic space .
if @xmath159 is compact , then the two notions are in fact equivalent .
also note that for every @xmath295 in a geodesic space and @xmath303 , @xmath304 is the closure of @xmath305 .
essentially all metric spaces @xmath159 that we consider in this paper are in fact compact geodesic spaces .
a path @xmath279,x)$ ] is a _ local geodesic _ between @xmath295 and @xmath296
if @xmath298 , @xmath299 , and for any @xmath306 $ ] there is a neighborhood @xmath137 of @xmath307 in @xmath276 $ ] such that @xmath308 is a geodesic .
it is then straightforward that @xmath309 .
( our terminology differs from that of @xcite , where this would be called a geodesic .
we also note that we do not require @xmath295 and @xmath296 to be distinct . )
an _ embedded cycle _ is the image of a continuous injective function @xmath310 , where @xmath311 .
the length @xmath312 is the length of the path @xmath313\to x$ ] defined by @xmath314 for @xmath315 .
it is easy to see that this length depends only on the embedded cycle @xmath316 rather than its particular parametrisation .
we call it the _ length _ of the embedded cycle , and write @xmath317 for this length . a metric space with no embedded cycle is called _ acyclic _ , and a metric space with exactly one embedded cycle is called _
unicyclic_. a metric space @xmath318 is an _
@xmath3-tree _ if it is an acyclic geodesic metric space . if @xmath159 is an @xmath3-tree then for @xmath319 , the _ degree _ @xmath320 of @xmath295 is the number of connected components of @xmath321 .
is a point of degree @xmath68 ; we let @xmath322 be the set of leaves of @xmath191 .
a metric space @xmath159 is an _
@xmath3-graph _ if it is locally an @xmath3-tree in the following sense .
note that by definition an @xmath3-graph is connected , being a geodesic space .
[ def : r - graph ] a compact geodesic metric space @xmath159 is an _ @xmath3-graph _ if for every @xmath270 , there exists @xmath323 such that @xmath324 is an @xmath3-tree .
let @xmath318 be an @xmath3-graph and fix @xmath270 .
the degree of @xmath295 , denoted by @xmath320 and with values in @xmath325 , is defined to be the degree of @xmath295 in @xmath326 for every @xmath327 small enough so that @xmath328 is an @xmath3-tree , and this definition does not depend on a particular choice of @xmath327 .
if @xmath329 and @xmath330 , we can likewise define the degree @xmath331 of @xmath295 in @xmath293 as the degree of @xmath295 in the @xmath3-tree @xmath332 , where @xmath333 is the connected component of @xmath293 that contains @xmath295 , for any @xmath327 small enough . obviously , @xmath334 whenever @xmath335 .
let @xmath336 an element of @xmath322 is called a _ leaf of @xmath191 _ , and the set @xmath337 is called the _ skeleton of @xmath191_. a point with degree at least @xmath4 is called a _
branchpoint of @xmath191_. we let @xmath338 be the set of branchpoints of @xmath191 . if @xmath191 is , in fact , an @xmath3-tree , then @xmath337 is the set of points whose removal disconnects the space , but this is not true in general . alternatively
, it is easy to see that @xmath339 where for @xmath294 , @xmath340 denotes the collection of all geodesic arcs between @xmath295 and @xmath296 .
since @xmath159 is separable , this may be re - written as a countable union , and so there is a unique @xmath197-finite borel measure @xmath341 on @xmath169 with @xmath342 for every injective path @xmath343 , and such that @xmath344 .
the measure @xmath341 is the hausdorff measure of dimension @xmath68 on @xmath191 , and we refer to it as the _ length measure _ on @xmath169
. if @xmath159 is an @xmath3-graph then the set @xmath345 is countable ( as is classically the case for compact @xmath3-trees ) , and hence this set has measure zero under @xmath341 .
[ def : core ] let @xmath159 be an @xmath3-graph . its _ core _ , denoted by @xmath346 , is the union of all the simple arcs having both endpoints in embedded cycles of @xmath191 .
if it is non - empty , then @xmath347 is an @xmath3-graph with no leaves .
the last part of this definition is in fact a proposition , which is stated more precisely and proved below as proposition [ sec : skeleton - core - kernel ] .
since the core of @xmath169 encapsulates all the embedded cycles of @xmath169 , it is intuitively clear that when we remove @xmath346 from @xmath169 , we are left with a family of @xmath3-trees .
this can be formalized as follows .
fix @xmath348 , and let @xmath280 be a shortest path from @xmath295 to @xmath346 , i.e. , a geodesic from @xmath295 to @xmath349 , where @xmath349 is chosen so that @xmath312 is minimum .
( recall that @xmath350 ) is a closed subspace of @xmath169 ) .
this shortest path is unique , otherwise we would easily be able to construct an embedded cycle @xmath44 not contained in @xmath346 , contradicting the definition of @xmath346 .
let @xmath351 be the endpoint of this path not equal to @xmath295 , which is thus the unique point of @xmath346 that is closest to @xmath295 .
by convention , we let @xmath352 if @xmath353 .
we call @xmath351 the point of attachment of @xmath295 .
[ sec : structure - r - graphs-6 ] the relation @xmath354 is an equivalence relation on @xmath169 . if @xmath355 $ ] is the equivalence class of @xmath295 , then @xmath356,d)$ ] is a compact @xmath3-tree .
the equivalence class @xmath355 $ ] of a point @xmath357 is a singleton if and only if @xmath358 .
the fact that @xmath359 is an equivalence relation is obvious . fix any equivalence class @xmath355 $ ] .
note that @xmath355\cap { \ensuremath{\mathrm{core}}}({\ensuremath{\mathrm{x}}})$ ] contains only the point @xmath351 , so that @xmath355 $ ] is connected and acyclic by definition .
hence , any two points of @xmath355 $ ] are joined by a unique simple arc ( in @xmath355 $ ] ) .
this path is moreover a shortest path for the metric @xmath360 , because a path starting and ending in @xmath355 $ ] , and visiting @xmath361 $ ] , must pass at least twice through @xmath351 ( if this were not the case , we could find an embedded cycle not contained in @xmath346 ) .
the last statement is easy and left to the reader .
[ sec : r - trees - r-1 ] if @xmath159 is an @xmath3-graph , then @xmath346 is the maximal closed subset of @xmath169 having only points of degree greater than or equal to @xmath66 .
if @xmath293 is closed and strictly contains @xmath346 , then we can find @xmath330 such that @xmath362 is maximal . then @xmath363 $ ]
is included in the set of points @xmath364 $ ] such that the geodesic arc from @xmath296 to @xmath351 does not pass through @xmath295 .
this set is an @xmath3-tree in which @xmath295 is a leaf , so @xmath365 .
note that this characterisation is very close to the definition of the core of a ( discrete ) graph .
another important structural component is @xmath366 , the set of points of @xmath346 such that @xmath367 is connected .
figure [ fig : rgraph ] summarizes the preceding definitions .
the space @xmath366 is not connected or closed in general . clearly , a point of @xmath366 must be contained in an embedded cycle of @xmath169 , but the converse is not necessarily true .
a partial converse is as follows .
[ sec : skeleton - core - kernel-2 ] let @xmath353 have degree @xmath368 and suppose @xmath295 is contained in an embedded cycle of @xmath169
. then @xmath369 .
let @xmath44 be an embedded cycle containing @xmath295 .
fix @xmath370 , and let @xmath371 be geodesics from @xmath372 to their respective closest points @xmath373 . note that @xmath374 is distinct from @xmath295 because otherwise , @xmath295 would have degree at least @xmath4 .
likewise , @xmath375 .
let @xmath376 be a parametrisation of the arc of @xmath44 between @xmath374 and @xmath377 that does not contain @xmath295 , then the concatenation of @xmath371 and the time - reversal of the path @xmath376 is a path from @xmath296 to @xmath378 , not passing through @xmath295 .
hence , @xmath321 is connected .
let us now discuss the structure of @xmath346 .
equivalently , we need to describe @xmath3-graphs with no leaves , because such graphs are equal to their cores by corollary [ sec : r - trees - r-1 ] . a _
graph with edge - lengths _ is a triple @xmath379 where @xmath380 is a finite connected multigraph , and @xmath381 for every @xmath382 . with every such object , one can associate an @xmath3-graph without leaves , which is the _ metric graph _ obtained by viewing the edges of @xmath380 as segments with respective lengths @xmath383 . formally , this @xmath3-graph is the metric gluing of disjoint copies @xmath384 of the real segments @xmath385,e\in e$ ] according to the graph structure of @xmath380 .
we refer the reader to @xcite for details on metric gluings and metric graphs . in section [ sec : rtreesrgraphs ]
, we will prove the following result .
[ sec : structure - r - graphs-4 ] an @xmath3-graph with no leaves is either a cycle , or is the metric gluing of a finite connected multigraph with edge - lengths in which all vertices have degree at least @xmath4 .
the associated multigraph , without the edge - lengths , is called the kernel of @xmath169 , and denoted by @xmath386 . the precise definition of @xmath387 , and the proof of theorem [ sec : structure - r - graphs-4 ] , both appear in section [ sec : core - metric - gluing ] .
-graph @xmath159 , emphasizing the structural components .
@xmath346 is in thick line ( black and red ) , with @xmath366 in red .
the subtrees hanging from @xmath346 are in thin blue line .
kernel vertices are represented as large dots .
an example of the projection @xmath388 is provided . ] for a connected multigraph @xmath131 , the _ surplus _ @xmath389 is @xmath390 . for an @xmath3-graph @xmath159 ,
we let @xmath391 if @xmath387 is non - empty . otherwise , either @xmath159 is an @xmath3-tree or @xmath346 is a cycle .
in the former case we set @xmath392 ; in the latter we set @xmath393 . since the degree of every vertex in @xmath387 is at least @xmath4 , we have @xmath394 , and so if @xmath395 we have @xmath396 with equality precisely if @xmath387 is @xmath4-regular .
let @xmath131 be a finite connected multigraph .
let @xmath397 be the set of of all edges @xmath398 such that @xmath399 is connected .
if @xmath400 , then @xmath149 contains at least one cycle and @xmath397 is non - empty . in this case , let @xmath401 be a uniform random edge in @xmath397 , and let @xmath402 be the law of the multigraph @xmath403 .
if @xmath404 , then @xmath402 is the dirac mass at @xmath149 . by definition
, @xmath405 is a markov kernel from the set of graphs with surplus @xmath406 to the set of graphs with surplus @xmath407 .
writing @xmath408 for the @xmath0-fold application of the kernel @xmath405 , we have that @xmath409 does not depend on @xmath0 for @xmath410 .
we define the kernel @xmath411 to be equal to this common value : a graph has law @xmath411 if it is obtained from @xmath149 by repeatedly removing uniform non - disconnecting edges .
[ sec : cycle - break - algor ] the probability distribution @xmath411 is the law of the minimum spanning tree of @xmath149 , when the edges @xmath412 are given exchangeable , distinct random edge - weights .
we prove by induction on the surplus of @xmath149 the stronger statement that @xmath411 is the law of the minimum spanning tree of @xmath149 , when the weights of @xmath397 are given exchangeable , distinct random edge - weights . for @xmath404
the result is obvious .
assume the result holds for every graph of surplus @xmath406 , and let @xmath149 have @xmath413 .
let @xmath401 be the edge of @xmath397 with maximal weight , and condition on @xmath401 and its weight .
then , note that the weights of the edges in @xmath414 are still in exchangeable random order , and the same is true of the edges of @xmath415 . by the induction hypothesis , @xmath416 is the law of the minimum spanning tree of @xmath417 .
but @xmath401 is not in the minimum spanning tree of @xmath149 , because by definition we can find a path between its endpoints that uses only edges having strictly smaller weights .
hence , @xmath416 is the law of the minimum spanning tree of @xmath149 . on the other hand , by exchangeability ,
the edge @xmath401 of @xmath397 with largest weight is uniform in @xmath397 , so the unconditional law of a random variable with law @xmath418 is @xmath419 .
there is a continuum analogue of the cycle - breaking algorithm in the context of @xmath3-graphs , which we now explain .
recall that @xmath366 is the set of points @xmath295 of the @xmath3-graph @xmath318 such that @xmath420 and @xmath321 is connected . for @xmath421 ,
we let @xmath422 be the space @xmath191 `` cut at @xmath295 '' .
formally , it is the metric completion of @xmath423 , where @xmath424 is the intrinsic distance : @xmath425 is the minimal length of a path from @xmath296 to @xmath374 that does not visit @xmath295 .
[ sec : cutting - cycles - an ] a point @xmath270 in a measured @xmath3-graph @xmath204 is a _ regular point _ if @xmath426 , and moreover @xmath427 and @xmath368 . a marked space @xmath428 , where @xmath159 is an @xmath3-graph and @xmath295 is a regular point , is called _
safely pointed_. we say that a pointed @xmath3-graph @xmath185 is safely pointed if @xmath429 is safely pointed .
if @xmath295 is a regular point then @xmath15 induces a measure ( still denoted by @xmath15 ) on the space @xmath430 with the same total mass
. we will give a precise description of the space @xmath431 in section [ sec : cutting - procedure ] : in particular , it is a measured @xmath3-graph with @xmath432 . note that if @xmath433 and if @xmath434 is the length measure restricted to @xmath366 , then @xmath435-almost every point is regular .
also , @xmath435 is a finite measure by theorem [ sec : structure - r - graphs-4 ] .
therefore , it makes sense to let @xmath436 be the law of @xmath430 , where @xmath295 is a random point of @xmath169 with law @xmath437 . if @xmath392 we let @xmath438 .
again , @xmath439 is a markov kernel from the set of measured @xmath3-graphs with surplus @xmath406 to the set of measured @xmath3-graphs of surplus @xmath440 , and @xmath441 for every @xmath442 : we denote this by @xmath443 . in section
[ sec : ckcc ] we will give details of the proofs of the aforementioned properties , as well as of the following crucial result .
for @xmath444 we let @xmath445 be the set of measured @xmath3-graphs with @xmath446 and whose core , seen as a graph with edge - lengths @xmath447 , is such that @xmath448 ( if @xmath393 , this should be understood as the fact that @xmath346 is a cycle with length in @xmath449 $ ] . )
[ sec : breaking - cycles - r-1 ] fix @xmath444 .
let @xmath450 be a sequence of measured @xmath3-graphs in @xmath445 , converging as @xmath2 to @xmath451 in @xmath100 .
then @xmath452 converges weakly to @xmath443 .
we can view any finite connected multigraph @xmath131 as a metric space @xmath453 , where @xmath454 is the least number of edges in any chain from @xmath144 to @xmath134 .
we may also consider the metric graph @xmath455 associated with @xmath149 by treating edges as segments of length @xmath68 ( this is sometimes known as the _ cable system _ for the graph @xmath149 @xcite ) .
then @xmath455 is an @xmath3-graph .
note that @xmath456 and , in fact , @xmath455 contains an isometric copy of @xmath453 .
also , temporarily writing @xmath457 for the _ graph - theoretic _ core of @xmath149 , that is , the maximal subgraph of @xmath149 of minimum degree two , it is straightforwardly checked that @xmath458 is isometric to @xmath459 .
conversely , let @xmath159 be an @xmath3-graph , and let @xmath460 be the set of points in @xmath169 with degree at least three .
we say that @xmath159 _ has integer lengths _ if all local geodesics between points in @xmath460 have lengths in @xmath461 .
let @xmath462 and note that if @xmath159 is compact and has integer lengths then necessarily @xmath463 and @xmath464 .
the removal of all points in @xmath465 separates @xmath169 into a finite collection of paths , each of which is either an open path of length one between two points of @xmath465 , or a half - open path of length strictly less than one between a point of @xmath465 and a leaf .
create an edge between the endpoints of each such _ open _ path , and call the collection of such edges @xmath466 .
then let @xmath467 we call the multigraph @xmath468 the _ graph corresponding to @xmath191 _
( see figure [ fig : graphspace ] ) .
@xmath469{graphspace1.pdf } & \includegraphics[width=0.3\textwidth]{graphspace2.pdf } & \includegraphics[width=0.3\textwidth]{graphspace3.pdf } \end{array}$ ] now , fix an @xmath3-graph @xmath470 which has integer lengths and surplus @xmath471 .
let @xmath472 be the points sampled by the successive applications of @xmath439 to @xmath191 : given @xmath473 , the point @xmath474 is chosen according to @xmath475 on @xmath476 , where @xmath477 is the space @xmath191 cut successively at @xmath478 .
note that @xmath479 can also naturally be seen as a point of @xmath169 for @xmath480 . since the length measure of @xmath465 is @xmath148 ,
almost surely @xmath481 for all @xmath482 .
thus , each point @xmath479 , @xmath482 , falls in a path component of @xmath483 which itself corresponds uniquely to an edge in @xmath484 .
note that the edges @xmath485 , @xmath482 , are distinct by construction .
then let @xmath486 , and for @xmath487 , write @xmath488 by construction , the graph @xmath489 is connected and has surplus precisely @xmath490 , and in particular @xmath491 is a spanning tree of @xmath468 .
let @xmath492 be the random @xmath3-graph resulting from the application of @xmath493 , that is obtained by cutting @xmath191 at the points @xmath494 in our setting .
[ prop : cut - close ] we have @xmath495 . first , notice that @xmath491 and @xmath496 are isomorphic as graphs , so isometric as metric spaces .
also , as noted in greater generality at the start of the subsection , we automatically have @xmath497 . [
prop : cut - dist ] the graph @xmath496 is identical in distribution to the minimum - weight spanning tree of @xmath468 when the edges of @xmath498 are given exchangeable , distinct random edge weights .
when performing the discrete cycle - breaking on @xmath468 , the set of edges removed from @xmath468 is identical in distribution to the set @xmath499 of edges that are removed from @xmath468 to create @xmath491 , so @xmath491 has the same distribution as the minimum spanning tree by proposition [ sec : cycle - break - algor ] .
furthermore , as noted in the proof of the preceding proposition , @xmath491 and @xmath496 are isomorphic .
we end this section by mentioning the operation of gluing , which in a vague sense is dual to the cutting operation .
if @xmath194 is an @xmath3-graph and @xmath302 are two distinct points of @xmath169 , we let @xmath500 be the quotient metric space @xcite of @xmath159 by the smallest equivalence relation for which @xmath295 and @xmath296 are equivalent . this space is endowed with the push - forward of @xmath15 by the canonical projection @xmath38 .
it is not difficult to see that @xmath500 is again an @xmath3-graph , and that the class of the point @xmath501 has degree @xmath502 .
similarly , if @xmath503 is a finite set of unordered pairs @xmath504 with @xmath505 in @xmath169 , then one can identify @xmath479 and @xmath506 for each @xmath13 , resulting in an @xmath3-graph @xmath507 .
we are now ready to state and prove the main results of this paper .
we begin by recalling from the introduction that we write @xmath26 for the mst of the complete graph on @xmath0 vertices and @xmath59 for the measured metric space obtained from @xmath26 by rescaling the graph distance by @xmath60 and assigning mass @xmath61 to each vertex .
recall that @xmath36 is the erds
rnyi random graph . for @xmath70
, we write @xmath508 for the components of @xmath73 listed in decreasing order of size ( among components of equal size , list components in increasing order of smallest vertex label , say ) . for each @xmath74 , we then write @xmath75 for the measured metric space obtained from @xmath76 by rescaling the graph distance by @xmath60 and giving each vertex mass @xmath77 , and let @xmath78 in a moment , we will state a scaling limit result for @xmath105 ; before we can do so , however , we must introduce the limit sequence of measured metric spaces @xmath509 .
we will do this somewhat briefly , and refer the interested reader to @xcite for more details and distributional properties .
first , consider the stochastic process @xmath510 defined by @xmath511 where @xmath512 is a standard brownian motion .
consider the excursions of @xmath513 above its running minimum ; in other words , the excursions of @xmath514 above 0 .
we list these in decreasing order of length as @xmath515 where , for @xmath74 , @xmath516 is the length of @xmath517 .
( we suppress the @xmath125-dependence in the notation for readability . ) for definiteness , we shift the origin of each excursion to 0 , so that @xmath518 \to { \mathbb{r}}_+$ ] is a continuous function such that @xmath519 and @xmath520 otherwise .
now for @xmath74 and for @xmath521 $ ] , define a pseudo - distance via @xmath522 declare that @xmath523 if @xmath524 , so that @xmath359 is an equivalence relation on @xmath525 $ ]
. now let @xmath526/\!\!\sim$ ] and denote by @xmath527 \to \mathcal{t}^i$ ] the canonical projection
. then @xmath528 induces a distance on @xmath529 , still denoted by @xmath528 , and it is standard ( see , for example , @xcite ) that @xmath530 is a compact @xmath3-tree . write @xmath531 for the push - forward of lebesgue measure on @xmath525 $ ] by @xmath532 , so that @xmath533 is a measured @xmath3-tree of total mass @xmath516 .
we now decorate the process @xmath534 with the points of an independent homogeneous poisson process in the plane
. we can think of the points which fall under the different excursions separately .
in particular , to the excursion @xmath517 , we associate a finite collection @xmath535 of points of @xmath525 \times [ 0 , \infty)$ ] which are the poisson points shifted in the same way as the excursion @xmath517 .
( for definiteness , we list the points of @xmath536 in increasing order of first co - ordinate . )
conditional on @xmath537 , the collections @xmath538 of points are independent .
moreover , by construction , given the excursion @xmath517 , we have @xmath539 .
let @xmath540 and note that , by the continuity of @xmath517 , @xmath541 almost surely .
let @xmath542 then @xmath543 is a collection of unordered pairs of points in the @xmath3-tree @xmath529 .
we wish to glue these points together in order to obtain an @xmath3-graph , as in section [ subsec : gluing ] .
we define a new equivalence relation @xmath359 by declaring @xmath523 in @xmath529 if @xmath544
. then let @xmath545 be @xmath546 , let @xmath547 be the quotient metric @xcite , and let @xmath548 be the push - forward of @xmath531 to @xmath545 .
then set @xmath549 and @xmath550 .
we note that for each @xmath74 , the measure @xmath531 is almost surely concentrated on the leaves of @xmath529 . as a consequence ,
@xmath548 is almost surely concentrated on the leaves of @xmath545 .
given an @xmath3-graph @xmath191 , write @xmath551 for the minimal length of a core edge in @xmath191 .
then @xmath552 whenever @xmath387 is non - empty .
we use the convention that @xmath553 if @xmath554 and @xmath555 if @xmath191 has a unique embedded cycle @xmath44 . recall also that @xmath471 denotes the surplus of @xmath191 .
[ thm : gnp - converge ] fix @xmath70 .
then as @xmath62 , we have the following joint convergence @xmath556 the first convergence takes place in the space @xmath557 .
the others are in the sense of finite - dimensional distributions .
let @xmath558 .
corollary 2 of @xcite gives the following joint convergence @xmath559 where the first convergence is in @xmath560 and the second is in the sense of finite - dimensional distributions .
( of course , @xmath561 and @xmath562 . )
theorem 1 of @xcite extends this to give that , jointly , @xmath563 in the sense of @xmath564 , where for @xmath565 , @xmath566 .
we need to improve this convergence from @xmath564 to @xmath567 .
first we show that we can get ghp convergence componentwise .
we do this in two lemmas .
[ lem : measured_identifications ] suppose that @xmath568 and @xmath569 are measured @xmath3-trees , that @xmath570 are pairs of points in @xmath571 and that @xmath572 are pairs of points in @xmath573 . then if @xmath574 and @xmath575 are the measured metric spaces obtained by identifying @xmath479 and @xmath506 in @xmath571 and @xmath576 and @xmath577 in @xmath573 , for all @xmath578 , we have @xmath579 where @xmath580 and similarly for @xmath581 .
let @xmath165 and @xmath582 be a correspondence and a measure which realise the gromov
prokhorov distance between @xmath583 and @xmath584 ; write @xmath585 for this distance .
note that by definition , @xmath586 and @xmath587 for @xmath588 . now make the vertex identifications in order to obtain @xmath589 and @xmath590 ;
let @xmath591 and @xmath592 be the corresponding canonical projections .
then @xmath593 is a correspondence between @xmath589 and @xmath590 .
let @xmath594 be the push - forward of the measure @xmath582 by @xmath595
. then @xmath596 and @xmath597 . moreover , by lemma 21 of @xcite , we have @xmath598 . the claimed result follows .
[ lem : componentwise ] fix @xmath74 .
then as @xmath62 , @xmath599 in @xmath600 .
this proof is a fairly straightforward modification of the proof of theorem 22 in @xcite , so we will only sketch the argument .
consider the component @xmath601 .
since we have fixed @xmath125 and @xmath13 , let us drop them from the notation and simply write @xmath602 for the component , and similarly for other objects .
write @xmath603 for the size of @xmath602 and @xmath604 for its surplus .
we can list the vertices of this graph in depth - first order as @xmath605 .
let @xmath606 be the graph distance of vertex @xmath607 from @xmath608 .
then it is easy to see that @xmath609 encodes a tree @xmath610 on vertices @xmath605 with metric @xmath611 such that @xmath612 .
we endow @xmath610 with a measure by letting each vertex of @xmath610 have mass @xmath77 .
next , let the pairs @xmath613 give the indices of the surplus edges required to obtain @xmath614 from @xmath610 , listed in increasing order of first co - ordinate .
in other words , to build @xmath602 from @xmath610 , we add an edge between @xmath615 and @xmath616 for each @xmath617 ( and re - multiply distances by @xmath1 ) .
recall that to get @xmath618 from @xmath614 , we rescale the graph distance by @xmath60 and assign mass @xmath77 to each vertex .
it is straightforward that @xmath618 is at ghp distance at most @xmath619 from the metric space @xmath620 obtained from @xmath610 by _ identifying _ vertices @xmath615 and @xmath616 for all @xmath617 . from the proof of theorem 22 of @xcite
, we have jointly @xmath621 by skorokhod s representation theorem , we may work on a probability space where these convergences hold almost surely .
consider the @xmath3-tree @xmath622 encoded by @xmath623 and recall that @xmath624 is the canonical projection @xmath625
\to \mathcal{t}$ ] .
we extend @xmath624 to a function on @xmath25 by letting @xmath626 .
let @xmath627 be the function defined by @xmath628 .
set @xmath629,|t - t'|\leq \delta_n\},\ ] ] where @xmath630 converges to @xmath148 slowly enough , that is , @xmath631 then @xmath632 is a correspondence between @xmath610 and @xmath571 that contains @xmath633 and @xmath634 for every @xmath635 , and with distortion going to @xmath148 .
next , let @xmath636 be the push - forward of lebesgue measure on @xmath637 $ ] under the mapping @xmath638 .
then the discrepancy of @xmath636 with respect to the uniform measure @xmath639 on @xmath610 and the image @xmath15 of lebesgue measure by @xmath624 on @xmath571 is @xmath640 , and @xmath641 . for all large enough @xmath0 , @xmath642 , so
let us assume henceforth that this holds .
then , writing @xmath643 and @xmath644 , we have @xmath645 almost surely , as @xmath62 . by lemma [ lem : measured_identifications ]
we thus have @xmath646 almost surely , as @xmath647 .
since @xmath648 , it follows that @xmath649 as well . by ( [ eq : aldous_joint_convergence ] ) , ( [ eq : abbrgo ] ) , lemma [ lem : componentwise ] and skorokhod s representation theorem , we may work in a probability space in which the convergence in ( [ eq : aldous_joint_convergence ] ) and in ( [ eq : abbrgo ] ) occur almost surely , and in which for every @xmath74 we almost surely have @xmath650 as @xmath62 .
now , for each @xmath74 , @xmath651 the proof of theorem 24 from @xcite shows that almost surely @xmath652 and ( [ eq : abbrgo ] ) then implies that almost surely @xmath653 the @xmath654 convergence of the masses entails that almost surely @xmath655 and ( [ eq : aldous_joint_convergence ] ) then implies that almost surely @xmath656 hence , on this probability space , we have @xmath657 almost surely . combined with ( [ eq : coordconv ] )
, this implies that in this space , almost surely @xmath658 the convergence of @xmath659 to @xmath660 follows from ( [ eq : aldous_joint_convergence ] ) . if @xmath13 is such that @xmath661 then , by ( [ eq : aldous_joint_convergence ] ) , we almost surely have @xmath662 for all @xmath0 sufficiently large . in this case ,
@xmath663 and @xmath664 are the lengths of the unique cycles in @xmath75 and in @xmath665 , respectively .
now , @xmath666 almost surely in @xmath230 , and it follows easily that in this space , @xmath667 almost surely , for @xmath13 such that @xmath661 . finally , by theorem 4 of @xcite , @xmath668 is bounded away from zero in probability .
so by skorokhod s representation theorem , we may assume our space is such that almost surely @xmath669 in particular , it follows from the above that , for any @xmath74 with @xmath670 , there is almost surely @xmath139 such that @xmath671 and @xmath672 for all n sufficiently large .
corollary [ sec : cutt - safely - point-1 ] ( i ) then yields that in this space , @xmath673 almost surely . together , the two preceding paragraphs establish the final claimed convergence . for completeness
, we note that this final convergence may also be deduced without recourse to the results of @xcite ; here is a brief sketch , using the notation of the previous lemma .
it is easily checked that the points of the kernels of @xmath674 and @xmath675 correspond to the identified vertices @xmath676 and @xmath677 , and those vertices of degree at least @xmath4 in the subtrees of @xmath678 spanned by the points @xmath679 and @xmath680 respectively .
these trees are combinatorially finite trees ( i.e. , they are finite trees with edge - lengths ) , so the convergence of the marked trees @xmath681 to @xmath682 entails in fact
the convergence of the same trees marked not only by @xmath683 but also by the points of degree @xmath4 on their skeletons .
write @xmath684 for these enlarged collections of points .
then one concludes by noting that @xmath685 ( resp.@xmath686 ) is the minimum quotient distance , after the identifications @xmath676 ( resp .
@xmath677 ) between any two distinct elements of @xmath687 ( resp . @xmath581 ) .
this entails that @xmath663 converges almost surely to @xmath664 for each @xmath74 .
the above description of the sequence @xmath104 of random @xmath3-graphs does not make the distribution of the cores and kernels of the components explicit .
( clearly the kernel of @xmath688 is only non - empty if @xmath670 and its core is only non - empty if @xmath689 . )
such an explicit distributional description was provided in @xcite , and will be partially detailed below in section [ sec : properties ] .
recall that @xmath40 is the minimum spanning forest of @xmath36 and that we write @xmath690 for the components of @xmath691 listed in decreasing order of size .
for each @xmath74 we write @xmath692 for the measured metric space obtained from @xmath693 by rescaling the graph distance by @xmath60 and giving each vertex mass @xmath77 .
we let @xmath694 recall the cutting procedure introduced in section [ sec : breaking - cycles - r ] , and that for an @xmath3-graph @xmath191 , we write @xmath492 for a random variable with distribution @xmath695 . for @xmath74 , if @xmath696 , let @xmath697 .
otherwise , let @xmath698 , where the cutting mechanism is run independently for each @xmath13 .
we note for later use that the mass measure on @xmath699 is almost surely concentrated on the leaves of @xmath699 , since this property holds for @xmath688 , and @xmath688 may be obtained from @xmath699 by making an almost surely finite number of identifications .
[ thm : msf - converge ] fix @xmath70 . then as @xmath62 , @xmath700 in the space @xmath95 .
write @xmath701 with the convention that @xmath702 when @xmath703 . likewise , for @xmath704 let @xmath705 .
we work in a probability space in which the convergence statements of theorem [ thm : gnp - converge ] are all almost sure . in this probability space ,
by theorem 5.19 of @xcite we have that @xmath706 is almost surely finite and that @xmath707 almost surely . by theorem [ thm : gnp - converge ] , almost surely @xmath685 is bounded away from zero for all @xmath708 .
it follows from theorem [ sec : breaking - cycles - r-1 ] that almost surely for every @xmath74 we have @xmath709 propositions [ prop : cut - close ] and [ prop : cut - dist ] then imply that we may work in a probability space in which almost surely , for every @xmath74 , @xmath710 now , for each @xmath74 , we have @xmath711 moreover , for each @xmath712 the right - hand side is bounded above by @xmath713 since @xmath706 is almost surely finite , as in the proof of theorem [ thm : gnp - converge ] we thus almost surely have that @xmath714 which combined with ( [ eq : msf - converge ] ) shows that in this space , almost surely @xmath715 in this section , we study the largest component @xmath716 of the minimum spanning forest @xmath717 obtained by partially running kruskal s algorithm , as well as its analogue @xmath718 for the random graph
. it will be useful to consider the random variable @xmath719 which is the smallest number @xmath720 such that @xmath718 is a subgraph of @xmath721 for every @xmath722 .
in other words , in the race of components , @xmath719 is the last instant where a new component takes the lead .
it follows from theorem 7 of @xcite that @xmath723 is tight , that is @xmath724 ( this result is stated in @xcite for the other erds rnyi random graph model , @xmath725 , rather than @xmath36 , but it is standard that results for the former have equivalents for the latter ; see @xcite for more details . ) in the following , if @xmath726 is a real function , we write @xmath727 if there exist positive , finite constants @xmath728 such that @xmath729 in the following lemma , we write @xmath730 for the hausdorff distance between @xmath731 and @xmath59 , seen as subspaces of @xmath59 . obviously , @xmath732 .
[ lem : cauchy - property2]for any @xmath733 and @xmath734 large enough , we have @xmath735 in the course of the proof of lemma [ lem : cauchy - property2 ] , we will need the following estimate on the length of the longest path outside the largest component of a random graph within the critical window .
[ lem : long_path - improved ] for all @xmath736 there exists @xmath734 such that for all @xmath737 and all @xmath0 sufficiently large , the probability that a connected component of @xmath738 aside from @xmath718 contains a simple path of length at least @xmath739 is at most @xmath740 .
the proof of lemma [ lem : long_path - improved ] follows precisely the same steps as the proof of lemma 3 ( b ) of @xcite , which is essentially the special case @xmath741 . , but their proof does imply the present stronger result . for the careful reader , the key point is that the last estimate in theorem 19 of @xcite is a specialisation of a more general bound , theorem 11 ( iii ) of @xcite . using the more general bound in
the proof is the only modification required to yield the above result . ] since no new idea is involved , we omit the details .
fix @xmath742 and for @xmath743 , let @xmath744 .
let @xmath745 be the smallest @xmath13 for which @xmath746 .
lemma 4 of @xcite ( proved via prim s algorithm ) states that @xmath747 } } = o(n^{1/6 } ( \log n)^{7/2});\ ] ] this is established by proving the following stronger bound , which will be useful in the sequel : @xmath748 let @xmath749 be the event that some component of @xmath750 aside from @xmath751 contains a simple path with more than @xmath752 edges and let @xmath753 lemma [ lem : long_path - improved ] entails that , for @xmath754 sufficiently large , for all @xmath0 , and all @xmath755 , @xmath756 where the last inequality holds for all @xmath13 sufficiently large . for fixed @xmath757 , if @xmath758 then for all @xmath759 $ ] we have @xmath760 if , moreover , @xmath761 , then we have @xmath762 the latter inequality holding for @xmath763 .
finally , fix @xmath764 and let @xmath765 be such that @xmath766 . since @xmath767 as @xmath2 , we certainly have @xmath768 for all @xmath0 large enough .
furthermore , @xmath769 for all @xmath125 large enough and all @xmath0 such that @xmath770 , by ( [ eqn : tdistance ] ) .
it then follows from and the tightness of @xmath723 that there exists a constant @xmath771 such that for all @xmath734 large enough , @xmath772 letting @xmath0 tend to infinity proves the lemma .
we are now in a position to prove a partial version of our main result . in what follows
, we write @xmath773 , @xmath774 and @xmath775 for the metric spaces obtained from @xmath59 , @xmath115 and @xmath120 by ignoring their measures .
[ lem : firstdraft ] there exists a random compact metric space @xmath776 such that , as @xmath62 , @xmath777 moreover , as @xmath99 , @xmath778 recall that the metric space @xmath230 is complete and separable .
theorem [ thm : msf - converge ] entails that @xmath779 as @xmath62 in @xmath780 .
the stated results then follow from this , lemma [ lem : cauchy - property2 ] and the principle of accompanying laws ( see theorem 3.1.14 of @xcite or theorem 9.1.13 in the second edition ) .
let @xmath114 be the measured metric space obtained from @xmath115 by rescaling so that the total mass is one ( in @xmath115 we gave each vertex mass @xmath77 ; now we give each vertex mass @xmath116 ) .
[ lem : mst - lim ] for any @xmath112 , @xmath781 in order to prove this proposition , we need some notation and a lemma .
let @xmath782 be the subgraph of @xmath67 with edge set @xmath783 .
then @xmath784 is a forest which we view as rooted by taking the root of a component to be the unique vertex in that component which was an element of @xmath785 . for @xmath786 ,
let @xmath787 be the number of nodes in the component @xmath788 of @xmath784 rooted at @xmath134 .
the fact that the random variables @xmath789 are exchangeable will play a key role in what follows .
[ lem : exchbitssmall ] for any @xmath790 , @xmath791 let @xmath792 be the event that vertices 1 and 2 lie in the same component of @xmath793 .
note that , conditional on @xmath794 , the event @xmath792 occurs with probability at least @xmath795 for sufficiently large @xmath0 .
so , in order to prove the lemma it suffices to show that @xmath796 in order to prove ( [ eqn:12diffcmpts ] ) , we consider the following modification of prim s algorithm .
we build the mst conditional on the collection @xmath797 of trees .
we start from the component containing vertex 1 in @xmath798 . to this component
, we add the lowest weight edge connecting it to a new vertex .
this vertex lies in a new component of @xmath798 , which we add in its entirety , before again seeking the lowest - weight edge leaving the tree we have so far constructed .
we continue in this way until we have constructed the whole mst .
( observe that the components we add along the way may , of course , be singletons . )
note that if we think of prim s algorithm as a discrete - time process , with time given by the number of vertices added so far , then this is simply a time - changed version which looks only at times when we add edges of weight _ strictly greater than @xmath799_. this is because when prim s algorithm first touches a component of @xmath797 , it necessarily adds all of its edges before adding any edges of weight exceeding @xmath800 .
for @xmath801 , write @xmath802 for the tree constructed by the modified algorithm up to step @xmath13 and let @xmath485 be the edge added at step @xmath13 .
the advantage of the modified approach is that , for each @xmath74 , we can calculate the probability that the endpoint of @xmath485 which does not lie in @xmath803 touches @xmath804 , given that it has not at steps @xmath805 .
recall that , at each stage of prim s algorithm , we add the edge of minimal weight leaving the current tree .
we are thinking of this tree as a collection of components of @xmath797 connected by edges of weight strictly greater than @xmath800 .
in general , different sections of the tree built so far are subject to different conditionings depending on the weights of the connecting edges and the order in which they were added .
in particular , the endpoint of @xmath485 contained in @xmath803 is more likely to be in a section with a lower weight - conditioning .
however , the _ other _ endpoint of @xmath485 is equally likely to be any of the vertices of @xmath806 because all that we know about them is that they lie in ( given ) components of @xmath807 .
formally , let @xmath808 .
let @xmath809 be the component containing 1 in @xmath797 .
recursively , for @xmath578 , let * @xmath485 be the smallest - weight edge leaving @xmath803 and * @xmath802 be the component containing 1 in the graph with edge - set @xmath810 .
the graph with edge - set @xmath811 is precisely @xmath67 .
let @xmath812 be the first index for which @xmath813 , so that @xmath812 is the time at which the component containing @xmath68 attaches to @xmath785 . for each @xmath578 , the endpoint of @xmath485 not in @xmath803
is uniformly distributed among all vertices of @xmath814 .
so , conditionally given @xmath815 and on @xmath816 , the probability that @xmath812 takes the value @xmath13 is @xmath817
. therefore , @xmath818 by theorem 2 of @xcite ( see also lemma 3 of @xcite ) , for all @xmath790 , @xmath819 using and , it follows that for any @xmath790 , there exists @xmath820 such that @xmath821 next , let @xmath822 be a uniformly random element of @xmath823 , and let @xmath824 be the size of the component of @xmath807 that contains @xmath822 .
theorem a1 of @xcite shows that @xmath825 } } < \infty,\ ] ] which implies that @xmath826 } } < \infty.\ ] ] for each @xmath827 , given that @xmath828 , the difference @xmath829 is stochastically dominated by @xmath824 , so that @xmath830}}\leq i{\ensuremath{\mathbb{e}\left[l_\lambda\right]}}\ , .\ ] ] by and markov s inequality , there exists @xmath831 such that @xmath832 the graph @xmath833 with edge - set @xmath834 forms part of the component containing 1 in @xmath793 ; indeed , the endpoint of @xmath835 not contained in @xmath836 is the root of this component .
write @xmath837 for this root vertex .
now consider freezing the construction of the mst via the modified version of prim s algorithm at time @xmath812 and constructing the rest of the mst using the modified version of prim s algorithm starting now from vertex 2 .
let @xmath838 .
let @xmath839 be the component containing 2 in the graph with edge - set @xmath840 .
recursively , for @xmath841 , let * @xmath842 be the smallest - weight edge leaving @xmath843 and * @xmath844 be the component containing 2 in the graph with edge - set @xmath845 .
let @xmath846 be the first index for which @xmath847 has an endpoint in @xmath848 , and let @xmath849 be the first index for which @xmath850 has an endpoint in @xmath851 .
recall that @xmath792 is the event that 1 and 2 lie in the same component of @xmath793 .
if @xmath792 occurs then we necessarily have @xmath852 . to prove ( [ eqn:12diffcmpts ] ) it therefore suffices to show that @xmath853 in order to do so , we first describe how the construction of @xmath854 conditions the locations of attachment of the edges @xmath855 . as in the introduction , for @xmath856
, @xmath857 is the weight of edge @xmath401 , and unconditionally these weights are i.i.d .
uniform@xmath32 $ ] random variables .
write @xmath858 , and for @xmath859 , let @xmath860 .
in particular , @xmath861 .
after @xmath854 is constructed , for each @xmath862 , the conditioning on edges incident to @xmath863 is as follows .
* every edge between @xmath864 and @xmath865 has weight at least @xmath866 .
* for each @xmath867 , every edge between @xmath865 and @xmath868\setminus v(c_j)$ ] has weight at least @xmath869 .
in particular , ( b ) implies all edges from @xmath865 to @xmath868\setminus v(c_{i_1})$ ] are conditioned to have weight at least @xmath870 .
this entails that components which are added later have lower weight - conditioning .
in particular , there is no conditioning on edges from @xmath861 to @xmath868\setminus v(c_{i_1})$ ] ( except the initial conditioning , that all such edges have weight at least @xmath800 , which comes from conditioning on @xmath871 ) .
it follows that under the conditioning imposed by the construction of @xmath854 , it is _ not _ the case that for @xmath872 , the endpoint of @xmath873 _ outside _
@xmath844 is uniformly distributed among @xmath874 .
however , the conditioning precisely biases these endpoints away from the sets @xmath865 with @xmath828 ( but not away from @xmath875 ) . as a consequence , for each @xmath841 , conditional on the edge set @xmath845 and on the event @xmath876 , the probability that @xmath877 is at least @xmath878 and the probability that @xmath879 is at most @xmath880 .
hence , @xmath881 and so , by ( [ eqn : luczak ] ) , we obtain that for any @xmath790 there exists @xmath882 such that @xmath883 moreover , @xmath884 note that @xmath885 . also , just as for the components @xmath802 ,
given that @xmath886 , the difference @xmath887 is stochastically dominated by @xmath824 and so we obtain the analogue of ( [ eq : mst - lim3 ] ) : there exists @xmath888 such that @xmath889 hence , from this and ( [ eq : mst - lim3 ] ) , we see that there exists @xmath890 such that @xmath891 together , ( [ eq : ibound ] ) and ( [ eq : jbound ] ) establish ( [ eq : mst - lim5 ] ) and complete the proof .
armed with this lemma , we now turn to the proof of proposition [ lem : mst - lim ] . fix @xmath112 and let @xmath892 be the minimal number of open balls of radius @xmath893 needed to cover the ( finite ) space @xmath59 .
this automatically yields a covering of @xmath894 by @xmath892 open balls of radius @xmath893 since @xmath110 is included in @xmath59 . from this covering
, we can easily construct a new covering @xmath895 of @xmath110 by sets of diameter at most @xmath896 which are pairwise disjoint .
let @xmath897 and let @xmath898 which defines a correspondence between @xmath110 and @xmath59 .
moreover , its distortion is clearly at most @xmath899 .
therefore , by lemma [ lem : cauchy - property2 ] , @xmath900 next , write @xmath901 by @xmath902 . since @xmath903 are exchangeable , theorem 16.23 of kallenberg @xcite entails that for any @xmath790 , @xmath904 as soon as we have that for all @xmath790 , @xmath905 which is precisely the content of lemma [ lem : exchbitssmall ] .
now define a measure @xmath636 on @xmath906 by @xmath907 note that @xmath908 by definition .
moreover , the marginals of @xmath636 are given by @xmath909 and @xmath910 therefore , the discrepancy @xmath911 of @xmath912 with respect to the uniform measures on @xmath59 and @xmath110 is at most @xmath913 which ( by relabelling the elements of @xmath894 so that the vertices in each @xmath914 have consecutive labels and using exchangeability ) is bounded above by @xmath915 then @xmath916 but now recall that @xmath917 is the minimal number of open balls of radius @xmath893 needed to cover @xmath59 .
let @xmath918 be the same quantity for @xmath776 . then by lemma [ lem : firstdraft ] , @xmath919 , which easily implies that @xmath920 .
in particular , by ( [ eq : distortionsmall ] ) and ( [ eq : kallenberg ] ) @xmath921 and the right - hand side converges to 0 as @xmath922 .
let @xmath923 be the measured metric space obtained from @xmath97 by renormalising the measure to be a probability .
[ thm : big_statement ] there exists a random compact measured metric space @xmath54 of total mass 1 such that as @xmath62 , @xmath924 in the space @xmath600 .
moreover , as @xmath99 , @xmath925 in the space @xmath600 . finally , writing @xmath926 , we have @xmath927 in @xmath928 , where @xmath776 is as in lemma [ lem : firstdraft ] .
recall that the metric space @xmath58 is a complete and separable .
theorem [ thm : msf - converge ] entails that @xmath929 as @xmath62 in @xmath930 .
the stated results then follow from this , proposition [ lem : mst - lim ] and the principle of accompanying laws ( see theorem 3.1.14 of @xcite or theorem 9.1.13 in the second edition ) .
finally , we observe that , analogous to the fact that @xmath110 is a subspace of @xmath59 , we can view @xmath931 as a subspace of @xmath54 .
( we emphasize that this does not follow from theorem [ thm : big_statement ] . ) to this end , we briefly introduce the marked gromov hausdorff topology of ( * ? ? ?
* section 6.4 ) .
let @xmath932 be the set of ordered pairs of the form @xmath933 , where @xmath318 is a compact metric space and @xmath329 is a compact subset of @xmath169 ( such pairs are considered up to isometries of @xmath191 ) .
a sequence @xmath934 of such pairs converges to a limit @xmath933 if there exist correspondences @xmath935 whose restrictions to @xmath936 are correspondences between @xmath937 and @xmath293 , and such that @xmath938 .
( in particular , this implies that @xmath937 converges to @xmath293 for the gromov
haudorff distance , when these spaces are equipped with the restriction of the distances on @xmath939 . )
moreover , a set @xmath940 is relatively compact if and only if @xmath941 is relatively compact for the gromov
hausdorff topology .
recall the definition of the tight sequence of random variables @xmath723 at the beginning of this section . by taking subsequences
, we may assume that we have the joint convergence in distribution @xmath942 for the product topology on @xmath943 .
on the right - hand side should , in principle , depend on @xmath125 , but obviously these spaces are almost surely all isometric .
] this coupling of course has the properties that @xmath944 and that @xmath945 for every @xmath946 . combining this with lemma [ lem : cauchy - property2 ]
we easily obtain the following .
[ sec : largest - tree - minimum ] there exists a probability space on which one may define a triple @xmath947 with the following properties : ( i ) @xmath948 is an a.s .
finite random variable ; ( ii ) @xmath944 , @xmath945 and @xmath949 for every @xmath946 ; and ( iii ) for every @xmath950 and @xmath951 large enough , @xmath952 in particular , @xmath953 as @xmath119 for the marked gromov - hausdorff topology .
in this section we give some properties of the limiting metric space @xmath954 .
we start with some general properties that @xmath954 shares with the brownian crt of aldous @xcite : [ thm : basic_properties ] @xmath54 is a measured @xmath3-tree which is almost surely binary and whose mass measure is concentrated on its leaves . by the second distributional convergence in theorem [ thm : big_statement ] , we may ( and will ) work in a space in which we almost surely have @xmath955 . since it is the gromov hausdorff limit of the sequence of @xmath3-trees @xmath956 , @xmath954 is itself an @xmath3-tree ( see for instance @xcite ) . for fixed @xmath70 , each component of @xmath107
is obtained from @xmath957 , the scaling limit of @xmath958 , using the cutting process . from the construction of @xmath957 detailed in section [ sec : scaling - limit - erdhos ] , it is clear that @xmath959 almost surely does not contain points of degree more than three , and so @xmath960 is almost surely binary .
next , let us work with the coupling @xmath961 , of proposition [ sec : largest - tree - minimum ] .
we can assume , using the last statement of this proposition and the skorokhod representation theorem , that @xmath962 a.s . in @xmath932 .
now suppose that @xmath963 has a point @xmath964 of degree at least @xmath965 with positive probability . on this event , we can find four points @xmath966 of the skeleton of @xmath963 , each having degree @xmath66 , and such that the geodesic paths from @xmath479 to @xmath964 have strictly positive lengths and meet only at @xmath964 .
but for @xmath125 large enough , @xmath967 all belong to @xmath968 , as well as the geodesic paths from @xmath969 to @xmath964 .
this contradicts the fact that @xmath931 is binary .
hence , @xmath54 is binary almost surely .
let @xmath295 and @xmath970 be sampled according to the probability measures on @xmath54 and on @xmath971 , respectively . for the remainder of the proof we abuse notation by writing @xmath972 and @xmath973 for the marked spaces ( random elements of @xmath974 ) obtained by marking at the points @xmath295 and @xmath970 .
then we may , in fact , work in a space in which almost surely @xmath975 as noted earlier , the mass measure on @xmath971 is almost surely concentrated on the leaves of @xmath971 , and it follows that for each fixed @xmath125 , @xmath970 is almost surely a leaf .
let @xmath976 \to m~\mbox{a geodesic with } x \in { \ensuremath{\mathrm{im}}}(f)\right\}\ , , \ ] ] so , in particular , @xmath977 precisely if @xmath295 is a leaf . for each fixed @xmath125 , since @xmath970 is almost surely a leaf , it is straightforward to verify that almost surely @xmath978 but then taking @xmath99 along any countable sequence shows that @xmath977 almost surely . to distinguish @xmath954 from aldous crt , we look at a natural notion of fractal dimension , the _ minkowski ( or box - counting ) dimension _
given a compact metric space @xmath191 and @xmath979 , let @xmath980 be the minimal number of open balls of radius @xmath981 needed to cover @xmath191 .
we define the lower and upper minkowski dimensions by @xmath982 if @xmath983 , then this value is called the minkowski dimension and is denoted @xmath984 .
[ prop : box - counting ] the minkowski dimension of @xmath54 exists and is equal to @xmath4 almost surely .
since the brownian crt @xmath64 satisfies @xmath985 almost surely ( ( * ? ? ?
* corollary 5.3 ) ) , we obtain the following result , which gives a negative answer to a conjecture of aldous @xcite .
[ cor : notcrt ] for any random variable @xmath986 , the laws of @xmath54 and of @xmath987 , the metric space @xmath64 with distances rescaled by @xmath65 , are mutually singular .
the proof relies on an explicit description of the components of @xmath128 , given in @xcite .
we only give a partial statement , since that is all that we need here .
note that , given @xmath988 , the kernel @xmath989 is a 3-regular multigraph with @xmath990 edges and hence @xmath991 vertices .
fix @xmath720 and @xmath992 , and @xmath405 a @xmath4-regular multigraph with @xmath990 edges .
label the edges of @xmath405 by @xmath993 arbitrarily .
construction 1 . : : independently sample random variables @xmath994 and @xmath995 . attach a line - segment of length @xmath996 in the place of edge @xmath997 in @xmath405 , for @xmath998 .
construction 2 . : : sample @xmath999 and , given @xmath1000 , let @xmath1001 be independent crt s with masses given by @xmath1002 respectively .
for @xmath1003 , let @xmath1004 be two independent points in @xmath1005 , chosen according to the normalized mass measure .
take the metric gluing of @xmath1006 induced by the graph structure of @xmath405 , by viewing @xmath1007 as the extremities of the edge @xmath13 .
here we should recall some of the basic properties of the crt @xmath64 , referring the reader to , e.g. , @xcite for more details .
if @xmath1008 is a standard normalized brownian excursion then @xmath64 is the quotient space of @xmath32 $ ] endowed with the pseudo - distance @xmath1009 , by the relation @xmath1010 .
it is seen as a measured metric space by endowing it with the mass measure which is the image of lebesgue measure on @xmath32 $ ] by the canonical projection @xmath1011\to \mathscr{t}$ ] .
it is also naturally rooted at the point @xmath1012 .
likewise , the crt with mass @xmath197 , denoted by @xmath1013 , is coded in a similar fashion by ( twice ) a brownian excursion conditioned to have duration @xmath197 . by scaling properties of brownian excursion
, this is the same as multiplying distances by @xmath1014 in @xmath64 , and multiplying the mass measure by @xmath197 .
[ sec : prop - scal - limit-1 ] the metric space obtained by construction 1 ( resp . construction 2 ) has same distribution as @xmath1015 ( resp .
@xmath1016 ) , given @xmath1017 , @xmath1018 and @xmath1019 .
the proof of proposition [ prop : box - counting ] builds on this result and requires a couple of lemmas .
recall the notation @xmath430 from section [ sec : breaking - cycles - r ] .
[ sec : prop - scal - limit-2]let @xmath1020 be a safely pointed @xmath3-graph and fix @xmath303 . then @xmath1021 .
this lemma will be proved in section [ sec : cutting - procedure ] , where we give a more precise description of @xmath430 .
the next lemma is a concentration result for the mass and surplus of @xmath1016 .
this should be seen as a continuum analogue of similar results in @xcite .
we stress that these bounds are far from being sharp , and could be much improved by a more careful analysis . in the rest of this section ,
if @xmath1022 is a family of positive random variables and @xmath1023 is a positive function , we write @xmath1024 if for all @xmath1025 , @xmath1026\right ) } } = { \mathrm{oe}}(\lambda)\ , .\ ] ] note that this only constrains the above probability for large @xmath125 .
[ sec : prop - scal - limit ] it is the case that @xmath1027 we use the construction of @xmath128 described in section [ sec : scaling - limit - erdhos ] . recall that @xmath1028 is a standard brownian motion , that @xmath1029 , and that @xmath1030 .
note that , letting @xmath1031 we have @xmath1032 . considering first @xmath1033 , by symmetry , the reflection principle and scaling we have that @xmath1034 and this is @xmath1035 since @xmath1036 is gaussian . turning to @xmath1037 , note that letting @xmath1038 , then @xmath1039 is a standard brownian motion by the markov property .
hence , on the event @xmath1040 , the probability that @xmath1041 for some @xmath1042 is at most @xmath1043 .
we deduce that @xmath1032 from the fact that @xmath1044 has an exponential distribution , see e.g. @xcite . on @xmath1045 , @xmath1046
from which it is elementary to obtain that if @xmath1047 , the following properties hold . *
the excursion @xmath1048 of @xmath1049 that straddles the time @xmath125 has length in @xmath1050 $ ] . *
all other excursions of @xmath1049 have length at most @xmath1051 . *
the area of @xmath1048 is in @xmath1052 $ ] .
note that ( i ) and ( ii ) imply that , for @xmath1053 , on @xmath1045 , the excursion @xmath1048 of @xmath1049 is the longest , which we previously called @xmath1054 , and which encodes the component @xmath1016 of @xmath128 .
this implies that @xmath1055 , since @xmath1056 is precisely the length of @xmath1054 .
finally , recall that , given @xmath1054 , @xmath1057 has a poisson distribution with parameter equal to the area of @xmath1054 . therefore , standard large deviation bounds together with ( iii ) imply that @xmath1058 . in this proof
, we always work with the coupling from proposition [ sec : largest - tree - minimum ] , but for convenience omit the decorations from the notation , e.g. , writing @xmath54 in place of @xmath776 or of @xmath963 .
in particular , this allows us to view @xmath931 as a subspace of @xmath54 for every @xmath946 .
since @xmath931 is obtained from @xmath1016 by performing the cutting operation of section [ sec : breaking - cycles - r ] , lemma [ sec : prop - scal - limit-2 ] implies that for every @xmath303 , @xmath1059 next , by viewing @xmath1015 as a graph with edge - lengths , we obtain that @xmath1060 is at least equal to the number @xmath1061 of edges of @xmath1015 that have length at least @xmath1062 , since the open balls with radius @xmath1063 centred at the midpoints of these edges are pairwise disjoint .
now fix @xmath1064 and a @xmath4-regular multigraph @xmath405 with @xmath990 edges , and recall the notation of * construction 1*. given that @xmath1065 and @xmath1066 , the edge - lengths of @xmath1015 are given by @xmath1067 , and we conclude that ( still conditionally ) @xmath1068 note that this does not depend on @xmath405 but only on @xmath197 and on @xmath184 . now
@xmath1069 can be represented as the sum of @xmath1070 independent random variables with distribution @xmath1071 , which have mean @xmath68 , and by standard large deviation results this implies that @xmath1072}{\ensuremath{\mathbb{p}\left(\gamma_k < \lambda^3\right)}}={\mathrm{oe}}(\lambda)\ , .\ ] ] hence , by first conditioning on @xmath1073 and using lemma [ sec : prop - scal - limit ] , for any given @xmath1074 , @xmath1075}}{\ensuremath{\mathbb{p}\left(n'(1/\lambda)<c\lambda^3\ , |\ , \mathrm{mass}(\mathscr{g}_\lambda^1)=\sigma , s(\mathscr{g}_\lambda^1)=k\right)}}+{\mathrm{oe}}(\lambda)\nonumber\\ & \leq \sup_{\substack{\sigma\geq \lambda\\k\in [ \lambda^3/2,\lambda^3]}}{\ensuremath{\mathbb{p}\left(|\{i\in
\{1,\ldots,3k-3\}:y_i\sqrt{\sigma\gamma_k}>2/\lambda\}|<c\lambda^3\right)}}+{\mathrm{oe}}(\lambda ) \nonumber\\ & \leq \sup_{k\in [ \lambda^3/2,\lambda^3]}{\ensuremath{\mathbb{p}\left(|\{i\in \{1,\ldots,3k-3\}:y_i>2/\lambda^3\}|<c\lambda^3\right)}}+{\mathrm{oe}}(\lambda ) \nonumber\end{aligned}\ ] ] we now use that @xmath1076 is distributed as @xmath1077 , where @xmath1078 are independent exponential@xmath1079 random variables .
standard large deviations results for gamma random variables imply that @xmath1080}{\ensuremath{\mathbb{p}\left(\gamma_1+\ldots+\gamma_{3k-3 } > 4\lambda^3\right ) } } = { \mathrm{oe}}(\lambda)\ , .\ ] ] from this we obtain @xmath1081}{\ensuremath{\mathbb{p}\left(|\{i\in \{1,\ldots,3k-3\}:y_i>2/\lambda^3\}|<c\lambda^3\right)}}}\nonumber\\ & \leq \sup_{k\in [ \lambda^3/2,\lambda^3]}{\ensuremath{\mathbb{p}\left(|\{i\in \{1,\ldots,3k-3\}:\gamma_i>8\}|<c\lambda^3\right)}}+{\mathrm{oe}}(\lambda)\nonumber\end{aligned}\ ] ] and this is @xmath1035 for @xmath1082 , since @xmath1083 is bin@xmath1084 distributed .
it follows that for such @xmath44 , @xmath1085 , which with implies that @xmath1086 we obtain by the borel cantelli lemma that @xmath1087 for all @xmath1088 sufficiently large . by sandwiching @xmath1089 between consecutive integers , this yields that almost surely @xmath1090
we now prove the upper bound from proposition [ prop : box - counting ] .
recall the definition of @xmath948 from proposition [ sec : largest - tree - minimum ] .
fix @xmath1091 and an integer @xmath1092 .
we work conditionally on the event @xmath1093 .
next , fix @xmath323 .
if @xmath1094 is a covering of @xmath931 by balls of radius @xmath1095 then , since @xmath1096 , the centres @xmath1097 of these balls are elements of @xmath54 . on the event @xmath1098
, whose complement has conditional probability @xmath1035 by proposition [ sec : largest - tree - minimum ] , the balls with centres @xmath1097 and radius @xmath1099 then form a covering of @xmath54 . hence @xmath1100 where in the penultimate step we used lemma [ sec : prop - scal - limit-2 ] and the fact that @xmath931 is obtained from @xmath1016 by performing @xmath1057 cuts , and in the last step we used the fact that @xmath1058 from lemma [ sec : prop - scal - limit ] .
to estimate @xmath1101 , we now use * construction 2 * to obtain a copy of @xmath1016 conditioned to satisfy @xmath1065 and @xmath1066 , where @xmath405 is a @xmath4-regular multigraph with @xmath990 edges . recall that we glue @xmath990 brownian crt s @xmath1102 along the edges of @xmath405 .
these crt s are conditionally independent given their masses @xmath1103 , and @xmath1000 has dirichlet@xmath1104 distribution .
( here we include the mass in the notation because it will vary later on . )
if each of these trees has diameter less than @xmath1095 , then clearly we can cover the glued space by @xmath990 balls of radius @xmath1095 , each centred in a distinct tree @xmath1105 .
therefore , by first conditioning on @xmath1056 and on @xmath1057 , and then using lemma [ sec : prop - scal - limit ] , @xmath1106 } } { \ensuremath{\mathbb{p}\left(n(\mathscr{g}_\lambda^1,1/\lambda^{1-{\ensuremath{\epsilon}}})>3\lambda^3\ , |\ , \mathrm{mass}(\mathscr{g}_\lambda^1)=\sigma , s(\mathscr{g}_\lambda^1)=k\right)}}+{\mathrm{oe}}(\lambda)\nonumber\\ & \leq \sup_{\substack{\sigma\leq 3\lambda\\k\in [ \lambda^3/2,\lambda^3 ] } } { \ensuremath{\mathbb{p}\left(\max_{1\leq i\leq 3k-3}{\mathrm{diam}}(\mathscr{t}^{(i)}_{\sigma x_i})>1/\lambda^{1-{\ensuremath{\epsilon}}}\right)}}+{\mathrm{oe}}(\lambda)\end{aligned}\ ] ] we can represent @xmath1000 as @xmath1077 , where @xmath1078 are i.i.d . random variables with distribution @xmath1107 . hence @xmath1108 standard large deviations results for gamma random variables then entail that for all @xmath112 , @xmath1072}{\ensuremath{\mathbb{p}\left(\max_{1\leq i\leq 3k-3}x_i > 1/\lambda^{3-{\ensuremath{\epsilon}}}\right)}}={\mathrm{oe}}(\lambda)\ , , \ ] ] which in turn implies that @xmath1109}}{\ensuremath{\mathbb{p}\left(\max_{1\leq i\leq 3k-3}{\mathrm{diam}}(\mathscr{t}^{(i)}_{\sigma x_i})>1/\lambda^{1-{\ensuremath{\epsilon}}}\right ) } } } \nonumber\\ & \leq \sup_{k\in [ \lambda^3/2,\lambda^3]}{\ensuremath{\mathbb{p}\left(\max_{1\leq i\leq 3k-3}x_i>1/\lambda^{3-{\ensuremath{\epsilon}}}\right ) } } + { \ensuremath{\mathbb{p}\left(\max_{1\leq i\leq 3\lambda^3}{\mathrm{diam}}(\mathscr{t}^{(i)}_{3/\lambda^{2-{\ensuremath{\epsilon}}}})>1/\lambda^{1-{\ensuremath{\epsilon}}}\right)}}\nonumber\end{aligned}\ ] ] where we used that , by scaling , @xmath1110 is stochastically increasing in @xmath197 , and @xmath1111 are independent crt s , each with mass @xmath1112 . using this bound and brownian scaling , it follows that @xmath1113}}{\ensuremath{\mathbb{p}\left(\max_{1\leq i\leq 3k-3}{\mathrm{diam}}(\mathscr{t}^{(i)}_{\sigma x_i})>1/\lambda^{1-{\ensuremath{\epsilon}}}\right ) } } \leq { \mathrm{oe}}(\lambda)+1-\left(1-{\ensuremath{\mathbb{p}\left({\mathrm{diam}}(\mathscr{t}>\lambda^{{\ensuremath{\epsilon}}/2}/\sqrt{3})\right)}}\right)^{3\lambda^3}\ , .\ ] ] next , it is well - known that the height of @xmath64 , that is , the maximal distance from the root to another point , is theta - distributed : @xmath1114 since @xmath1115 it follows that @xmath1116 we obtain that is @xmath1035 , and ( [ eq:1729 ] ) then yields that @xmath1117 .
by , we then have @xmath1118 therefore , the borel cantelli lemma implies that @xmath1119 a.s . for every integer @xmath1120 large enough
this implies that , conditionally on @xmath1093 , @xmath1121 almost surely for every @xmath323 , by sandwiching @xmath1089 between integers in @xmath1122 . since @xmath948 is almost surely finite and @xmath734 was arbitrary , this then holds unconditionally for any @xmath323 .
this concludes the proof of proposition [ prop : box - counting ] .
in this section , we investigate @xmath3-graphs and prove the structure theorems claimed in section [ sec : r - trees - r ] . in this section , @xmath318 is an @xmath3-graph .
the _ girth _ of @xmath191 is defined by @xmath1123 if @xmath159 is an @xmath3-graph , then by definition @xmath1124 is an @xmath3-tree for every @xmath270 and for some function @xmath1125 .
the balls @xmath1126 form an open cover of @xmath169 . by extracting a finite sub - cover
, we see that there exists @xmath323 such that for every @xmath270 , the space @xmath328 is an @xmath3-tree .
we let @xmath1127 be the supremum of all numbers @xmath323 with this property .
it is immediate that @xmath1128 .
in fact , it is not difficult to show that @xmath1129 and that @xmath1130 is an @xmath3-tree .
more precisely , the closed ball @xmath1131 is also a ( compact ) @xmath3-tree , since it is the closure of the corresponding open ball .
these facts are not absolutely crucial in the arguments to come , but they make some proofs more elegant , so we will take them for granted and leave their proofs to the reader , who is also referred to proposition 2.2.15 of @xcite .
[ sec : r - graphs ] if @xmath279,x)$ ] is a local geodesic in @xmath191 , then for every @xmath1132 $ ] , the restriction of @xmath280 to @xmath1133\cap[a , b]$ ] is a geodesic .
in particular , if @xmath44 is an embedded cycle and @xmath1134 , then @xmath44 contains a geodesic arc of length @xmath1135 with mid - point @xmath295 .
the function @xmath280 is injective on any interval of length at most @xmath1135 , since otherwise we could exhibit an embedded cycle with length at most @xmath1136 . in particular
, @xmath280 is injective on the interval @xmath1133\cap[a , b]$ ] , and takes values in the @xmath3-tree @xmath1137 , so that its image is a geodesic segment , and since @xmath280 is parameterized by arc - length , its restriction to the above interval is an isometry .
this proves the first statement .
for the second statement , note that every injective path @xmath1138,x)$ ] parameterized by arc - length is a local geodesic since , for every @xmath307 , the path @xmath280 restricted to @xmath1133\cap [ a , b]$ ] is an injective path in the @xmath3-tree @xmath1137 parameterized by arc - length , and hence is a geodesic . if now @xmath1139 is an injective continuous function inducing the embedded cycle @xmath44 , it suffices to apply the previous claim to a parametrisation by arc - length mapping @xmath148 to @xmath295 of the function @xmath1140 . in this section
, @xmath318 is again an @xmath3-graph .
recall that @xmath346 is the union of all arcs with endpoints in embedded cycles .
[ sec : skeleton - core - kernel ] the set @xmath346 is a finite union of embedded cycles and simple arcs that are disjoint from the embedded cycles except at their endpoints .
moreover , the space @xmath347 is an @xmath3-graph with no leaves .
assume , for a contradiction , that the union of all embedded cycles can not be written as a finite union of embedded cycles .
then we can find an infinite sequence @xmath1141 of embedded cycles such that @xmath1142 is non - empty for every @xmath1143 , and thus contains at least one point @xmath479 .
up to taking subsequences , one can assume that @xmath479 converges to some point @xmath295 , and that @xmath1144 for every @xmath708 .
let @xmath1145 be a geodesic from @xmath295 to @xmath479 : this geodesic takes its values in the @xmath3-tree @xmath1146 . since @xmath1147 , by proposition [ sec : r - graphs ] we can find two geodesic paths starting from @xmath479 , meeting only at @xmath479 , with length @xmath1148 , and taking values in @xmath1149 .
at least one of these paths @xmath1150 does not pass through @xmath295 , and so the concatenation @xmath1151 of @xmath1145 and @xmath1152 is an injective path parameterized by arc - length starting from @xmath295 and with length @xmath1127 .
so it is , in fact , a geodesic path , since it takes its values in @xmath1146 .
we let @xmath506 be the endpoint of @xmath1151 , so that @xmath1153 for every @xmath708 .
now , we observe that if @xmath1154 , the paths @xmath1151 and @xmath1155 both start from the same point @xmath295 , but since @xmath1152 takes values in @xmath1156 , since @xmath1157 passes through @xmath1158 , and since @xmath1159 , these paths are disjoint outside the ball @xmath1160 .
this implies that @xmath1161 for every @xmath1154 , and contradicts the compactness of @xmath169 .
therefore , the union @xmath1162 of all embedded cycles is closed and has finitely many connected components . by definition
, @xmath346 is the union of @xmath1162 together with all simple arcs with endpoints in @xmath1162 . obviously , in this definition
, we can restrict our attention to simple arcs having only their endpoints in @xmath1162 .
so let @xmath1163 with @xmath1164 be linked by an simple arc @xmath65 taking its values outside @xmath1162 , except at its endpoints .
necessarily , @xmath295 and @xmath296 must be in disjoint connected components of @xmath1162 , because otherwise there would exist a path from @xmath295 to @xmath296 in @xmath1162 whose concatenation with @xmath1165 would create an embedded cycle not included in @xmath1162 . furthermore , there can exist at most one arc @xmath65 , or else it would be easy to exhibit an embedded cycle not included in @xmath1162 .
so we see that @xmath346 is a finite union of simple arcs and embedded cycles , which is obviously connected , and is thus closed .
any point in @xmath346 has degree at least @xmath66 by definition .
it remains to check that the intrinsic metric on @xmath346 is given by @xmath360 itself .
let @xmath1166 and @xmath1165 be a geodesic path from @xmath295 to @xmath296 .
assume that @xmath1165 takes some value @xmath1167 outside @xmath346 .
let @xmath1168 and @xmath1169 , so that @xmath1170 .
since @xmath346 is connected , we can join @xmath1171 and @xmath1172 by a simple arc included in @xmath346 , and the union of this arc with @xmath1173 is an embedded cycle not contained in @xmath346 , a contradiction . in this section , @xmath191 is an @xmath3-graph with no leaves .
we now start to prove theorem [ sec : structure - r - graphs-4 ] on the structure of such @xmath3-graphs .
the set @xmath1174 of branchpoints of @xmath169 forms the vertex set of @xmath387 .
[ sec : structure - r - graphs-1 ] the set @xmath338 is finite , and @xmath1175 for every @xmath1176 . by proposition
[ sec : skeleton - core - kernel ] , the number of cycles of @xmath191 is finite .
we assume that @xmath191 is not acyclic , and argue by induction on the maximal number of independent embedded cycles , that is , of embedded cycles @xmath1177 such that @xmath1178 for @xmath588 . plainly , we may and will assume that for all @xmath1179 , either @xmath1156 is disjoint from @xmath1180 , or @xmath1181 is a simple arc of the form @xmath1182 , where @xmath1183\to x$ ] satisfies @xmath1184 .
the result is trivial if @xmath191 is unicyclic ( @xmath1185 ) .
suppose @xmath169 has @xmath184 independent embedded cycles @xmath1177 as above .
consider the smallest connected subset @xmath170 of @xmath169 containing @xmath1186 : this subset is the union of @xmath1186 with some simple arcs having only their endpoints as elements of @xmath1187 , and is a closed subset of @xmath169 .
if @xmath1188 does not intersect @xmath170 , then there exists a unique simple arc with one endpoint @xmath277 in @xmath1188 and the other endpoint @xmath278 in @xmath170 , and disjoint from @xmath1189 elsewhere . then @xmath1190 must be elements of @xmath338 : @xmath277 is the only element of @xmath338 in @xmath1188 , we have @xmath1191 , and @xmath1192 .
therefore , the number of points in @xmath338 is at most @xmath1193 , where @xmath223 is the set @xmath170 endowed with the intrinsic metric inherited from @xmath191 .
this is an @xmath3-graph without leaves and with ( at most ) @xmath1194 independent cycles . if on the other hand @xmath1195 , then by assumption we have @xmath1196 , where @xmath65 is a sub - arc of @xmath1188 disjoint from @xmath170 except at its endpoints @xmath1190 .
the latter are elements of @xmath338 , and satisfy @xmath1197 and similarly for @xmath278 ( note that @xmath1190 may be equal ) .
after we remove @xmath1198 from @xmath169 , we are left with an @xmath3-graph @xmath223 ( in the induced metric ) without leaves , and with at most @xmath1194 independent cycles .
the result follows by induction on @xmath184 .
if @xmath191 is unicyclic , then @xmath169 is in fact identical to its unique embedded cycle @xmath44 . in this case , @xmath1199 , and we let @xmath1200 .
if @xmath191 has at least two distinct embedded cycles , then the previous proof entails that @xmath1201 , and more precisely that every embedded cycle contains at least one point of @xmath338 .
the set @xmath1202 has finitely many connected components ( in fact , there are precisely @xmath1203 components , as the reader is invited to verify ) , which are simple arcs of the form @xmath1182 , where @xmath1183\to x$ ] is such that @xmath1165 is injective on @xmath1204 , such that @xmath1205 , and such that @xmath1206 .
we let @xmath466 be the set of the closures of these connected components , i.e. the arcs @xmath1207)$ ] with the above notation , which are called the _
kernel edges_. the multigraph @xmath386 is the _ kernel _ of @xmath191 , where the vertices incident to @xmath1208 are , of course , the endpoints of @xmath401 .
an orientation of the edge @xmath401 is the choice of a parametrisation @xmath1183\to x$ ] of the arc @xmath401 or its reversal @xmath1209 , considered up to reparametrisations by increasing bijections from @xmath32 $ ] to @xmath32 $ ] .
if @xmath401 is given an orientation , then its endpoints are distinguished as the source and target vertices , and are denoted by @xmath1210 , respectively .
the next proposition then follows from the definition of @xmath338 .
[ sec : structure - r - graphs-3 ] the kernel of a non - unicyclic @xmath3-graph without leaves is a multigraph of minimum degree at least 3 . finally , we prove theorem [ sec : structure - r - graphs-4 ] .
assume that @xmath191 is a non - unicyclic @xmath3-graph without leaves , and let @xmath1211 be the lengths of the kernel edges .
note that if @xmath1212 , then @xmath1213 where @xmath1214 is a chain from @xmath295 to @xmath296 if it is possible to orient @xmath1215 in such a way that @xmath1216 and @xmath1217 for every @xmath1218 .
of course , it suffices to restrict the infimum to those chains that are simple , in the sense that they do not visit the same vertex twice .
since there are finitely many simple chains , the above infimum is , in fact , a minimum .
next , if @xmath295 and @xmath296 are elements of @xmath401 and @xmath1219 respectively , consider an arbitrary orientation of @xmath1220 .
then a shortest path from @xmath295 to @xmath296 either stays in @xmath401 ( in this case @xmath1221 ) , or passes through at least one element of @xmath338 incident to @xmath401 , and likewise for @xmath1219 .
therefore , @xmath1222 where we let @xmath1223 be the length of the arc of @xmath401 between @xmath277 and @xmath278 if @xmath1224 , and @xmath1225 otherwise .
it is shown in ( * ? ? ?
* section 3 ) that this formula gives the distance for the metric gluing of the graph with edge - lengths @xmath447 .
this proves theorem [ sec : structure - r - graphs-4 ] . in this section
, we show that kernels of @xmath3-graphs are stable under small perturbations in the gromov
hausdorff metric , under an assumption which says , essentially , that the girth is uniformly bounded away from @xmath148 .
recall from section [ sec : breaking - cycles - r ] that @xmath445 is the set of measured @xmath3-graphs @xmath191 such that @xmath1226 where it is understood in this definition that the unicyclic @xmath3-graphs ( those with surplus @xmath68 ) are such that their unique embedded cycle has length in @xmath449 $ ] .
it follows that the sets @xmath1227 , are decreasing , with union the set of all measured @xmath3-graphs .
if @xmath160 $ ] is an @xmath3-graph , we write @xmath160\in { \mathcal{a}}_r$ ] if @xmath1228\in { \mathcal{a}}_r$ ] .
note that an element @xmath1229 has @xmath1230 .
likewise , we let @xmath1231 be the set of ( isometry equivalence classes of ) safely pointed measured @xmath3-graphs @xmath1232 with @xmath451 ( recall definition [ sec : cutting - cycles - an ] ) , and say that a pointed @xmath3-graph @xmath1233 if @xmath1234 . a subset @xmath65 of @xmath169 is said to be in correspondence with a subset @xmath1235 of @xmath170 via @xmath1236 if @xmath1237 is a correspondence between @xmath65 and @xmath1235 .
let @xmath191 and @xmath223 be @xmath3-graphs with surplus at least @xmath66 .
given @xmath1238 , for @xmath112 we say that @xmath165 is a _
( of @xmath191 and @xmath223 ) if @xmath1239 , and there exists a multigraph isomorphism @xmath1240 between @xmath387 and @xmath1241 such that : 1 . for every @xmath1242 , @xmath1243 .
2 . for every @xmath1208 , the edges @xmath401 and @xmath1244 are in correspondence via @xmath165 , and @xmath1245 if @xmath1246 , an @xmath327-overlay is a correspondence with distortion at most @xmath327 , such that the unique embedded cycles @xmath1247 of @xmath191 and @xmath223 are in correspondence via @xmath165 , and @xmath1248
finally , if @xmath1249 then an @xmath327-overlay is just a correspondence of distortion at most @xmath327 .
[ lem : deltaover ] fix @xmath1250 . for every @xmath323 there exists @xmath1251 such that if @xmath318 and @xmath1252 are elements of @xmath1253 and @xmath1254 has @xmath1255
, then there exists an @xmath327-overlay @xmath1256 with @xmath1257 .
we say that a sequence of graphs with edge - lengths @xmath1258 converges to the graph with edge - lengths @xmath379 if @xmath1259 and @xmath380 are isomorphic for all but finitely many @xmath1260 , through an isomorphism @xmath1261 such that @xmath1262 as @xmath2 for every @xmath398 .
we now state some consequences of proposition [ lem : deltaover ] which are used in the proof of theorem [ thm : gnp - converge ] and in section [ sec : cuttingrgs ] , before proceeding to the proof of proposition [ lem : deltaover ] .
recall the definition of the distances @xmath1263 from section [ sec : notions - convergence ] .
[ sec : cutt - safely - point-1 ] fix @xmath1264 .
let @xmath1265 and @xmath204 be elements of @xmath445 .
suppose that @xmath1266 as @xmath2 .
+ ( i ) then @xmath1267 converges to @xmath387 as a graph with edge - lengths . as a consequence , @xmath1268 , and writing @xmath1269 ( resp .
@xmath435 ) for the restriction of the length measure of @xmath1270 ( resp . @xmath191 ) to @xmath1271 ( resp .
@xmath366 ) , it holds that @xmath1272 \(ii ) let @xmath1273 be a random variable in @xmath1274 with distribution @xmath1275 and @xmath295 be a random variable in @xmath191 with distribution @xmath437 . then as @xmath62 , @xmath1276 in the space @xmath1277 .
the above results rely on the following lemma . given metric spaces @xmath159 and @xmath177 , @xmath1236 and @xmath303 , let @xmath1278 @xmath1279 is the @xmath981-enlargement of @xmath165 with respect to the product distance .
note that if @xmath165 is a correspondence between @xmath169 and @xmath170 , then @xmath1279 is also a correspondence for every @xmath303 .
moreover , @xmath1280 .
a mapping @xmath1281\to [ a',b']$ ] is called _ bi - lipschitz _
if @xmath1282 is a bijection such that @xmath1282 and @xmath1283 are lipschitz , and we call the quantity @xmath1284\big\}\ ] ] the _ bi - lipschitz constant _ of @xmath1282 . by convention , we let @xmath1285 if @xmath1282 is not a bijection , or not bi - lipschitz .
[ sec : cutting - r - graphs ] fix @xmath444 and let @xmath1286 .
suppose there exists a correspondence @xmath165 between @xmath169 and @xmath170 such that @xmath1287 .
let @xmath1288 be two distinct points in @xmath169 , and let @xmath280 be a local geodesic from @xmath295 to @xmath296 .
let @xmath1289 be such that @xmath1290 .
then there exists a local geodesic @xmath1291 from @xmath1292 to @xmath378 with @xmath1293 and a bi - lipschitz mapping @xmath1294\to [ 0,{\ensuremath{\mathrm{len}}}(f')]$ ] such that @xmath1295 for every @xmath1296 $ ] , and @xmath1297 note that the second part of the statement also implies a lower bound on the length of @xmath1291 , namely , @xmath1298 which is , of course , useless when @xmath1299 .
let us first assume that @xmath1300 , so in particular @xmath1301 and @xmath280 is the geodesic from @xmath295 to @xmath296 .
we have @xmath1302 so that @xmath1292 and @xmath378 are linked by a unique geodesic @xmath1291 .
set @xmath1303 for @xmath1304 . from the preceding chain of inequalities
, we obtain that @xmath1305 fix @xmath1306 and let @xmath1307 be such that @xmath1308 .
then @xmath1309 , so that @xmath1307 belongs to the @xmath3-tree @xmath1310 .
let @xmath377 be the ( unique ) point of @xmath1311 that is closest to @xmath1307 .
then a path from @xmath1292 or @xmath378 to @xmath1307 must pass through @xmath377 , from which we have @xmath1312 therefore , @xmath1313 so after a short calculation we get that @xmath1314 from this we obtain @xmath1315 so that , in conjunction with , we have @xmath1316 we next assume that @xmath1317 .
fix an integer @xmath1318 such that @xmath1319 and let @xmath1320 and @xmath1321 for @xmath1322 . by proposition
[ sec : r - graphs ] , since @xmath280 is a local geodesic and @xmath1323 for every @xmath1324 , the restriction @xmath1325}$ ] must be a shortest path and so @xmath1326\ , .\ ] ] letting @xmath1327 be a point such that @xmath1328 ( where we always make the choice @xmath1329 and @xmath1330 ) , we have @xmath1331 , so that we can consider the unique geodesic @xmath1332 between @xmath1327 and @xmath1333 . the concatenation of the paths @xmath1334 is not necessarily a local geodesic , but by excising certain parts of it we will be able to recover a local geodesic between @xmath1292 and @xmath378 . for each @xmath1335 ,
the sets @xmath1336 and @xmath1337 are included in the @xmath3-tree @xmath1338 , and the concatenation of @xmath1339 and @xmath1332 is a path from @xmath1340 to @xmath1333 which , as such , must contain the image of the geodesic @xmath1341 between these points .
let @xmath1342 be the unique point of @xmath1343 that is closest to @xmath1327 , and let @xmath1344 .
then @xmath1345 so that , for @xmath1346 , @xmath1347 if @xmath1348 then @xmath1349 however , since @xmath1350 and @xmath1351 , we have @xmath1352 so , in fact , @xmath1353 and , in particular , @xmath1354 does not lie on the shortest path between @xmath1355 and @xmath576 . from this
it follows that if @xmath1291 denotes the concatenation of the geodesic @xmath1356 between @xmath1342 and @xmath1357 , for @xmath1358 , then @xmath1291 is a local geodesic between @xmath1292 and @xmath378 .
its length is certainly bounded by the sum of the lengths of the paths @xmath1359 , so that @xmath1360 as claimed .
next , we let @xmath1361 , so that @xmath1362 if @xmath1294\to [ 0,{\ensuremath{\mathrm{len}}}(f')]$ ] is the concatenation of the mappings @xmath1363 , then @xmath1282 is bi - lipschitz with the same upper - bound for @xmath1364 as for each @xmath1365 . finally , we note that @xmath1366 is the concatenation of the paths @xmath1367 .
if @xmath1368 , we let @xmath1307 be such that @xmath1369 and let @xmath377 be the point in @xmath1370 that is closest to @xmath1307 .
then similar arguments to before entail that @xmath1371 , so that @xmath1372 which implies that @xmath1373 and we conclude that @xmath1374 , and so that @xmath1375 for every @xmath1376 $ ] .
we will prove this result only when one of @xmath191 and @xmath223 ( and then , in fact , both ) has surplus at least @xmath66 , leaving the similar and simpler case of surplus @xmath68 to the reader ( the case of surplus @xmath148 is trivial ) .
also , we may assume without loss of generality that @xmath1377 .
fix @xmath1378 , and fix any @xmath1379 .
also , fix @xmath1380 and a correspondence @xmath1381 with @xmath1382 .
list the elements of @xmath338 as @xmath1383 , and fix elements @xmath1384 of @xmath170 with @xmath1385 for each @xmath1386 . since @xmath1387 and @xmath1383 are pairwise at distance at least @xmath981 , @xmath1384 are pairwise at distance at least @xmath1388 and , in particular , are all distinct .
next , for every @xmath1208 , say with @xmath1389 , fix a local geodesic @xmath1390 between @xmath1391 and @xmath1392 with @xmath1393 and @xmath1394 . by lemma
[ sec : cutting - r - graphs ] , there exists a geodesic @xmath1395 from @xmath1396 to @xmath1397 and a bi - lipschitz mapping @xmath1398\to [ 0,{\ensuremath{\mathrm{len}}}(f''_e)]$ ] with @xmath1399 and such that @xmath1400 for every @xmath1401 $ ] . in particular , it follows that @xmath1402
. then we claim that for @xmath585 small enough , the following two properties hold . 1 . for every @xmath1208
, the path @xmath1403 is injective .
2 . for @xmath1404 with @xmath1405
, we have @xmath1406 to establish the first property , suppose that @xmath1407 for some @xmath1408 and distinct @xmath1409 $ ] . for concreteness ,
let us assume that @xmath1410 and @xmath1411 . since @xmath1395 is a local geodesic , this implies that @xmath1412 .
moreover , since @xmath1413 and since @xmath1414 , we have @xmath1415 on the other hand , we have @xmath1416 and since @xmath1417 and @xmath1418 , @xmath1419 .
but if @xmath1420 $ ] , then @xmath1421 , because a path from @xmath1422 to @xmath1423 is either a subarc of @xmath401 , or passes through both vertices @xmath1391 and @xmath1392 .
it follows that @xmath1424 , in contradiction with .
this yields that property 1 holds .
the argument for property 2 is similar : for every @xmath1425 and @xmath1426 , there exist @xmath1427 and @xmath1428 such that @xmath1429 and @xmath1430 are in @xmath1431 .
then the distance from @xmath1432 to @xmath338 is at least @xmath1433 so that @xmath1434 . from this
, we deduce that @xmath1435 .
next , for every @xmath1436 , consider the points @xmath1437 for which @xmath1410 , as well as the points @xmath1438 for which @xmath1439 .
these points are on the boundary of the ball @xmath1440 , which we recall is an @xmath3-tree .
let @xmath1441 be the subtree of @xmath1440 spanned by these points .
then property 1 above shows that @xmath1442 induces a closed subgraph of @xmath177 without leaves , and so this subgraph is in fact a subgraph of @xmath1443 .
furthermore , property 2 implies that the points of degree at least @xmath4 in this subgraph can only belong to @xmath1444 .
since any such point is then an element of @xmath1445 and @xmath1446 , we see that each @xmath1441 can contain at most one element of @xmath1445 . on the other hand
, each @xmath1441 must contain at least one element of @xmath1445 because @xmath1441 has at least three leaves ( since @xmath1391 has degree at least @xmath4 ) .
thus , each @xmath1441 contains exactly one element of @xmath1445 , which we denote by @xmath1447 .
next , for @xmath1448 , if @xmath1449 , then we let @xmath1450 be the simple path from @xmath1451 to @xmath1452 that has a non - empty intersection with @xmath1395 .
it is clear that this path is well - defined and unique . letting @xmath1453 for @xmath1454 , and letting @xmath1455 for @xmath1456
, we have therefore defined a multigraph homomorphism from @xmath387 to @xmath1241 , and this homomorphism is clearly injective . by symmetry of the roles of @xmath169 and @xmath170 , we see that @xmath1457 and @xmath1458 , and so @xmath1240 must , in fact , be a multigraph isomorphism . finally , since @xmath1459 , we have @xmath1460 by our choice of @xmath585 .
but , by construction , @xmath1461 , since the endpoints of @xmath1244 each have distance at most @xmath893 from an endpoint of @xmath1462 .
it follows that @xmath1463 .
finally , since every point of @xmath1244 is within distance @xmath893 of @xmath1464 and @xmath1465 and @xmath1466 are in correspondence via @xmath1431 , it follows that @xmath401 and @xmath1244 are in correspondence via @xmath1467 .
since @xmath1468 , this completes the proof .
again we only consider the case @xmath1469 , the case @xmath393 being easier ( and the case @xmath392 trivial ) .
let @xmath1470 and @xmath191 be as in the statement of corollary [ sec : cutt - safely - point-1 ] .
let @xmath1471 and @xmath1472 be sequences of correspondences and of measures , respectively , such that @xmath1473 and @xmath1474 each converge to @xmath148 as @xmath2 .
the fact that @xmath1267 converges to @xmath387 as a graph with edge - lengths is then an immediate consequence of proposition [ lem : deltaover ] : for each @xmath0 sufficiently large , simply replace @xmath632 by @xmath1475 , where @xmath1476 is an @xmath1477-overlay of @xmath191 and @xmath1270 , for some sequence @xmath1478 ( we may assume @xmath1479 ) .
we continue to write @xmath632 instead of @xmath1475 , and note that enlarging @xmath632 diminishes @xmath1480 .
in particular , we obtain that for all large enough @xmath0 , there is an isomorphism @xmath1261 from @xmath1481 to @xmath1482 such that @xmath1483 converges to @xmath148 for every @xmath1208 . the fact that @xmath1268 is immediate .
we now fix a particular orientation of the edges , and view @xmath1261 as an isomorphism of oriented graphs , in the sense that @xmath1484 . for each @xmath1456 , let @xmath1390 be a local geodesic between @xmath1485 and @xmath1486 with @xmath1394 and @xmath1487 and , for each @xmath1488 and @xmath1489 , define @xmath1390 accordingly . then for each @xmath0 sufficiently large , define a mapping @xmath1490 with domain @xmath1491)$ ] by setting @xmath1492 for each @xmath1456 and each @xmath1493 . by considering a small enlargement of @xmath632 , or , equivalently , by letting @xmath1477 tend to zero sufficiently slowly
, we may assume without loss of generality that @xmath1494 for all @xmath1495 .
this comes from the fact that @xmath401 and @xmath1496 are in correspondence via @xmath632 ; we leave the details of this verification to the reader .
it follows that the relation @xmath1497 is a subset of @xmath632 .
let @xmath1498 be the set of edges @xmath1456 whose removal from @xmath466 does not disconnect @xmath387 .
clearly , @xmath1499 , and the measure @xmath435 is carried by @xmath1500 ( in fact , it is carried by the subset of points of @xmath1501 with degree @xmath66 , by proposition [ sec : skeleton - core - kernel-2 ] ) .
let @xmath1502 be the restriction of @xmath435 to the set @xmath1503 , which has total mass @xmath1504 .
we consider the push - forward @xmath1505 of @xmath1502 by the mapping @xmath1506 from @xmath169 to @xmath1507 .
then the second marginal of @xmath1505 is the restriction of @xmath1269 to @xmath1508 , so that @xmath1509 the latter converges to @xmath148 by the convergence of the edge - lengths .
it only remains to note that @xmath1510 by construction .
this yields ( i ) .
finally , ( i ) implies that @xmath1511 converges to @xmath1512 in the metric @xmath1513 introduced in section [ sec : notions - convergence ] , and ( ii ) then follows from proposition [ sec : grom - hausd - prokh ] .
in this section , we will consider a simple cutting procedure on @xmath3-graphs , and study how this procedure is perturbed by small variations in the gromov hausdorff distance .
let @xmath159 be an @xmath3-graph , and let @xmath369 .
we endow the connected set @xmath1514 with the intrinsic distance @xmath1515 : more precisely , @xmath425 is defined to be the minimal length of an injective path not visiting @xmath295 .
this is indeed a minimum because there are finitely many injective paths between @xmath296 and @xmath374 in @xmath169 , as a simple consequence of theorem [ sec : structure - r - graphs-4 ] applied to @xmath1516 .
the space @xmath1517 is not complete , so we let @xmath422 be its metric completion as in section [ sec : breaking - cycles - r ] .
this space is connected , and thus easily seen to be an @xmath3-graph .
we call it the _ @xmath3-graph @xmath159 cut at the point @xmath295_. from now on
, we will further assume that @xmath368 , so that @xmath185 is safely pointed as in definition [ sec : cutting - cycles - an ] . in this case , one can provide a more detailed description of @xmath422 .
a cauchy sequence @xmath1518 in @xmath423 is also a cauchy sequence in @xmath159 , since @xmath1519 . if its limit @xmath296 in @xmath159 is distinct from @xmath295 , then it is easy to see that @xmath1520 , by considering a ball @xmath1521 not containing @xmath295 within which @xmath1522 .
so let us assume that @xmath1518 converges to @xmath295 for the distance @xmath360 . since @xmath295 has degree @xmath66 , the @xmath3-tree @xmath1523 has exactly two components , say @xmath1524 .
it is clear that @xmath1525 for every @xmath1526 .
since @xmath1518 is a cauchy sequence for @xmath1517 , we conclude that it must eventually take all its values in precisely one of @xmath1527 and @xmath1528 , let us say @xmath1527 for definiteness .
note that the restrictions of @xmath360 and @xmath1515 to @xmath1527 are equal , so that if @xmath1529 is another cauchy sequence in @xmath423 which converges in @xmath159 to @xmath295 and takes all but a finite number of values in @xmath1527 , then @xmath1530 , and so this sequence is equivalent to @xmath1518 .
we conclude that the completion of @xmath1531 adds exactly two points to @xmath321 , corresponding to classes of cauchy sequences converging to @xmath295 in @xmath159 `` from one side '' of @xmath295 .
so we can write @xmath1532 and describe @xmath1533 as follows : * if @xmath1534 then @xmath1535 is the minimal length of a path from @xmath296 to @xmath374 in @xmath169 not visiting @xmath295 . * if @xmath1536 then @xmath1537 is the minimal length of an injective path from @xmath295 to @xmath296 in @xmath169 which takes its values in the component @xmath1527 on some small initial interval @xmath1538 , and similarly for @xmath1539 with @xmath1540 . * finally , @xmath1541 is the minimal length of an embedded cycle passing through @xmath295 .
if @xmath1232 is a pointed measured metric space such that @xmath185 is a safely pointed @xmath3-graph , and @xmath427 , then the space @xmath422 carries a natural measure @xmath209 , such that @xmath1542 and , for any open subset @xmath1543 not containing @xmath1544 and @xmath1545 , @xmath1546 if on the right - hand side we view @xmath65 as an open subset of @xmath169 .
consequently , there is little risk of ambiguity in using the notation @xmath15 instead of @xmath209 for this induced measure .
we finish this section by proving lemma [ sec : prop - scal - limit-2 ] on the number of balls required to cover the cut space .
let @xmath1547 be a covering of @xmath191 by open balls of radius @xmath303 , centred at @xmath1097 respectively . by definition ,
any point of @xmath169 can be joined to the centre of some ball @xmath749 by a geodesic path of length @xmath1548 .
if such a path does not pass through @xmath295 , then it is also a geodesic path in @xmath430 .
now since @xmath1094 is a covering of @xmath191 , this implies that any point @xmath296 in @xmath169 can either * be joined to some point @xmath479 by a path of length @xmath1548 that does not pass through @xmath295 , * or can be joined to @xmath295 through a path @xmath1165 of length @xmath1548 . in the first case , this means that @xmath296 belongs to the ball with centre @xmath479 and radius @xmath981 in @xmath430 . in the second case ,
depending on whether the initial segment of @xmath1165 belongs to @xmath1527 or @xmath1528 , this means that @xmath296 belongs to the ball with centre @xmath1544 or @xmath1545 with radius @xmath981 in @xmath430 .
this yields a covering of @xmath430 with at most @xmath1549 balls , as desired .
conversely , it is clear that if @xmath1318 balls are sufficient to cover @xmath430 then the same is true of @xmath191 , because distances are smaller in @xmath191 than in @xmath430 ( if @xmath295 is identified with the points @xmath1550 ) .
the following statement will be used in conjunction with corollary [ sec : cutt - safely - point-1 ] ( ii ) . recall the definition of @xmath1231 from the start of section [ sec : convergence - core ] .
[ thm : safelypointed ] fix @xmath1551 . let @xmath1552 and @xmath1232 be elements of @xmath1231 .
suppose that @xmath1553 and that @xmath1554 for every @xmath0 .
then @xmath1555 our proof of theorem [ thm : safelypointed ] hinges on two lemmas ; to state these lemmas we require a few additional definitions .
let @xmath1556 and recall the definition of the projection @xmath1557 . for @xmath323 , write @xmath1558 so that @xmath1559 .
the sets @xmath1560 decrease to the singleton @xmath1561 as @xmath1562 , because @xmath368 .
consequently , @xmath1563 converges to @xmath148 as @xmath1562 . for @xmath323
sufficiently small , the set @xmath1564 , endowed with the intrinsic metric , is an @xmath3-graph .
in fact , it is easy to see that for @xmath1565 , this intrinsic metric is just the restriction of @xmath1533 to @xmath1566 .
-graph : @xmath346 is in thicker line . ] let us assume that @xmath1567 .
let @xmath1568 be the two points of @xmath346 at distance @xmath327 from @xmath295 , labelled in such a way that , in the notation of section [ sec : cutting - procedure ] , @xmath1569 is the point closest to @xmath1544 in @xmath1570 , or in other words , such that @xmath1571 . for @xmath1572 let @xmath19 be the geodesic arc between @xmath1573 and @xmath295 in @xmath169 .
we let @xmath1574 , which we see as a subset of @xmath1570 .
see figure [ fig : notationcut ] for an illustration .
now let @xmath1575 . just as we defined the space @xmath1576 , we define the space @xmath1577 , with @xmath1578 .
we likewise define the sets @xmath1579 and @xmath1580 and the points @xmath1581 for @xmath1582 as above .
we will use the same notation @xmath153 for the projection @xmath1583 .
[ sec : stab - cutt - proc ] fix @xmath790 . if @xmath165 is a @xmath585-overlay of @xmath191 and @xmath223 then for every @xmath1584 , we have @xmath1585 .
let @xmath1586 be such that @xmath1587 . since @xmath165 is a @xmath585-overlay , @xmath1588 .
in particular , if @xmath1589 then we have @xmath1590 . otherwise , a geodesic from @xmath378 to @xmath1586 must pass through @xmath1591 and @xmath1592 , so that @xmath1593 on the other hand , since @xmath165 is an @xmath585-overlay , we know that @xmath346 and @xmath1443 are in correspondence via @xmath165 , which implies that @xmath1594 so that @xmath1595 . in all cases , we have @xmath1585 , as claimed . [
sec : stab - cutt - proc-2 ] fix @xmath1551 . for all @xmath112 there
exists @xmath1596 such that if @xmath1597 then @xmath1598 since @xmath1597 we can find @xmath1599 with @xmath1600 and with @xmath1601 , and a measure @xmath582 with @xmath1602 and @xmath1603 .
fix @xmath1604 such that @xmath1605 and @xmath1606 . by choosing @xmath1607 sufficiently small
, it follows by proposition [ lem : deltaover ] that there exists a @xmath585-overlay @xmath165 of @xmath191 and @xmath223 with @xmath1608 , so in particular @xmath1601 and @xmath1609 .
we also remark that @xmath1610 .
we next modify @xmath165 to give a correspondence between @xmath1566 and @xmath1611 by letting @xmath1612 where for @xmath1572 , we define @xmath1613 to verify that @xmath1614 is indeed a correspondence between @xmath1566 and @xmath1611 , it suffices to check that there does not exist @xmath1615 for which @xmath1616 , and similarly that there does not exist @xmath1617 for which @xmath1618 . in the first case
this is immediate since @xmath1619 , so for all @xmath1620 with @xmath1621 we have @xmath1622 .
a symmetric argument handles the second case .
we next estimate the distortion of @xmath1614 when @xmath1566 and @xmath1611 are endowed with the metrics @xmath1533 and @xmath1623 respectively . to this end , let @xmath1624 .
we have to distinguish several cases .
the simplest case is when @xmath1625 and @xmath1626 are , in fact , both in @xmath165 . in particular ,
@xmath1627 and @xmath1628 .
let @xmath280 be a geodesic from @xmath296 to @xmath374 in @xmath1566 , i.e. a local geodesic in @xmath1566 not passing through @xmath295 and with minimal length .
let @xmath1291 be the path from @xmath378 to @xmath377 associated with @xmath280 as in lemma [ sec : cutting - r - graphs ] , which we may apply since @xmath1629 .
we claim that @xmath1291 does not pass through @xmath1292 .
indeed , if it did , then we would be able to find a point @xmath1630 such that @xmath1631 .
since also @xmath1632 , it would follow that @xmath1633 contradicting the fact that @xmath280 is a path in @xmath1566 . by lemma [ sec : cutting - r - graphs ]
, we deduce that @xmath1634 where at the last step we use that @xmath1635 .
let us now consider the cases where @xmath1636 , still assuming that @xmath1637 .
there are two possibilities . 1
. there exists @xmath1638 with @xmath1639 and @xmath1572 , and so @xmath1640 .
there exists @xmath1641 with @xmath1642 and @xmath1643 , and so @xmath1644 .
let us consider the first case , assuming @xmath1645 for definiteness .
the argument leading to is still valid , with @xmath1586 replacing @xmath378 . using @xmath1646
, we obtain @xmath1647 in the second case ( still assuming @xmath1645 without loss of generality ) , we have to modify the argument as follows .
we consider a geodesic @xmath280 from @xmath1648 to @xmath374 in @xmath422
. we @xmath1291 be the associated path from @xmath378 to @xmath377 ( again using lemma [ sec : cutting - r - graphs ] ) , and claim that @xmath1649 .
otherwise , @xmath280 would visit a point at distance less than @xmath1650 from @xmath295 .
on the other hand , the point of @xmath301 that is closest to @xmath295 is @xmath1651 .
but by lemma [ sec : stab - cutt - proc ] , we have @xmath1652 finally , since @xmath1653 we obtain that @xmath1654 , and the argument leading to yields @xmath1655 arguing similarly when @xmath1626 is no longer assumed to belong to @xmath165 , we obtain the following bound for every @xmath1656 : @xmath1657 writing @xmath1658 , by symmetry we thus conclude that @xmath1659 where the distortion is measured with respect to the metrics @xmath1533 and @xmath1623 .
now let @xmath1660 be the @xmath1661-enlargement of @xmath1614 with respect to @xmath1533 and @xmath1623 . since @xmath1614 is a correspondence between @xmath1566 and @xmath1611 , and all points of @xmath1570 ( resp .
@xmath1662 ) have distance at most @xmath1661 from @xmath1566 ( resp .
@xmath1611 ) under @xmath1533 ( resp .
@xmath1623 ) , we have that @xmath1660 is a correspondence between @xmath1570 and @xmath1662 , of distortion at most @xmath1663 finally , since @xmath1610 and @xmath1664 , and since @xmath1665 , we have @xmath1666 since @xmath1667 , the lemma then follows from the two preceding offset equations and the definition of the distance @xmath56 . fix @xmath112 . under the hypotheses of the theorem , for all @xmath0 large enough , by lemma [ sec : stab - cutt - proc-2 ] we have @xmath1668 it
is easily checked that , for all @xmath112 , @xmath1669 which both converge to @xmath148 as @xmath1670 .
the result follows .
let @xmath1020 be a safely pointed @xmath3-graph , and write @xmath435 for the length measure restricted to @xmath366 .
then @xmath1671 is an @xmath3-graph with @xmath432 . indeed , if @xmath401 is the edge of @xmath387 that contains @xmath295 , then it is easy to see that @xmath1672 is the graph obtained from @xmath191 by first deleting the interior of the edge @xmath401 , and then taking the kernel of the resulting @xmath3-graph .
taking the kernel of a graph does not modify its surplus , and so the surplus diminishes by @xmath68 during this operation , which corresponds to the deletion of the edge @xmath401 .
moreover , we see that @xmath445 is stable under this operation , in the sense that if @xmath1673 , then for every @xmath295 such that @xmath185 is safely pointed , the space @xmath422 is again in @xmath445
. indeed , the edges in @xmath1672 are either edges of @xmath387 , or a concatenation of edges in @xmath387 , and so the minimum edge - length can only increase . on the other hand , the total core length and surplus can only decrease .
let us now consider the following random cutting procedure for @xmath3-graphs .
if @xmath159 is an @xmath3-graph which is not an @xmath3-tree , then it contains at least one cycle , and by proposition [ sec : skeleton - core - kernel-2 ] , @xmath341-almost every point of any such cycle is in @xmath366 .
consequently , the measure @xmath1674 is non - zero , and we can consider a point @xmath295 chosen at random in @xmath366 with distribution @xmath437 . then @xmath185 is a.s . safely pointed by proposition [ sec : skeleton - core - kernel-2 ] .
let @xmath436 be the distribution of @xmath422 . by convention ,
if @xmath191 is an @xmath3-tree , we let @xmath438 . by combining corollary [ sec : cutt - safely - point-1 ] ( ii ) with theorem [ thm : safelypointed ] , we immediately obtain the following statement . [
sec : randomly - cutting - r-1 ] fix @xmath303 , and let @xmath1675 and @xmath191 be elements of @xmath445 such that @xmath1676 as @xmath2 . then @xmath1677 in @xmath58 , as @xmath62 .
in particular , @xmath439 defines a markov kernel from @xmath445 to itself for every @xmath981 .
since each application of this kernel decreases the surplus by @xmath68 until it reaches @xmath148 , it makes sense to define @xmath443 to be the law of @xmath1678 for every @xmath1679 , where @xmath1680 denotes the @xmath1681-fold composition of @xmath439 .
the next corollary follows immediately from proposition [ sec : randomly - cutting - r-1 ] by induction .
[ cor : randomly - cutting - r-2 ] fix @xmath303 , and let @xmath1675 and @xmath191 be elements of @xmath445 with @xmath1676 as @xmath2 .
then @xmath1682 in @xmath58 , as @xmath62 .
this proves theorem [ sec : breaking - cycles - r-1 ] .
lab was supported for this research by nserc discovery grant and by an fqrnt nouveau chercheur grant , and thanks both institutions for their support .
lab and cg were supported for this research by royal society international exchange award ie111100 .
nb acknowledges anr-09-blan-0011 .
cg is grateful to universit paris - sud for making her professeur invit for march 2011 which enabled work on this project to progress .
she is also grateful for support from the university of warwick .
gm acknowledges the support of grants anr-08-blan-0190 and anr-08-blan-0220 - 01 .
he is grateful to pims ( cnrs umi 3069 ) and ubc , vancouver , for hospitality in the year 20112012 , and for the support of universit paris - sud . , _ the objective method : probabilistic combinatorial optimization and local weak convergence _ , in probability on discrete structures , vol . 110 of encyclopaedia math .
sci . , springer , berlin , 2004 , pp . 172 .
height 2pt depth -1.6pt width 23pt , _ the continuum random tree .
an overview _ , in stochastic analysis ( durham , 1990 ) , vol .
167 of london math .
lecture note ser . , cambridge university press , cambridge , 1991 , pp . 2370 . ,
_ on the history of combinatorial optimization ( till 1960 ) _ , in discrete optimization , k. aardal , g. l. nemhauser , and r. weismantel , eds . , vol .
12 of handbooks in operations research and management , north holland , 2005 , pp . 168 . | consider the minimum spanning tree ( mst ) of the complete graph with @xmath0 vertices , when edges are assigned independent random weights . endow this tree with the graph distance renormalized by @xmath1 and with the uniform measure on its vertices .
we show that the resulting space converges in distribution as @xmath2 to a random measured metric space in the gromov
hausdorff prokhorov topology .
we additionally show that the limit is a random binary @xmath3-tree and has minkowski dimension @xmath4 almost surely .
in particular , its law is mutually singular with that of the brownian continuum random tree or any rescaled version thereof .
our approach relies on a coupling between the mst problem and the erds
rnyi random graph .
we exploit the explicit description of the scaling limit of the erds
rnyi random graph in the so - called critical window , established in @xcite , and provide a similar description of the scaling limit for a `` critical minimum spanning forest '' contained within the mst .
[ 0.41 ] figure 1 : a simulation of the minimum spanning tree on @xmath5 .
black edges have weights less than @xmath6 ; for coloured edges , weights increase as colours vary from from red to purple . |
the group of classical @xmath0 bootis stars comprises late b to early f - type , population i stars , with moderate to extreme ( up to a factor of 100 ) surface underabundances of most fe - peak elements and solar abundances of lighter elements ( c , n , o , and s ) .
they are rare , with a maximum of about 2% of all objects in the relevant spectral domain , between the zero- and terminal - age main - sequence ( zams and tams ) , are found to be such objects ( @xcite ) .
@xcite suggested that the peculiar chemical abundances on the stellar surfaces are due to selective accretion of circumstellar ( cs ) material .
due to gravitational settling and radiative acceleration , it is then mixed in the shallow convection zone of the star .
this explains why the anomalous abundance pattern is similar to that found in the gas phase of the interstellar medium ( ism ) in which refractory elements like iron and silicon have condensed into dust grains .
later on , @xcite and @xcite developed a model which describes the interaction of the star with its local ism and/or cs environment .
as a result , different levels of underabundance are produced by different amounts of accreted material relative to the photospheric mass .
the small fraction of this star group on the main - sequence ( ms ) is explained by the low probability of a star - cloud interaction and by the effects of meridional circulation , which dissolves any accretion pattern a few million years after the accretion has stopped .
the hot end of this model is due to significant stellar winds for stars with @xmath16@xmath1712000k whereas the cool end , at about 6500k , is defined by convection which prevents the accreted material manifesting at the stellar surface .
strong support for the selective accretion scenario has been given by @xcite who found that half of their sample of herbig ae / be stars exhibit the characteristic @xmath0 bootis type abundance pattern .
we know that the density of cs material around herbig ae / be stars is very high , perfectly suited as the source for accretion .
almost all @xmath0 bootis stars are located within the classical @xmath15 scuti/@xmath1 doradus instability strip .
@xcite presented a detailed analysis of their pulsational behaviour .
they concluded that at least 70% of the group members inside the classical instability strip pulsate , and they do so with high overtone p - modes ( [email protected] ) . in this paper , we present the newly detected @xmath1 doradus pulsational characteristics of the classical @xmath0 bootis star , hd 54272 .
interesting enough , it was misidentified as a rr lyrae type star by @xcite .
we performed a detailed time series analysis of the wide angle search for planets ( wasp ) and all sky automated survey ( asas ) photometric data .
four statistically significant frequencies and their combinations were detected .
our findings are discussed in comparison with the first hybrid @xmath0 bootis/@xmath1 doradus star , hr 8799 .
the latter is a very interesting young object hosting at least four planets and a massive dusty debris disk @xcite .
hd 54272 , on the other hand , is a fast rotating star almost at the tams with no detected ir excess . from spectroscopy
, we deduce a metallicity of @xmath140.8 to @xmath141.1dex compared to the sun . this fact together with the high projected rotational velocities , make this object an interesting and important test case for models dealing not only with @xmath1 doradus pulsation , but also with selective accretion in the presence of meridional circulation .
.the astrophysical parameters of hd 54272 ( * ? ? ?
* this work ) and hr 8799 @xcite .
the errors in the final digits of the corresponding quantity are given in parenthesis . [ cols="<,^,^ " , ]
hd 54272 ( bd+19 1620 , v=8.80mag ) was first classified as a member of the @xmath0 bootis group by @xcite . for this star , neither a hipparcos parallax nor a radial velocity is available .
@xcite searched for @xmath15 scuti type pulsations and found a null result with an upper limit of 1.4mmag within a time series of 2.6 hours .
it has to be emphasized that long term trends , such as the here reported @xmath1 doradus pulsation , were not investigated .
the only detailed follow - up analysis of the astrophysical parameters ( table [ stars_param ] ) for hd 54272 were presented by @xcite .
the mass , effective temperature and surface gravity are typical for an early f - type ms star . from isochrone fitting , they deduce that it has already spent 83% of its ms life - time . in this work
we use the refined yerkes classification ( often denoted as ` mk system ' ) scheme introduced by @xcite .
it uses the notation ` k ' , ` h ' , and ` m ' for the classification of the ca ii k , hydrogen , and metal lines according to standard stars .
it allows for a more detailed stellar classification than the classical yerkes system .
the spectrum of hd 54272 was observed at the osservatorio astronomico di padova - asiago with the 182 cm telescope using the boller & chivens spectrograph ( 600 lines mm@xmath3 grating ) which gave a nominal resolution of 2.0pixel@xmath3 ( dispersion of 85mm@xmath3 ) and a spectral coverage of about 1200 . in fig .
[ spec_hd54272 ] the observed spectrum is shown together with two standards ( hd 23194 and hd 113139 ) .
the latter were taken from @xcite and have a resolution of 1.8pixel@xmath3 . clearly , the high rotational velocity and the low metallicity of hd 54272 is visible .
it is classified as ` ka5hf2ma5v @xmath0 boo ' . in order to check the astrophysical parameters of hd 54272 and to search for an ir excess , we made use of the spectral energy distribution ( sed ) fitting tool by @xcite .
as input data we used the available @xmath19 , 2mass and wise photometry .
the effective temperature and surface gravity are well reproduced with the best fitting model .
furthermore , no ir excess was detected up to 22.1@xmath20 m .
we next investigated the metallicity of hd 54272 in more detail .
for this , a synthesized spectrum was computed using the program spectrumgrayro / spectrum / spectrum.html ] @xcite and modified versions of the atlas9 code taken from the vienna new model grid of stellar atmospheres , nemo @xcite .
we used a stellar atmosphere with the following parameters ( table [ stars_param ] ) : = 7000k , = 3.8 , and = 2kms@xmath3 .
the synthetic spectrum was first folded with the instrumental profile and then with different rotational profiles yielding a best fit for 250kms@xmath3 with an uncertainty of about 25kms@xmath3 . to test these parameters ,
a grid of atmospheres with effective temperatures and surface gravities around the input values were applied .
the hydrogen lines are best fitted with the original values with the constraint that they are not sensitive to . to estimate the [ m / h ] value , we used different models from + 0 to @xmath142dex .
figure [ synthetic ] shows the result for [ m / h][email protected] and @xmath141.0dex , respectively .
the median of the difference between the observed and synthetic spectrum with [ m / h][email protected] is only 0.2% .
therefore we adopted this values for hd 54272 . for the estimation of its error
, we used models with @xmath140.5 and @xmath141.5dex and calculated , again , the differences of the synthetic and observed spectra .
we notice that the difference with the more abundant model is smaller than the one with the more underabundant one . from this , we conclude that the metallicity range is between @xmath140.8 and @xmath141.1dex .
figure [ synthetic_two ] shows the effect of the underabundant model for a typical metallic - line region around 4600 .
we checked if the high rotational velocity is affecting the derived metallicity .
because no reliable stellar atmospheres taking into account the rotation , are available , coplanar ones are used .
assuming that the line - of - sight is almost equator - on , the models of @xcite are used to find the and values at the poles . from their table 1
, the model for a spectral type of a5 and a ratio of the angular velocity to the critical angular velocity of 0.9 almost perfectly fits the parameters of hd 54272 .
it implies = 9130k and = 4.25 at the poles .
again , we applied the above mentioned method to generate synthetic spectra using a different set of astrophysical parameters each 15@xmath21 along the disk and assuming rigid body rotation .
for the final spectrum , we integrated all individual spectra over the stellar disk .
figure [ synthetic_two ] shows the effect of the rotation using stellar atmospheres with [ m / h][email protected] . due to the high rotation velocity and the low abundance , the effect is only small and still in the range of the derived uncertainty
this is line with the results of @xcite investigated the effect of stellar rotation on the elemental abundance for a - type stars .
they found no trends for values up to 250kms@xmath3 ( see fig .
10 therein ) . as a conclusion ,
we are confident that the derived metal - weakness is indeed intrinsic and not due to the inadequate treatment of stellar rotation . according to the formula that the rotation becomes imperceptible when @xmath22@xmath23@xmath24=85mm@xmath3 , we deduce a lower limit of 85kms@xmath3 for our spectrum . from our spectral synthesis
, we derive a @xmath22 of about 250@xmath1325kms@xmath3 . using the mass and radius as listed in table [ stars_param ]
, we calculate a break - up velocity for hd 54272 of 313kms@xmath3 . adopting an equatorial line - of - sight
, this implies that hd 54272 rotates at least with 80% of its break - up velocity .
such a high velocity is exceptional not only for @xmath1 doradus pulsators @xcite , but also for ms stars close to the tams , in general @xcite
. the astrophysical characteristics of hd 54272 fit within the @xmath0 bootis group .
in addition , we investigated the @xmath22 distribution of the @xmath0 bootis stars , taken from @xcite , and apparent ` normal ' type stars .
the latter are from the list of @xcite which covers the corresponding spectral range , excluding the late b - type stars and objects marked as ` cb ' ( binaries ) and ` cp ' ( peculiar stars ) .
the samples consist of 42 @xmath0 bootis and 953 ` normal ' stars , respectively .
figure [ vsini ] shows the distributions of both groups .
we performed a pair - sample t - test @xcite which results that both distributions are on a 0.05 level , identical . therefore
, @xmath22 seems not to be an intrinsically ` critical ' astrophysical parameter for the formation and stability of the @xmath0 bootis phenomenon .
* however , the fast rotation of hd 54272 is not the common characteristics for @xmath0 bootis stars ( fig . [ vsini ] ) .
but there is at least one similar member , hd 193256 ( hip 100286 ) , with a @xmath22 of about 250kms@xmath3 .
it is also close to the tams but hotter ( 7700k ) .
* hr 8799 ( hd 218396 , v=5.95mag ) is an exceptional , close - by ( about 40pc ) , star classified as ` ka5hf0ma5v @xmath0 boo ' by @xcite .
they derived the astrophysical parameters as listed in table [ stars_param ] .
the typical @xmath0 bootis type abundance pattern was later confirmed by @xcite .
the membership of this star to the columba association which has an age of about 30myr is still a matter of debate @xcite .
moreover , @xcite summarizes the age determinations for this star ( see table 1 therein ) which range from 30myr to 1.6gyr .
the latter has been deduced from an asteroseismic study by @xcite .
we adopted the most probable one , @xmath18100myr , as argued and analysed by @xcite .
hr 8799 is hosting at least four planets and a massive dusty debris disk @xcite . the @xmath1 doradus type pulsation of hr 8799
was studied by @xcite .
they found four significant frequencies ( 1.9791 , 1.7268 , 1.6498 , and 0.2479d@xmath3 ) with corresponding amplitudes in johnson @xmath25 between 15.98 and 5.66mmag .
@xcite concluded from a detailed asteroseismic study that hr 8799 is intrinsically metal - weak which is in contrary to the common accepted theory that the @xmath0 bootis phenomenon is restricted to the photosphere alone .
however , one has to keep in mind that there are at least two important free parameters ( the inclination angle and the correct mode identification ) which are vital for the asteroseismic analysis .
such an intrinsic metal - weakness , compared to the sun , is still in the range which is found in the solar neighborhood @xcite .
if we accept this fact then also a non - solar isochrone has to be used to derive the age of hr 8799 .
for a given temperature and luminosity , a lower metal isochrone would result in an older age .
for the listed metallicity , it would naturally explain the often quoted discrepancy between the age of the columba association ( 30myr ) and some other determinations indicating an older age ( 100myr ) .
furthermore , we have to ask ( i ) if all ` classical ' @xmath0 bootis stars are intrinsic metal - weak or not , and ( ii ) if hr 8799 is a true member of the group . lllr & frequency & amplitude & + & [ d@xmath3 ] & [ mag ] & + + & + & + @xmath9 & 1.410117(2 ) & 0.02128(7 ) & 16.0 + @xmath10 & 1.283986(5 ) & 0.0138(1 ) & 6.3 + @xmath27 & 1.293210(4 ) & 0.01454(7 ) & 6.8 + @xmath11 & 1.536662(9 ) & 0.00664(7 ) & 5.5 + @xmath12 & 1.15722(1 ) & 0.00546(8 ) & 5.9 + @xmath28 & 0.22658(1 ) & 0.00523(8 ) & 6.4 + + & + & + @xmath9 & 1.41011(1 ) & 0.0232(1 ) & 7.8 + @xmath10 & 1.28399(2 ) & 0.0147(1 ) & 6.7 + @xmath27 & 1.293(6 ) & 0.009(3 ) & 4.8 + & 1.89(3 ) & 0.007(2 ) & 4.4 + & 5.01(2 ) & 0.007(2 ) & 4.0 + + distributions of @xmath0 bootis ( upper panel ) and apparent ` normal ' type ( lower panel ) stars .
according to a pair - sample t - test , both distributions are on a 0.05 level , identical.,width=283 ] to @xmath28.,width=283 ]
we analysed the photometric data from wasp @xcite and the all sky automated survey ( asas , e.g. * ? ? ?
* ; * ? ? ?
* ) surveys .
the wasp data , with the time base of 1566 d , are strongly affected by instrumental artifacts .
therefore , a fourier model with six harmonics was fitted to the data and measurements deviating more than 2@xmath29 were removed .
this procedure was applied three times . for the asas data ( time base of 2545 d ) , no such procedure was necessary , but the best quality data with flag ` a ' and ` b ' were used and few outliers were removed manually . since asas gives information about the brightness in five different apertures , a weighted mean value was used . a frequency analysis was performed using period04 @xcite software , which was also used for the determination of the uncertainties via monte carlo simulations .
asas data showed only five peaks : @xmath9 , @xmath10 , @xmath27 and two peaks with no clear interpretation .
the latter are probably due to the intrinsic characteristics of the data set and the accuracy of the individual measurements . in the wasp data ,
six significant peaks were identified ( see table [ frek ] ) . to estimate the reliability of these frequencies
, the data were prewhitened with their various combinations .
frequencies , which were left out , were always detected again .
in addition , we performed the complete time series analysis for two sub data sets .
the overall data set is divided by the time of observations to hjd 2450000 + [ 5137:5186 ] and [ 5500:5624 ] , respectively .
this procedure should avoid the influence of seasonal effects on the fourier analysis and the detection of possible spurious frequencies . within the errors ,
all frequencies listed in table [ frek ] were found in both data sets . from fig .
[ fourier ] it is apparent that frequency spectra are strongly affected by daily aliases , which can be expected when analysing data from single - side observations .
this makes the frequency identification somewhat ambiguous and also @xmath30 d@xmath3 should be considered as possible frequencies . without any additional observations peaks can be hardly interpreted and it is only possible to guess potential scenarios .
nevertheless , peaks with the highest amplitudes correspond to the same frequencies in asas and wasp data .
therefore peaks that were identified ( table [ frek ] ) should be of higher reliability than their 1-day aliases .
considering the frequencies in table [ frek ] , some interesting features appear .
the peaks @xmath10 , @xmath11 and @xmath12 are almost equally spaced with @xmath31 d@xmath3 ( calculated as average of spacings between frequencies ) with respect to @xmath9 : @xmath32 , @xmath33 and @xmath34 . since the spectral window of the data set ( fig .
[ fourier ] detail in top panel ) do not show any suspicious peak near 0.1265 d@xmath3 , we consider this spacing as intrinsic to the pulsation of the star .
this symmetrical pattern is similar to those of modulated rr lyrae stars @xcite , where the side - peaks are products of amplitude and/or phase modulation of the basic pulsation frequency with modulation period corresponding to the spacing between these peaks ( in our case , @xmath35 d ) .
such an interpretation of the frequency spectra , probably led to the misclassification of hd 54272 as a modulated rr lyrae type star with a pulsation period of 0.7789192 d ( @xmath36 in our identification , * ? ? ?
* ) . as fig .
[ phase ] shows , the mean amplitude of light changes is only about 0.06 mag , which is too small for rr lyraes @xcite .
however , omitting small amplitude and two other frequencies , which were found , the light curve shape could easily be interpreted as a modulated rr lyrae star @xcite .
we investigated if the found frequencies could be caused by rotation . from the observational point of view ,
such rotational induced variability was never found for members of the @xmath0 bootis group before @xcite . in general , as further ingredients , a local stellar magnetic field which produces spots on the surface ( as well as elemental overabundaces ) and slow rotation in order to guarantee stability are needed @xcite .
this so - called oblique rotator model @xcite , explains the variations as a geometrical effect as the star rotates , using a simple dipolar geometry . because hd 54272 is a very fast rotator and magnetic fields
were never detected for @xmath0 bootis stars @xcite it is very unlikely that the oblique rotator model could be applied .
based on the projected rotational velocity , radius ( table [ stars_param ] ) , and @xmath37=0@xmath21 as well as the formula by @xcite , we derive a rotational period of 0.45d for hd 54272 .
larger values of the inclination would result in even shorter periods .
the fastest rotating objects with periods close to 0.5d are very hot and high mass he - strong stars @xcite .
the @xmath38 d@xmath3 aliases of the first five frequencies match the expected rotational period .
also the regular spacing between frequencies could possibly be the consequence of rotation .
however , if we assume that the variability is due to rotation then we need a new model to explain this phenomenon because the oblique rotator is , with our current knowledge , not applicable for hd 54272 .
recently , @xcite investigated the light curves of several thousands of a - type stars observed by the kepler space mission .
he found that almost 10% of a - type stars exhibit variations resembling those usually attributed to starspots in cool stars , including a few exhibiting travelling waves usually interpreted as differentially rotating starspots .
most of these variations can also not be explained by the above mentioned oblique rotator model .
so , if we interpret the found variability of hd 54272 due to rotation , it would be the first @xmath0 bootis star with starspots on its surface which would be a challenging example for models .
explanation of peaks corresponding to @xmath39 and @xmath40 as coupling terms , which are often reported in @xmath1 doradus stars @xcite and which implies nonlinearity in the oscillation ( similar behavior was reported in hr 8799 , * ? ? ?
* ) , fails , because @xmath41 and @xmath42 are not present in frequency spectra and none of daily aliases match the values needed for such consideration .
furthermore , we checked the location of our target within the @xmath43 versus @xmath44 diagram which is sensitive to surface gravity @xcite . for this purpose ,
we used data for ` normal ' @xcite , @xmath0 bootis @xcite , and rr lyrae stars @xcite together with the standard relation from @xcite . the photometric indices were either taken from the listed references or from the catalogue by @xcite .
the result is shown in fig .
hd 54272 and hr 8799 are located where the other @xmath0 bootis stars are situated .
hr 8799 lies a little bit below the standard line which could be due to its young age and the surrounding dense circumstellar material .
the @xmath0 bootis stars are clearly separated from the rr lyrae stars with two exceptions : hd 148638 and hd 193281 .
both stars are rather close visual binary systems with almost equally bright companions .
therefore , the available photometry could be influenced by the binarity .
as a conclusion , the presented analysis clearly shows that hd 54272 is not a rr lyrae but a @xmath1 doradus pulsator without a hint of @xmath15 scuti variability @xcite . according to epoch hjd 2455164.116 .
the bottom panels show the frequency spectra of unprewhitened data.,width=321 ] versus @xmath44 diagram for hd 54272 , hr 8799 , ` normal ' ( crosses , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , @xmath0 bootis ( filled triangles , * ? ? ?
* ) , and rr lyrae stars ( open circles , * ? ? ?
the solid line is the standard relation from @xcite .
the rr lyrae stars are clearly separated from the @xmath0 bootis stars and our target.,width=264 ]
we analysed the time series of the wasp and asas projects in order to shed more light on the pulsational behaviour of hd 54272 .
this object was classified as @xmath0 bootis star and rr lyrae variable .
a combination which , a - priori , exclude each other .
our detailed analysis established this object as a @xmath1 doradus pulsator with six detected frequencies , which could be possibly biased by one day aliases .
the amplitudes of detected peaks are between 5 and 21mmag .
the three frequencies with the highest amplitudes were detected in both data sets , independently .
this example is also important for other @xmath1 doradus stars which might be misidentified as rr lyrae variables . from spectroscopic data ,
we deduce a classification of ` ka5hf2ma5v @xmath0 boo ' for hd 54272 . a comparison with synthetic spectra yields a high rotational velocity ( 250@xmath1325kms@xmath3 ) and a metal deficiency of @xmath45m / h@xmath46 $ ] of about @xmath140.8 to @xmath141.1dex compared to the sun .
it is located very close to the tams .
these facts make hd 54272 an important test case for several different models including pulsation , rotation , diffusion , and selective accretion .
our results are compared with those from the first detected star which shows @xmath0 bootis and @xmath1 doradus characteristics : hr 8799 .
although the pulsational behaviours are very similar , the evolutionary statuses are quite different .
hr 8799 is a very young object hosting at least four planets and a massive dusty debris disk .
if we accept that selective accretion plays a key role for the @xmath0 bootis phenomenon , it seems to have a significant effect on @xmath1 doradus type pulsation .
this could be concluded from the fact that these two stars have widely different evolutionary statuses and astrophysical parameters , but a strikingly similar pulsational characteristics .
the detection of further members of the @xmath0 bootis group showing @xmath1 doradus type pulsation would put further constraints on models explaining and describing these phenomena .
the wasp consortium consists of astronomers primarily from universities of st andrews , keele , leicester , warwick , queens university belfast , the open university , isaac newton group la palma and instituto de astrofsica de canarias .
wasp - north is hosted by the issac newton group on la palma and wasp - south is hosted by saao .
funding for wasp comes from consortium universities and from the uk science and technology facilities council .
this project is financed by the somopro ii programme ( 3sga5916 ) .
the research leading to these results has acquired a financial grant from the people programme ( marie curie action ) of the seventh framework programme of eu according to the rea grant agreement no . 291782 .
the research is further co - financed by the south - moravian region .
it was also supported by the grant of czech ministry of education , youth and sports 7amb12at003 , 7amb14at015 , muni / a/0735/2012 , and the financial contributions of the austrian agency for international cooperation in education and research ( bg-03/2013 and cz-10/2012 ) .
we thank markus hareter and martin netopil for their assistance as well as the referee for valuable comments .
this work reflects only the author s views and the european union is not liable for any use that may be made of the information contained therein .
baines e. k. et al . , 2012 , apj , 761 , 57 balona l. a. , 2011 , mnras , 415 , 1691 balona l. a. , guzik j. a. , uytterhoeven k. , smith j. c. , tenenbaum p. , twicken j. d. , 2011 , mnras , 415 , 3531 benk j. m. , szab r. , papar m. , 2011 , mnras , 417 , 974 bohlender , d. a. , landstreet , j. d. , 1990 , mnras , 247 , 606 bruntt h. , de cat p. , aerts c. , 2008 , a&a , 478 , 487 crawford d. l. , 1979 , aj , 84 , 1858 esposito s. et al . , 2013 ,
a&a , 549 , a52 fernley j. , barnes t. g. , skillen i. , hawley s. l. , hanley c. j. , evans d. w. , solano e. , garrido r. , 1998 , a&a , 330 , 515 folsom c. p. , bagnulo s. , wade g. a. , alecian e. , landstreet j. d. , marsden s. c. , waite i. a. , 2012 , mnras , 422 , 2072 garrison r. f. , gray r .o . , 1994 , aj , 107 , 1556 gray r. o. , garrison r. f. , 1987 , apjs , 65 , 581 gray r. o. , garrison r. f. , 1989a , apjs , 69 , 301 gray r. o. , garrison r. f. , 1989b , apjs , 70 , 623 gray r. o. , corbally c. j. , 1994 , aj , 107 , 742 gray r. o. , kaye a. b. , 1999 , aj , 118 , 2993 gray r. o. , corbally c. j. , garrison r. f. , mcfadden m. t. , robinson p. e. , 2003 , aj , 126 , 2048 hauck b. , mermilliod m. , 1998 , a&as , 129 , 431 heiter u. et al . , 2002 , a&a 392 , 619 kamp i. , paunzen e. , 2002 , mnras , 335 , l45 lenz p. , breger m. , 2004 , comm .
, 146 , 53 luck r. e. , heiter u. , 2005 , aj , 129 , 1063 martinez - galarza j. r. , kamp i. , su k. y. l. , gspr a. , rieke g. , mamajek e. e. , 2009 , aj , 694 , 165 mestel l. , moss d. , 2005 , mnras , 361 , 595 michaud g. , charland y. , 1986 , apj , 311 , 326 mikulek z. , krtika j. , henry g. w. , de villiers s. n. , paunzen e. , zejda , m. 2010 , a&a , 511 , l7 moya a. , amado p. j. , barrado d. , garca hernndez a. g. , aberasturi m. , montesinos b. , aceituno f. , , 2010a , mnras , 405 , l81 moya a. , amado p. j. , barrado d. , garca hernndez a. g. , aberasturi m. , montesinos b. , aceituno f. , 2010b , mnras , 406 , 566 paunzen e. , 2001 , a&a , 373 , 633 paunzen e. et al .
, 2002a , a&a , 392 , 515 paunzen e. , iliev i. kh . , kamp i. , barzova i. s. , 2002b , mnras , 336 , 1030 philip a. g. , egret d. , 1980 , a&as , 40 , 199 pojmaski g. , 1997 , aca , 47 , 467 pojmaski g. , 2001 , iau colloq . 183 : small telescope astronomy on global scales , 246 , 53 pollacco d. l. et al . , 2006 , pasp , 118 , 1407 preston g. w. , 1971 , pasp , 83 , 571 rees d. g. , 1987 , foundations of statistics , chapman & hall , london robitaille t. p. , whitney b. a. , indebetouw r. , wood k. , 2007 , apjs , 169 , 328 royer f. , zorec j. , gmez a. e. 2007 , a&a , 463 , 671 sadakane k. , 2006 , pasj , 58 , 1023 skarka m. , 2013 , a&a , 549 , a101 slettebak a. , kuzma t. j. , collins g. w. , ii , 1980 , apj , 224 , 171 smith h. a. , 2003 , rr lyrae stars , cambridge university press , cambridge stibbs , d. w. n. , 1950 , mnras , 110 , 395 szczygie d. m. , fabrycky d. c. , 2007 , mnras , 377 , 1263 takeda y. , han i. , kang d .-
, lee b .- c .
, kim k .- m .
, 2008 , journal of korean astronomical society , 41 , 83 zerbi f. m. et al .
, 1999 , mnras , 303 , 275 zorec j. , royer f. , 2012 , a&a , 537 , a120 zuckerman b. , rhee j. h. , song i. , bessell m. s. , 2011 , apj , 732 , 61 | we detect the second known @xmath0 bootis star ( hd 54272 ) which exhibits @xmath1 doradus type pulsations .
the star was formerly misidentified as a rr lyrae variable .
the @xmath0 bootis stars are a small group ( only 2% ) of late b to early f - type , population i stars which show moderate to extreme ( up to a factor 100 ) surface underabundances of most fe - peak elements and solar abundances of lighter elements ( c , n , o , and s ) . the photometric data from the wide angle search for planets ( wasp ) and all sky automated survey ( asas ) projects were analysed .
they have an overlapping time base of 1566 d and 2545 d , respectively .
six statistically significant peaks were identified ( @xmath2d@xmath3 , @xmath4d@xmath3 , @xmath5d@xmath3 , @xmath6d@xmath3 , @xmath7d@xmath3 and @xmath8d@xmath3 ) .
the spacing between @xmath9 and @xmath10 , @xmath9 and @xmath11 , @xmath12 and @xmath10 is almost identical . since the daily aliasing is very strong , the interpretation of frequency spectra is somewhat ambiguous . from spectroscopic data ,
we deduce a high rotational velocity ( 250@xmath1325kms@xmath3 ) and a metal deficiency of about @xmath140.8 to @xmath141.1dex compared to the sun . a comparison with the similar star , hr 8799 , results in analogous pulsational characteristics but widely different astrophysical parameters . since both are @xmath0 bootis type stars , the main mechanism of this phenomenon , selective accretion , may severely influence @xmath1 doradus type pulsations .
[ firstpage ] techniques : photometric stars : chemically peculiar stars : variables : @xmath15 scuti stars : variables : rr lyrae stars : individual : hd 54272 |
in these notes we provide additional details for some results of the main text .
we first provide additional motivation for our definition of the entanglement entropy ( ee ) from the perspective of the @xmath28 quantum group ( qg ) symmetry , and we prove that @xmath222 .
next , we give more examples of the computation of the qg ee for larger larger systems in various representations .
we elaborate on the construction in the rsos case , detailing in particular the mapping between the rsos and loop model representations .
finally we discuss the emergence of a @xmath223 term in the non - compact case .
our definition of the ee relies on using a modified trace , known as a jones trace , in which a factor of the type @xmath224 is inserted under the usual trace symbol . to ensure that the resulting reduced density operator @xmath25 makes sense in the qg formalism
, we must ensure that it commutes with the generators of @xmath28 .
we therefore consider the xxz spin-@xmath75 chain , with boundary terms as described in the main text .
the hamiltonian commutes with the following generators : s^z&=&_i _ i^z + s^+&=&_i q^_1^z ... q^_i-1^z_i^+ q^-_i+1^z ... q^-_m^z + s^-&=&_i q^_1^z ...
q^_i-1^z_i^- q^-_i+1^z ... q^-_m^zsince the generators commute with the hamiltonian , they share the same right and left eigenvectors . as a consequence they commute with the density operator @xmath24 = 0,= .
we split the spin chain in two parts @xmath2 , @xmath4 and define the reduced density operator @xmath25 using a jones trace over the part @xmath4 .
we consider the case where @xmath2 is in the middle of the chain between @xmath225 and @xmath226 , so that @xmath227 and @xmath228 . thus _
a=_b q^2s^z_b_l-2s^z_b_r , s^z_b=_ib_i^z .
let us check that the generators of @xmath229 on the subsystem @xmath2 commute with the reduced density operator @xmath25 .
we have the following relations : s^z&=&s^z_b_l+s^z_a + s^z_b_r , + s^&=&s^_b_l q^-s^z_a - s^z_b_r+q^s^z_b_l s^_a q^-s^z_b_r + q^s^z_b_l+s^z_as^_b_r .
consider first @xmath230 : s^z_a _ a & = & _ b(s^z_aq^2s^z_b_l-2s^z_b_r ) + & = & _ b((s^z - s^z_b_l - s^z_b_r)q^2s^z_b_l-2s^z_b_r ) .
obviously @xmath231 , @xmath232 , @xmath230 and @xmath233 commute .
since @xmath233 also commutes with @xmath24 : s^z_a _ a & = & _ b(q^2s^z_b_l-2s^z_b_rs^z)- _ b(q^2s^z_b_l-2s^z_b_rs^z_b_l ) + & -&_b(q^2s^z_b_l-2s^z_b_ls^z_b_l ) . for the two last terms we performed a cyclic permutation under the trace
we can now sum all terms and this proves @xmath234=0 $ ] .
next we do the same for @xmath235 : s^+_a _ a & = & _ b(s^+_a q^2s^z_b_l-2s^z_b_r ) + & = & _ b((q^-s^z_b_l+s^z_b_rs^+-q^-s^z_a - s^z_b_ls^+_b_l + & & -q^s^z_a+s^z_b_rs^+_b_r)q^2s^z_b_l-2s^z_b_r ) + & & ( 1)-(2)-(3 ) .
the first term @xmath236 of the right - hand side reads ( 1)&=&_b(q^-s^z_b_l+s^z_b_rs^+q^2s^z_b_l-2s^z_b_r ) + & = & _ b(q^2s^z_b_l-2s^z_b_rs^+q^-s^z_b_l+s^z_b_r)thanks to the cyclic permutation under the trace and the commutation of @xmath24 and @xmath237 .
we then deal with the second term @xmath238 involving @xmath239 : ( 2)&=&_b(q^-s^z_a - s^z_b_ls^+_b_lq^2s^z_b_l-2s^z_b_r ) + & = & _ b(s^+_b_l q^-s^z_a q^s^z_b_l-2s^z_b_r ) + & = & _ b(q^-s^z+s^z_b_r+s^z_b_l q^s^z_b_l-2s^z_b_r s^+_b_l ) + & = & _ b(q^-s^z_b_r+s^z_b_l q^-s^z+s^z_b_l s^+_b_l ) + & = & _ b(q^-2s^z_b_r+2s^z_b_l s^+_b_lq^-s^z_a - s^z_b_l ) , thanks to cyclic permutations of the operators over the subsystem @xmath4 , the commutation of @xmath233 with @xmath24 and the commutation of @xmath231 with @xmath230 and @xmath232 .
similarly for the term @xmath240 involving @xmath241 : ( 3)&=&_b(q^s^z_a+s^z_b_rs^+_b_rq^2s^z_b_l-2s^z_b_r ) + & = & _ b(q^-2s^z_b_r+2s^z_b_l s^+_b_rq^s^z_a+s^z_b_r ) . by regrouping the terms we find the desired property @xmath242 .
a very similar computation can be done for @xmath243 .
a meaningful ee must satisfy , at the very least , the symmetry property @xmath244 , meaning that subsystem @xmath2 is as entangled with @xmath4 , as @xmath4 with @xmath2 .
we now show that this is the case for our qg ee .
let us consider the case @xmath245 .
the proof is then simple and can be extended by analytic continuation to complex @xmath71 . in this case
the hamiltonian is symmetric , and @xmath246 .
we again divide our system in two pieces @xmath2 and @xmath4 with a cut in the middle ( for more complicated cuts the argument is similar ) and write the state in the following way : = _ i , j_i , j_a_b . the bases @xmath247 and @xmath248 can be chosen such that they have a well - defined magnetization . as a consequence ,
since the groundstate @xmath7 is in the zero - magnetization sector , we can define those bases such that the matrix @xmath249 is block - diagonal and where each block corresponds to a sector of @xmath2 and @xmath4 with a well - defined magnetization .
when we perform a singular value decomposition ( svd ) we end up with = _
s__a_b , where @xmath250 and @xmath251 are eigenvectors of @xmath26 and @xmath252 ; they form orthonormal bases of @xmath2 and @xmath4 .
the density matrix @xmath24 is = _ ,
the reduced density matrices @xmath25 and @xmath253 read _ a & = & _ bq^-2s_b= _ s_^2 q^-2s^_b _ a_a , + _ b & = & _ aq^-2s_a= _ s_^2 q^2s^_a _ b_b .
since the ground state is in the @xmath254 sector @xmath255 and thus the two reduced density operators have the same spectra and define the same entropy .
this proves the statement @xmath76 in the case of a cut in the middle of the system . to keep the discussion in the main text as simple as possible ,
we have presented all explicit computations for a chain with just @xmath256 sites .
this is of course no limitation to applying our general definitions , and accordingly we give here a few examples for higher values of @xmath36 .
we consider the case of @xmath257 sites .
the basis of link states is : = , = + the hamiltonian @xmath258 has the following ground state @xmath259 , where @xmath260 and @xmath261 .
the density matrix @xmath24 is = ( ^2 + + + ) .
consider first a bipartition in which @xmath2 is the first site , and @xmath4 the remainder .
take the partial markov trace over the three last sites , we find the reduced density operator _ a=(^2n+2+n)=. this leads to @xmath262 , the same result as found in the main text for the ee of the first spin with @xmath256 .
next we take @xmath2 as the first two sites , to compute the entanglement at the middle of the system .
we trace the density operator over the two last sites : _ a & = & ( + ( ^2n+2 ) ) + & = & ( + ( ^2 n + 2 ) e_1 ) . we now need to take the logarithm of @xmath25 .
we notice the identity @xmath263 , where @xmath264 \draw[thick ] ( 1,1 . ) arc ( 0:-180:0.5 ) ; \draw[thick ] ( 1,-0.2 ) arc ( 0:180:0.5 ) ; \end{tikzpicture } } $ ] .
it is then easy to find that _
a = -^2 + ( 1+n ) e_1 .
[ logrhoalemma ] we can now compute @xmath265 as & - & + ( ( ^2n + 2)+ ( 1 + n ) + & + & ( ^2n+2)(1+n ) ) . tracing over @xmath2 we finally obtain _ a & = & -_a_a = - n((^2n+2 ) + & + & ( 1+n)+(^2n+2)(1+n ) ) + & = & - ( 1 + n)^2 + ^2 .
[ s4sites ] we have verified that this expression coincides with the result obtained by using the modified trace in the vertex model
. it also agrees with computations in the potts spin representation for @xmath266 integer ( see below ) . for larger @xmath36
it is hard to compute this final partial trace directly , since the form of @xmath267 will be substantially more complicated than ( [ logrhoalemma ] ) .
a much more convenient option is to recall that gluing corresponding sites on top and bottom of any word in the tl algebra means technically to take the so - called markov trace @xmath268 .
this in turn can be resolved as follows = _ j [ 2j+1]_q _ v_j , [ markovtracedecomp ] where @xmath269 is the usual matrix trace over the ( standard ) module @xmath270 with @xmath271 defect lines , and @xmath272_q = \tfrac{q^k - q^{-k}}{q - q^{-1}}$ ] are @xmath71-deformed numbers such that the loop weight @xmath273_q = q+q^{-1}$ ] . in the simple @xmath257 case considered above
, @xmath2 has just two sites so that @xmath274 and @xmath275 are both one - dimensional with bases @xmath276 \draw[thick ] ( 1,1 . ) arc ( 0:-180:0.5 ) ; \end{tikzpicture } } \}$ ] and @xmath277 respectively .
thus we have the matrices . _ a |_v_0= , . _ a |_v_2= and _ a_a & = & _ v_0_a_a + & + & ( n^2 - 1 ) _ v_2_a_a .
[ markovtrace1 ] we find in the end _
a_a & = & + & + & ( n^2 - 1 ) , which is the same as ( [ s4sites ] ) after simplification .
we have made similar computations for @xmath278 sites , for all choises of the bipartition @xmath279 , finding again perfect agreement between the results from the loop model ( with the markov trace ) and the vertex model ( with the modified trace ) .
the same computation can be conducted in the @xmath280-state potts spin representation for @xmath280 integer .
there are @xmath281 spins labelled @xmath282 ( with @xmath283 ) enjoying free boundary conditions .
the interactions @xmath77 take different expressions depending on the parity of @xmath30 .
we have @xmath284 , where @xmath285 detaches the @xmath128th spin from the rest ( the new spin freely takes any of the @xmath280 values ) ; while @xmath286 , where @xmath287 joins two neighbouring spins ( forcing them to take the same value ) .
let us note that the eigenenergies of ( [ hising ] ) are @xmath294 .
the first two ( and in particular the ground state energy ) are also found in the loop model , but the latter two are not .
as we have seen , this does not prevent us from finding the same @xmath295 , which is a property of @xmath7 .
on the other hand , one can check that @xmath25 has the same spectrum in the two representations .
these conclusions extend to @xmath296 : we find the same @xmath297 , and the eigenvalues of @xmath25 are the same ( up to multiplicities , and after the elimination of non - relevant zero eigenvalues ) . in the rsos construction a height @xmath298 is defined at each site @xmath299 , subject to the constraint @xmath300 for each @xmath301 .
we note that while the loop model is defined on @xmath36 strands , there are now @xmath302 rsos heights .
free boundary conditions for the first and last spins in the equivalent potts model ( i.e. , no defect lines in the loop model ) correspond @xcite to fixing @xmath303 .
more generally , having @xmath271 defect lines in the loop model would correspond to @xmath304 and @xmath305 . to explain the details we move to a slightly larger example , namely @xmath278 and @xmath306 ( i.e. , @xmath296 ) , in order to see all non - trivial features of the computation at work .
consider the following labelling of the rsos basis states : @xmath307 it is straightforward to find the normalised ground state @xmath7 in this basis and check that its eigenenergy coincides with that of the other representations .
we denote @xmath308 as usual .
consider first the bipartion `` 4 + 2 '' , where @xmath2 contains the first four sites and @xmath4 the last two .
the junction of the two intervals is at @xmath309 , and we write @xmath310 for the corresponding intermediate height which belongs to both @xmath2 and @xmath4 .
the sector @xmath311 has the basis @xmath312 , while @xmath313 has the basis @xmath314 . in each sector , the reduced density matrix @xmath315 is formed by tracing over the heights belonging to @xmath4 . however , in both cases the choice of boundary conditions ( @xmath316 ) , the sector label @xmath128 and the rsos constraint fully fix the @xmath4-heights , so the trace is trivial : . _ a |_v_0 & = & ( ) .
each @xmath315 has precisely one non - zero eigenvalue , @xmath317 .
applying the normalisation _
j = _ j / [ 1 + 2j]_q we find that they agree with the eigenvalues of @xmath25 restrained to the standard module @xmath270 with @xmath271 defect lines in the corresponding loop model computation . therefore @xmath295 can be computed by the decomposition ( [ markovtracedecomp ] ) of the markov trace , which writes explicitly as ( [ markovtrace1 ] ) in this simple case . to see a non - trivial trace over the @xmath4-system
, we consider the same example but with the `` 3 + 3 '' bipartition ( i.e. , @xmath318 ) .
the sector @xmath319 has the basis ( for the @xmath2-heights ) @xmath320 .
the first basis element corresponds to states @xmath321 and @xmath322 for the full system , while the second basis elements corresponds to @xmath323 and @xmath324 .
therefore , to form @xmath325 we must sum over those possibilities ( which corresponds to tracing over the free height @xmath326 ) : _
a(12 ) = ( we describe now the general construction of a modified trace in the rsos models that will enable us to extract the true central charge from the entanglement , even in the non - unitary case .
we thus return to the @xmath153 rsos model with @xmath327 and @xmath328 .
set @xmath329 .
the interactions @xmath77 satisfying ( [ tlrels ] ) propagate @xmath151 into @xmath330 and read @xmath331 . with the boundary conditions
@xmath332 , the ground state @xmath7 then has the same energy as in the other representations this is also true in the non - unitary cases @xmath333 , provided we resolve the square root as @xmath334 when @xmath335 . obtaining the reduced density matrix @xmath25 for a bipartition @xmath279
involves a subtle manipulation of the height @xmath336 situated at the junction between @xmath2 and @xmath4 . for each fixed @xmath337
, define @xmath338 as the usual trace of @xmath339 over the @xmath4-heights ( @xmath151 with @xmath340 ) .
thus @xmath338 is a matrix indexed by the @xmath2-heights ( @xmath151 with @xmath341 ) .
the label @xmath128 is the quantum group spin of the sector @xmath342 of the reduced density matrix , and corresponds to having @xmath271 defect lines in the loop model computation .
now let @xmath343 denote the set eigenvalues of @xmath338 . we claim that @xmath344_q$ ]
yield precisely the corresponding loop model eigenvalues ( disregarding any zero eigenvalues ) , and that the qg entropy @xmath345 can be constructed therefrom by computing the markov trace ( [ markovtracedecomp ] ) over @xmath2 in the same way as for the loop model . note that this implies the following relation @xmath346_q \lambda_k(j)\ln\lambda_k(j)= \\ & -&\sum_{j , k}\lambda_k(j)\ln\lambda_k(j ) + \sum_{j , k}\lambda_k(j)\ln[2j+1]_q \ , , \nonumber\end{aligned}\ ] ] where on the left we have the qg ee , and the first term on the right is the ` ordinary ' ee for the ( non - unitary ) rsos model .
the term on the left scales like @xmath347 , and the first term on the right like @xmath348 .
this implies that the second term on the right must also be proportional to @xmath349 in the non - unitary case . while this is not impossible in view of our knowledge of entanglement spectra @xcite , the result clearly deserves a more thorough study .
we discuss here in more detail the correspondence between the rsos and loop models for the calculation of the rnyi entropies . for simplicity
, we only consider open boundary conditions with the boundary heights @xmath350 fixed to @xmath88 and a cut on the edge of the system ( figure [ loopr ] ) .
loops surround clusters of constant height . when a loop makes a right ( resp
. left ) turn by bouncing off a piece of a cluster , it gets a weight @xmath351 ( resp .
@xmath352 ) where @xmath12 and @xmath353 are the heights of the adjacent clusters ( cluster of height @xmath353 on the left , and @xmath12 on the right ) .
the amplitude @xmath354 is defined by @xmath355 , with @xmath356 . after summing over all possible heights ,
loops pick a weight @xmath357 if they are homotopic to a point .
black edges along the planes represent boundary conditions ( height fixed to @xmath350 ) . the top and bottom are identified .
non - contractible loops always wind @xmath92 times .
moreover , only the outermost loop ( red here ) sees the boundary directly . ]
let us consider the rnyi entropies for @xmath358 . using the replica picture
, we must compute the weigh of loops on the @xmath92-sheeted surface shown in figure [ loopr ] for @xmath129 . the weight of a non - contractible ( resp .
contractible ) loop on this surface is @xmath359 ( resp .
@xmath360 ) , due to the @xmath150 winding of non - contractible loops ; this must finally be summed over all possible path in the dynkin diagram .
for instance , consider the case of figure [ loopr ] , with @xmath129 and two non - contractible loops .
the first loop is the boundary between a cluster of height @xmath361 on its left and @xmath87 on its right .
it picks up a factor @xmath362 .
the second loop can either surround a cluster of height @xmath88 or @xmath188 , and therefore gets a factor @xmath363 . in the general case , we consider heights living on the @xmath153 dynkin diagram , and define the following matrix @xmath364 for @xmath365 .
thus , @xmath366 is the adjacency matrix with the non - contractible loop weights on @xmath92 replicas .
the matrix element @xmath367 is the weight of the configuration with @xmath169 non - contractible loops , where the boundaries are fixed to @xmath350 and the last loop surrounds a cluster of height @xmath368 .
since we sum over the height of the last cluster and we fixed @xmath361 , the full weight is @xmath369 where @xmath370 and @xmath371 .
the weight of a set of @xmath169 contractible loops is then @xmath372 , where @xmath373 and @xmath374 are the right and left eigenvectors of @xmath366 associated to the eigenvalues @xmath375 , for @xmath376 .
we hence need to sum over sectors where the weight of non - contractible loops is given by the different eigenvalues of @xmath366 .
we notice that the characteristic polynomial depends only on the products @xmath377 ( expand by the minors of the first column ) .
the characteristic polynomial is hence unchanged if we replace @xmath366 by the usual adjacency matrix , with elements @xmath378 .
the spectra of the adjacency matrices of @xmath153 dynkin diagrams are @xmath379 .
the normalized eigenvectors of @xmath366 are found @xcite as _ i & = & ( ) ^n ( ) , + _ i & = & ( ) ^-n ( ) for @xmath380 . finally , the rsos partition function with @xmath92 replicas and a boundary is a sum of loop partition functions @xmath381 , where non - contractible loops get a weight @xmath382 , i.e. , @xmath383 .
the prefactor @xmath384 can be computed from the eigenvectors of @xmath366 : _ k&=&_free|_k_k| [ detailedalphak ] + & = & ( ) ^1-n_i=1^m()^n .
the dominant contribution comes from non - contractible loop with the largest possible weight , @xmath385 ; this is because the corresponding sector is associated with the smallest electric charge .
in the limit where the system size goes to infinity we thus have @xmath386 .
we note that the detailed coefficient @xmath384 will depend on the boundary condition imposed on the left of the system . for fixed height @xmath350
, we see that the prefactor in ( [ detailedalphak ] ) contributes a term @xmath387 .
recall now the expression ( see e.g. @xcite ) of the affleck - ludwig entropy @xcite we restrict here to the unitary case @xmath156 for simplicity : @xmath388^{1/4}\left [ 2{\sin{\pi\over m}\over\sin{\pi\over m+1}}\right]^{1/2}\sin{\pi \hbdy\over m+1 } \nonumber\ ] ] we see that the @xmath350 dependence of the @xmath389 contribution to the rnyi entropy matches the ( logarithm of ) the degeneracy factor @xmath390 .
meanwhile , it is well known that fixing the rsos height to @xmath350 corresponds to the boundary condition @xmath391 in the above notation , while it is also known that the conformal boundary condition contributes to the entanglement by a factor @xmath389 which is precisely the logarithm of the degeneracy factor the affleck - ludwig entropy @xcite .
our calculation thus reproduces this subtle aspect of the entanglement entropy as well .
we also note that , despite the relative freedom offered by the coefficients @xmath384 , there does not seem to be any satisfactory way to concoct a boundary condition for which the leading term @xmath392 cancels out for all @xmath92 . as an example of non - compact cft we consider the @xmath14 liouville theory , which can be obtained by taking the @xmath393 limit of the unitary cfts based on the @xmath153 rsos models @xcite .
using our lattice approach , it is easy to see which features might emerge in this limit . indeed , going back to the calculation in the preceding subsection , and writing the contributions from all possible loop weights , we get the partition function for the @xmath92-replica model in the form @xmath394 and @xmath395 the coefficients @xmath396 are difficult to evaluate : they depend not only on the combinatorics of the model , but also on the normalization in the continuum limit of the different insertions of lattice vertex operators necessary to give the correct weights to non - contractbile loops . recall @xmath397 and @xmath398 .
we now take the limit @xmath399 , following the construction of @xcite . to this end
, we have to make an ansatz for the coefficients @xmath396 .
many comments in the literature suggest that the dependency on @xmath92 is negligible .
assume for extra simplicity that the @xmath396 are essentially constant as a function of @xmath169 as well ( this is all up to a lattice - cutoff power - law dependency , which we put in the @xmath3 term ) .
we have then , replacing sums by integrals when @xmath400 is large , that @xmath401 note that we have extended the integral to infinity , while since obviously @xmath402 , it looks like it should run only up to @xmath403 .
there are two reasons for this :
one is that at large @xmath3 the behavior is dominated by the region of @xmath404 small anyhow .
the other is that we have , in fact , neglected all the contributions occurring from electric charges ( in the lattice derivation ) shifted by integers . accepting ( [ ncznint ] )
we find , after evaluating the gaussian integrals , the result @xmath405 note that there are in fact additional factors of @xmath400 cropping up when we transform the sums ( [ nczn])([ncz1 ] ) into integrals .
they will only affect the entanglement by @xmath389 terms , so we have neglected them .
note that the argument hinges crucially on the absence of a non - trivial ( power - law ) dependency of the @xmath396 on @xmath169 .
since these coefficients depend , in part , on the correspondence between lattice and continuum , this may well provide a non - universal contribution to the @xmath223 term . | entanglement entropy has proven invaluable to our understanding of quantum criticality .
it is natural to try to extend the concept to `` non - unitary quantum mechanics '' , which has seen growing interest from areas as diverse as open quantum systems , non - interacting electronic disordered systems , or non - unitary conformal field theory ( cft ) .
we propose and investigate such an extension here , by focussing on the case of one - dimensional quantum group symmetric or supergroup symmetric spin chains .
we show that the consideration of left and right eigenstates combined with appropriate definitions of the trace leads to a natural definition of rnyi entropies in a large variety of models .
we interpret this definition geometrically in terms of related loop models and calculate the corresponding scaling in the conformal case .
this allows us to distinguish the role of the central charge and effective central charge in rational minimal models of cft , and to define an effective central charge in other , less well understood cases .
the example of the @xmath0 alternating spin chain for percolation is discussed in detail .
the concept of entanglement entropy has profoundly affected our understanding of quantum systems , especially in the vicinity of critical points @xcite .
a growing interest in _ non - unitary quantum mechanics _ ( with non - hermitian `` hamiltonians '' ) stems from open quantum systems , where the reservoir coupling can be represented by hermiticity - breaking boundary terms @xcite .
another motivation comes from disordered non - interacting electronic systems in @xmath1 dimensions ( d ) where phase transitions , such as the plateau transition in the integer quantum hall effect ( iqhe ) , can be investigated using a supersymmetric formalism and dimensional reduction via 1d non - hermitian quantum spin chains with supergroup symmetry ( susy ) @xcite .
susy spin chains and quantum field theories with target space susy also appear in the ads / cft correspondence @xcite and in critical geometrical systems such as polymers or percolation @xcite .
quantum mechanics with non - hermitian but pt - symmetric `` hamiltonians '' also gains increased interest @xcite .
can entanglement entropy be meaningfully extended beyond ordinary quantum mechanics ?
we focus in this letter on critical 1d spin chains and the associated 2d critical statistical systems and cfts .
this is the area where our understanding of the ordinary case is the deepest , and the one with most immediate applications . for ordinary critical quantum chains ( gapless , with linear dispersion relation ) , the best known result concerns the entanglement entropy ( ee ) of a subsystem @xmath2 of length @xmath3 with the ( infinite ) rest @xmath4 at temperature @xmath5 .
let @xmath6 denote the reduced density operator , where @xmath7 is the normalized ground state and @xmath8 .
the ( von neumann ) ee then reads @xmath9 .
one has @xmath10 for @xmath11 , where @xmath12 is a lattice cutoff and @xmath13 the central charge of the associated cft . for the xxz chain , @xmath14 .
statistical mechanics is ripe with non - hermitian critical spin chains : the ising chain in an imaginary magnetic field ( whose critical point is described by the yang - lee singularity ) , the alternating @xmath0 chain describing percolation hulls @xcite , or the alternating @xmath15 chain describing the iqhe plateau transition @xcite .
the ising chain is conceptually the simplest , as it corresponds to a _ rational _ non - unitary cft . in this case ,
abstract arguments @xcite suggest replacing the unitary result by @xmath16 where @xmath17 is the _ effective _ central charge .
for instance , for the yang - lee singularity , @xmath18 but @xmath19 ; in this case ( [ ceffres ] ) was checked numerically @xcite .
it was also checked analytically for integrable realizations of the non - unitary minimal cft .
the superficial similarity with the result @xmath20 for the thermal entropy per unit length of the infinite chain at @xmath21 suggests that ( [ ceffres ] ) is a simple extension of the scaling of the ground - state energy in non - unitary cft @xcite .
but the situation is more subtle , as can be seen from the fact that the leading behavior of the ee is independent of the ( low - energy ) eigenstate in which it is computed @xcite .
there are two crucial conditions in the derivation of ( [ ceffres ] ) : the left and right ground states @xmath22 must be identical , and the full operator content of the theory must be known .
these conditions hold for minimal , rational cft , but in the vast majority of systems the operator content depends on the boundary conditions ( so it is unclear what @xmath17 is ) , and @xmath23 , begging the question of how exactly @xmath24 , @xmath25 and @xmath26 are defined . in this letter
we explore this vast subject by concentrating on non - hermitian models with susy or quantum group ( qg ) symmetry .
we extend the general framework of coulomb gas and loop model representations to ee calculations .
we derive ( [ ceffres ] ) for minimal non - unitary models , and define modified ee involving the true @xmath13 even in non - unitary cases .
we finally introduce a natural , non - trivial ee in susy cases , even when the partition function @xmath27 .
[ [ ee - and - qg - symmetry . ] ] ee and qg symmetry .
+ + + + + + + + + + + + + + + + + + + we first discuss the critical @xmath28 qg symmetric xxz spin chain @xcite .
let @xmath29 be pauli matrices acting on space @xmath30 and define the nearest neighbor interaction @xmath31 \nonumber \ ] ] with @xmath32 , @xmath33 .
the hamiltonian @xmath34 with @xmath35 describes the ordinary critical xxz chain on @xmath36 sites , but we add the hermiticity - breaking boundary term @xmath37 to ensure commutation with the @xmath28 qg ( whose generators are given in the supplemental material ( sm ) ) .
consider first 2 sites , that is @xmath38 .
@xmath39 is not hermitian ; its eigenvalues are real @xcite but its left and right eigenstates differ .
we restrict @xmath40 $ ] , so the lowest energy is @xmath41 ( the other eigenenergy is @xmath42 ) .
the right ground state , defined as @xmath43 , is @xmath44 .
we use the ( standard ) convention that complex numbers are conjugated when calculating the bra associated with a given ket ; therefore @xmath45 . the density matrix @xmath46 ( in the basis @xmath47 )
is normalized , @xmath48 .
taking subsystem a ( b ) as the left ( right ) spin , the reduced density operator is @xmath49 , and therefore @xmath50 this coincides with the well - known result for the @xmath51 symmetric ( hermitian ) xxx chain ( @xmath52 ) .
but since @xmath39 is non - hermitian , it is more correct to work with left and right eigenstates defined by @xmath53 and @xmath54 ( or @xmath55 , since @xmath56 ) .
restricting to the sector @xmath57 we have @xmath58 , @xmath59 denote the right eigenstates with energies @xmath60 . the left eigenstates @xmath61 , @xmath62 are obtained from ( [ 0rstate])([1rstate ] ) by @xmath63 .
normalizations are such that @xmath64 , and @xmath65 for @xmath66 . since @xmath67
we need both l and r eigenstates to build a projector onto the ground state .
we thus _ define _ @xmath68 and @xmath69 .
we justify the use of a modified trace shortly with both geometrical and qg considerations .
observe that @xmath70 is normalized for the modified trace ( note the opposite power of @xmath71 ) : @xmath72 .
we now define the ee as @xmath73 the result ( [ svnmod ] ) is more appealing that ( [ svn ] ) : it depends on @xmath71 through the combination @xmath74 which is the quantum dimension of the spin @xmath75 representation of @xmath28 .
note that ( [ svnmod ] ) satisfies @xmath76 ( see sm ) .
[ [ entanglement - and - loops . ] ] entanglement and loops .
+ + + + + + + + + + + + + + + + + + + + + + + eq .
( [ svnmod ] ) admits an alternative interpretation in terms of loop models .
since @xmath77 obey the temperley - lieb ( tl ) relations , @xmath78 & = & 0 \mbox { for } |i - j|>1 \ , , \label{tlrels}\end{aligned}\ ] ] their action can be represented in terms of diagrams : @xmath79 \draw[thick ] ( 1,1 . ) arc ( 0:-180:0.5 ) ; \draw[thick ] ( 1,-0.2 ) arc ( 0:180:0.5 ) ; \end{tikzpicture } } $ ] contracts neighboring lines , and multiplication means stacking diagrams vertically , giving weight @xmath80 to each closed loop .
the ground state of @xmath38 is @xmath81 \draw[thick ] ( 1,1 . )
arc ( 0:-180:0.5 ) ; \end{tikzpicture } } $ ] ( @xmath82 stands for loop ) .
we check graphically that @xmath83 . with
the scalar product ordinarily used in loop models ( see sm ) , @xmath84 is correctly normalized .
the density matrix is @xmath85 \draw[thick ] ( 1,1 . )
arc ( 0:-180:0.5 ) ; \draw[thick ] ( 1,-0.2 ) arc ( 0:180:0.5 ) ; \end{tikzpicture } } $ ] .
the partial trace @xmath86 glues corresponding sites on top and bottom throughout @xmath4 ( here site @xmath87 ) .
the resulting reduced density matrix acts only on @xmath2 ( site @xmath88 ) : @xmath89 .
the gluing of @xmath2 creates a loop of weight @xmath90 , so @xmath91 .
the agreement with ( [ svnmod ] ) is of course no accident . indeed , for any spin-@xmath75 hamiltonian expressed in the tl algebra ( and thus commuting with @xmath28 ) , the ee and in fact , the @xmath92-replica rnyi ( see below ) entropies obtained with the modified traces and with the loop construction coincide .
we shall call these _ qg entropies _ , and denote them @xmath93 . [
[ coulomb - gas - calculation - of - the - ee . ] ] coulomb gas calculation of the ee .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + for the critical qg invariant xxz chain with @xmath94 , the ee @xmath93 scales as expected in cft , but with the _ true _ central charge @xmath95 ( instead of @xmath96 ) , where we parametrized @xmath97 . the simplest argument for this claim is field theoretical .
we follow @xcite , where the rnyi ee , @xmath98 , is computed from @xmath92 copies of the theory on a riemann surface with two branch points a distance @xmath3 apart . as the density operator is obtained by imaginary time evolution , we must project , in the case of non - unitary cft , onto @xmath99 in the `` past '' and on @xmath61 in the `` future '' , to obtain @xmath100 .
we calculate the qg rnyi ee using the loop model .
the geometry of @xcite leads to a simple generalization of well - known partition function calculations @xcite : an ensemble of dense loops now lives on @xmath92 sheets ( with a cut of length @xmath3 ) , and _ each loop _ has weight @xmath90 .
let @xmath101 denote the partition function .
crucially , there are now _ two types _ of loops : those which do not intersect the cut close after winding an angle @xmath102 , but those which do close after winding @xmath103 . to obtain the rnyi ee
, we must find the dependence of @xmath101 on @xmath3 . to this end
we use the coulomb gas ( cg ) mapping @xcite .
the tl chain is associated with a model of oriented loops on the square lattice .
assign a phase @xmath104 to each left ( right ) turn . in the plane
, the number of left minus the number of right turns is @xmath105 , so the weight @xmath106 results from summing over orientations .
the oriented loops then provide a vertex model , hence a solid - on - solid model on the dual lattice .
dual height variables are defined by induction , with the ( standard ) convention that the heights across an oriented loop edge differ by @xmath107 .
in cg theory , the large - distance dynamics of the heights is described by a gaussian field @xmath108 with action @xmath109={g\over 4\pi}\int { \rm d}^2x \left[(\partial_x\phi)^2+(\partial_y\phi)^2\right]$ ] and coupling @xmath110 . with @xmath92 replicas ,
we get in this way @xmath92 bosonic fields @xmath111 . the crux of the matter is the cut : a loop winding @xmath92 times around one of its ends should still have weight @xmath90 , whilst , since @xmath112 on the riemann surface , it gets instead @xmath113 .
we repair this by placing _ electric charges _ at the two ends ( labelled @xmath114 ) of the cut , @xmath115 and @xmath116 , where @xmath117 will be determined shortly .
more precisely , we must insert the vertex operators @xmath118 $ ] before computing @xmath101 .
this choice leaves unchanged the weight of loops which do not encircle nor intersect the cut . a loop that surround _
both _ ends ( and thus , lives on a single sheet ) gathers @xmath119 from the turns , and @xmath120 from the vertex operators ( since the loop increases the height of points @xmath121 and @xmath122 by @xmath123 ) .
the two contributions give in the end @xmath124 , summing up to @xmath90 as required .
finally , for a loop encircling only _ one _ end we get phases @xmath125 , so the correct weight @xmath90 is obtained setting @xmath126 . to evaluate the @xmath101 we implement the sewing conditions on the surface , @xmath127 with @xmath128 mod @xmath92 , by forming combinations of the fields that obey twisted boundary conditions along the cut .
for instance , with @xmath129 , we form @xmath130 and @xmath131 . while @xmath132 does not see the cut , @xmath133 is now twisted : @xmath134 . for arbitrary @xmath92 , the field
@xmath135 does not see the cut , while the others are twisted by angles @xmath136 with @xmath137 . using that the dimension of the twist fields in a _ complex _
bosonic theory is @xcite @xmath138 we find that the twisted contribution to the partition function is @xmath139 with @xmath140 .
meanwhile , the field @xmath132 , which would not contribute to the ee for a free boson theory ( here @xmath141 ) , now yields a non - trivial term due to the vertex operators with @xmath142 : @xmath143 with @xmath144 . assembling everything we get @xmath145 .
inserting @xmath146 and @xmath147 gives the rnyi entropies @xmath148\ln l\ ] ] ( @xmath93 is obtained for @xmath149 ) , hence proving our claim .
replicas ( here @xmath129 ) , the black loop must wind @xmath150 times before closing onto itself .
the red loop surrounds both ends of the cut . ]
we emphasize that the @xmath28 spin chain differs from the usual one simply by the boundary terms @xmath151 .
these are not expected to affect the ordinary ee , and the central charge obtained via the density operator @xmath8 ( with @xmath152 , but normalized as in our introduction ) will be @xmath96 . [
[ entanglement - in - non - unitary - minimal - models . ] ] entanglement in non - unitary minimal models .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we now discuss the restricted solid - on - solid ( rsos ) lattice models , which provide the nicest regularization of non - unitary cfts . in these models , the variables are `` heights '' on an @xmath153 dynkin diagram , with boltzmann weights that provide yet another representation of the tl algebra ( [ tlrels ] ) , with parameter @xmath154 and @xmath155 .
the case @xmath156 is hermitian , while @xmath157 leads to negative weights , and hence a non - unitary cft .
one has @xmath158 , and , for @xmath157 , the effective central charge determined by the state of lowest conformal weight @xcite through @xmath159is @xmath160 .
the case @xmath161 gives the yang - lee singularity universality class discussed in the introduction .
defining the ee for rsos models is not obvious , since their hilbert space ( we use this term even in the non - unitary case ) is not a tensor product like for spin chains .
most recent numerical and analytical work however neglected this fact , and ee was defined using a straightforward partial trace , summing over all heights in @xmath4 compatible with those in @xmath2 . in this case , it was argued and checked numerically that @xmath162 in the unitary case , and @xmath163 in the non - unitary case .
note that @xmath13 matches that of the loop model based on the same tl algebra , with @xmath164 . for details on the qg ee in the rsos case , see the sm .
the rsos partition functions can be expressed in terms of loop model ones , @xmath165 . in the plane , the equivalence @xcite replaces equal - height clusters by their surrounding loops , which get the usual weight @xmath90 through an appropriate choice of weights on @xmath153 . with periodic boundary conditions , the correspondence is more intricate due to non - contractible clusters / loops . on the torus @xcite ,
@xmath165 is defined by giving each loop ( contractible or not ) weight @xmath90 , whereas for the rsos model contractible loops still have weight @xmath90 , but one sums over sectors where each non - contractible loop gets the weights @xmath166 for any @xmath167 .
the same sum occurs ( see sm for details ) when computing @xmath101 of the riemann surface with @xmath92 replicas : non - contractible loops are here those winding one end of the cut .
note also that @xmath168 for rsos models , so the imaginary - time definition of @xmath24 in unambiguous @xcite .
crucially , the sum over @xmath169 is dominated ( in the scaling limit ) by the sector with the largest @xmath170 , that is @xmath171 and @xmath172 . in the non - unitary case ( @xmath173 ) , @xmath174 , and the ee
is found by extending the above computation .
we have still @xmath175 , but now @xmath176 .
to normalize at @xmath177 , one must divide by @xmath178 to the power @xmath92 , with the same charges : @xmath179 whence the rnyi entropy @xmath180 .
hence our construction establishes the claim of @xcite .
[ [ ee - in - the - sl21-susy - chain . ] ] ee in the @xmath0 susy chain .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + percolation and other problems with susy ( see the introduction ) have @xmath27 , hence @xmath181 , and the ee scales trivially .
having a non - trivial quantity that distinguishes the many @xmath181 universality classes would be very useful .
we now show that , by carefully distinguishing left and right eigenstates , and using traces instead of supertraces , one can modify the definition of ee to build such a quantity .
we illustrate this by the @xmath0 alternating chain @xcite which describes percolation hulls .
this chain represents the tl algebra ( [ tlrels ] ) with @xmath182 , and involves the fundamental ( @xmath183 ) and its conjugate ( @xmath184 ) on alternating sites , with @xmath185 .
the 2-site hamiltonian , @xmath38 , restricted to the subspace @xmath186 ( where @xmath187 are bosonic and @xmath188 is fermionic ) , reads e_1 = |0_r0_l| = ( + + )
( + + ) the eigenvectors are @xmath189 and @xmath190 ; note that conjugation is supergroup invariant ( i.e. , @xmath191 ) . hence , despite the misleading expression , @xmath39 is not unitary .
the density operator is @xmath192 and satisfies @xmath193 . the reduced density operator @xmath194 .
if we define the rnyi ee also with the supertrace , we get @xmath195 for all @xmath92 .
it is more interesting ( and natural ) to take instead the _ normal trace _ of @xmath196 ; this requires a renormalization factor to ensure @xmath197 .
we obtain then @xmath198 and thus @xmath199 .
this equals the qg rnyi ee with @xmath200 .
this calculation carries over to arbitrary size .
one finds that @xmath201 with weight @xmath182 , _ provided _ non - contractible loops winding around one cut end in the replica calculation get the modified weight @xmath202 instead of @xmath90 .
we can then use the cg framework developed in the context of the non - unitary minimal models to calculate the scaling behavior .
we use ( [ nonunit ] ) , with @xmath203 for percolation ( @xmath182 ) , and @xmath204 .
it follows that @xmath205 is purely imaginary , and that @xmath206 with @xmath207 .
[ [ numerical - checks . ] ] numerical checks .
+ + + + + + + + + + + + + + + + + all these results were checked numerically . as an illustration ,
we discuss only the case @xmath208 , for which the rsos and loop models have @xmath209 , while @xmath210 for the rsos model . in the corresponding @xmath28 chain , we measured the ( ordinary ) ee as in ( [ svn ] ) , the qg rnyi ee @xmath211 as in ( [ svnmod ] ) , and the qg rnyi ee for the modified loop model where non - contractible loops have fugacity @xmath212 ( instead of @xmath213 ) .
this , recall , should coincide asymptotically with the rnyi ee for the rsos model .
results ( see figure [ fig1 ] ) fully agree with our predictions .
( @xmath214 ) , versus the length of the cut @xmath3 , for a chain with @xmath215 sites and open boundary conditions .
purple dots show the usual ee with the unmodified trace .
averaging over the parity oscillations ( solid curve ) reveals the scaling with @xmath216 .
red squares show the @xmath129 rnyi entropy , with the modified trace giving weight @xmath217 to non - contractible loops ; this scales with @xmath210 .
blue triangles again show @xmath211 , but with @xmath218 ; the scaling then involves the true central charge @xmath209 . ]
[ [ conclusion . ] ] _ conclusion . _
+ + + + + + + + + + + + + while we have mostly discussed the critical case , we stress that the qg ee can be defined also away from criticality .
an interesting example is the @xmath0 alternating chain , for which staggering makes the theory massive ( this corresponds to shifting the topological angle away from @xmath219 in the sigma - model representation ) .
properties of the qg rnyi ee along this ( and other ) rg flows will be reported elsewhere . to summarize , we believe that our analysis completes our understanding of ee in 1d by providing a natural extension to non - unitary models in their critical or near - critical regimes .
there are clearly many situations ( such as phenomenological `` hamiltonians '' for open systems ) where things will be very different , but we hope our work will provide the first step in the right direction .
our approach also provides a long awaited `` coulomb gas '' handle on the correspondence between lattice models and quantum information quantities . in the sm
we apply this to show that , in the case of non - compact theories , the well - known @xmath220 term will be corrected by @xmath221 terms ( with , most likely , a non - universal amplitude ) , in agreement with recent independent work @xcite .
_ acknowledgments _ : the work of hs and jlj was supported by the erc advanced grant nuqft .
the work of hs was also supported by the us department of energy ( grant number de - fg03 - 01er45908 ) .
we thank b. doyon and o. castro - alvaredo for inspiring discussions and comments .
99 p. calabrese , j. cardy and b. doyon ( eds . ) , _ entanglement entropy in extended quantum systems _ , j. phys . a : math . theor .
* 42 * ( 2009 ) , and references therein .
t. prozen , j. phys .
a : math . theor . * 48 * , 373001 ( 2015 ) .
m. zirnbauer , j. math .
phys .
* 38 * , 2007 ( 1997 ) , and references therein .
n. beisert et al .
, lett .
math .
phys . * 99 * , 3 ( 2012 ) .
t. quella and v. schomerus , j. phys .
a : math . theor . * 46 * , 494010 ( 2013 ) . g. parisi and n. sourlas , j. physique lett . * 41 * , 403 ( 1980 ) .
c.m .
bender , j. phys .
conf .
ser . * 631 * , 012002 ( 2015 ) .
n. read and h. saleur , nucl .
phys .
b * 613 * , 409 ( 2001 ) .
d. bianchini , o. castro - alvaredo , b. doyon , e. levi and f. ravanini , j. phys .
a : math . gen . * 48 * , 04ft01 ( 2015 ) .
d. bianchini , o. castro - alvaredo and b. doyon , nucl .
phys .
b * 896 * , 835 ( 2015 ) . c. itzykson , h. saleur and j.b .
zuber , euro .
phys .
lett .
* 2 * , 91 ( 1986 ) .
f. alcaraz , m. berganza and g. sierra , phys .
rev .
lett . * 106 * , 201601 ( 2011 ) .
v. pasquier and h. saleur , nucl .
phys .
b * 330 * , 523 ( 1990 ) .
a. morin - duchesne , j. rasmussen , p. ruelle and y. saint aubin , j. stat .
mech .
( 2016 ) 053105 .
j. cardy and p. calabrese , j. phys .
a : math . theor . * 42 * , 504005 ( 2009 ) .
p. di francesco , h. saleur and j.b .
zuber , j. stat .
phys . * 49 * , 57 ( 1987 ) .
b. nienhuis , j. stat .
phys . * 34 * , 731 ( 1984 ) .
j.l .
jacobsen , _ conformal field theory applied to loop models _ , in a.j .
guttmann ( ed . ) , _ polygons , polyominoes and polycubes _ , lecture notes in physics * 775 * , 347424 ( springer , 2009 ) .
d. friedan , e. martinec and s. shenker , nucl .
phys .
b * 282 * , 13 ( 1987 ) .
v. pasquier , j. phys .
a : math . gen . * 20 * , l1229 ( 1987 ) .
p. di francesco , h. saleur and j.b .
zuber , nucl .
phys .
b * 300 * , 393 ( 1988 ) .
d. bianchini and o. castro - alavaredo , _ branch point twist field correlators in the massive free boson theory _
, arxiv:1607.05656 ; o. blondeau fournier and b. doyon , _ expectation values of twist fields and universal entanglement saturation of the massive free boson _ , in preparation .
h. saleur and m. bauer , nucl .
phys .
b * 320 * , 591 ( 1989 ) .
p. calabrese and a. lefvre , phys .
rev . a * 78 * , 032329 ( 2008 ) .
s. fredenhagen , m.r .
gaberdiel and c. keller , j. phys .
a : math .
theor . * 42 * , 495403 ( 2009 ) .
i. affleck and a.w.w .
ludwig , phys .
rev .
lett . * 67 * , 161 ( 1991 ) .
i. runkel and g.m.t .
watts , jhep 09 ( 2001 ) 006 . |
@xmath1 denotes the space of square integrable functions @xmath3 with norm @xmath4 .
we use @xmath5 to mean @xmath6 .
the following result was announced in @xcite .
[ thmbumps]for any @xmath7 and any @xmath8 there exists @xmath9 and @xmath10 and @xmath11 such that@xmath12 since the span of the hermite functions is dense in @xmath13 we have for some @xmath14@xmath15 now use finite backward differences to approximate the derivatives .
we have for some small @xmath9@xmath16 + b_{2}\tfrac{1}{t^{2}}\left [ e^{-x^{2}}-2e^{-\left ( x - t\right ) ^{2}}+e^{-\left ( x-2t\right ) ^{2}% } \right ] \nonumber\\ & + b_{3}\tfrac{1}{t^{3}}\left [ e^{-x^{2}}-3e^{-\left ( x - t\right ) ^{2}% } + 3e^{-\left ( x-2t\right ) ^{2}}-e^{-\left ( x-3t\right ) ^{2}}\right ] + \cdots\nonumber\\ & = \overset{n}{\underset{n=0}{% % tcimacro{\tsum } % % beginexpansion { \textstyle\sum } % endexpansion } } b_{n}\frac{1}{t^{n}}\overset{n}{\underset{k=0}{% % tcimacro{\tsum } % % beginexpansion { \textstyle\sum } % endexpansion } } \left ( -1\right ) ^{k}\binom{n}{k}e^{-\left ( x - kt\right ) ^{2}}% \text{.}\label{pfmaincoeffs}%\end{aligned}\ ] ] this result may be surprising ; it promises we can approximate to any degree of accuracy a function such as the following characteristic function of an interval@xmath17 } \left ( x\right ) : = \left\ { \begin{array } [ c]{l}% 1\\ 0 \end{array } \right .
\begin{array } [ c]{l}% \text{for } x\in\left [ -10,-11\right ] \\
\text{otherwise}% \end{array}\ ] ] with support far from the means of the gaussians @xmath18 which are located in @xmath19 at the points @xmath20 .
the graphs of these functions @xmath21 are extremely simple geometrically , being gaussians with the same variance .
we only use the right translates , and they all shrink precipitously ( exponentially ) away from their means .
@xmath22{2.5261in}{\begin{center } \includegraphics [ natheight=9.364200 in , natwidth=13.749600 in , height=1.7244 in , width=2.5261 in ] % { i100.png}% \\ $ \sum a_n e^{-\left ( x - nt\right ) ^2}\approx$ characteristic function ?
\end{center}}}% % endexpansion\ ] ] _ surely there is a gap in this sketchy little proof ? _ no . we will , however , flesh out the details in section [ secappalgo ] . the coefficients @xmath23
are explicitly calculated and the @xmath2 convergence carefully justified . but these details are elementary .
we include them in the interest of appealing to a broader audience .
_ then is this merely another pathological curiosity from analysis ?
we probably need impractically large values of _ @xmath14 _ _ to approximate any interesting functions . _ _
no , @xmath14 need only be as large as the hermite expansion demands .
certainly this particular approach depends on the convergence of the hermite expansion , and for many applications hermite series converge slower than other fourier approximations
after all , hermite series converge on all of @xmath24 while , e.g. , trigonometric series focus on a bounded interval .
hermite expansions do have powerful convergence properties , though .
for example , hermite series converge uniformly on finite compact subsets whenever @xmath25 is twice continuously differentiable ( i.e. , @xmath26 ) and @xmath27 for some @xmath28 as @xmath29 .
alternately if @xmath25 has finitely many discontinuities but is still @xmath26 elsewhere and @xmath30 the expansion again converges uniformly on any closed interval which avoids the discontinuities @xcite , @xcite : .
if @xmath25 is smooth and properly bounded , the hermite series converges faster than algebraically @xcite .
_ then is the method unstable ? _
yes , there are two serious drawbacks to using theorem [ thmbumps ] . \1 .
numerical differentiation is inherently unstable .
fortunately we are estimating the derivatives of gaussians , which are as smooth and bounded as we could hope , and so we have good control with an explicit error formula .
it is true , though , that dividing by @xmath31 for small @xmath32 and large @xmath33 will eventually lead to huge coefficients @xmath23 and round - off error .
there are quite a few general techniques available in the literature for combatting round - off error in numerical differentiation .
we review the well - known @xmath33-point difference formulas for derivatives in section [ secimpconv ] .
the surprising approximation is only possible because it is weaker than the typical convergence of a series in the mean .
unfortunately@xmath34 theorem [ thmbumps ] requires recalculating all the @xmath23 each time @xmath14 is increased .
further , the @xmath23 are not unique .
the least squares best choice of @xmath23 are calculated in section [ secleastsqrs ] , but this approach gives an ill - conditioned matrix .
a different formula for the @xmath23 is given in theorem [ algbump ] which is more computationally efficient .
despite these drawbacks the result is worthy of note because of the new and unexpected opportunities which arise using an approximation method with such simple functions . in this vein ,
section [ seclowftrig ] details an interesting corollary of theorem [ thmbumps ] : apply the fourier transform to see that low - frequency trigonometric series are dense in @xmath13 with gaussian weight function .
in this section theorem [ algbump ] gives an explicit formula for the coefficients @xmath23 of theorem [ thmbumps ] .
let s review the details of the hermite - inspired expansion@xmath35 claimed in the proof .
the formula for these coefficients is@xmath36 be warned this is not precisely the standard hermite expansion , but a simple adaptation to our particular requirements .
let s check this formula for the @xmath37 using the techniques of orthogonal functions .
remember the following properties of the hermite polynomials @xmath38 ( @xcite , e.g. ) .
define @xmath39 .
the set of hermite functions@xmath40 is a well - known basis of @xmath1 and is orthonormal since@xmath41 this means given any @xmath42 it is possible to write@xmath43 @xmath44equality in the @xmath2 sense@xmath45 where@xmath46 the necessity of this formula for @xmath47 can easily be checked by multiplying both sides of @xmath48 by @xmath49 , integrating and applying @xmath50 .
however , we want@xmath51 so apply this process to @xmath52 .
but @xmath53 may not be @xmath2 integrable . if it is not , we must truncate it : @xmath54 } \left ( x\right ) $ ] is @xmath2 for any @xmath55 and @xmath56 } \underset{\epsilon/3}{\approx}f$ ] for a sufficiently large choice of @xmath57 .
now we get new @xmath47 as follows@xmath58 } \left ( x\right ) & = \overset{\infty}{\underset{n=0}{% % tcimacro{\tsum } % % beginexpansion { \textstyle\sum } % endexpansion } } c_{n}\tfrac{1}{\sqrt{n!2^{n}\sqrt{\pi}}}h_{n}\left ( x\right ) e^{-x^{2}% /2}\text{\qquad so}\\ f\left ( x\right ) \chi_{\left [ -m , m\right ] } \left ( x\right ) & = \overset{\infty}{\underset{n=0}{% % tcimacro{\tsum } % % beginexpansion { \textstyle\sum } % endexpansion } } c_{n}\tfrac{\left ( -1\right ) ^{n}}{\sqrt{n!2^{n}\sqrt{\pi}}}\left ( -1\right ) ^{n}h_{n}\left ( x\right ) e^{-x^{2}}=\overset{\infty}% { \underset{n=0}{% % tcimacro{\tsum } % % beginexpansion { \textstyle\sum } % endexpansion } } b_{n}\frac{d^{n}}{dx^{n}}e^{-x^{2}}%\end{aligned}\ ] ] where@xmath59 } \left ( x\right ) h_{n}\left ( x\right ) e^{-x^{2}/2}\left ( x\right ) dx\\ & = \tfrac{1}{\sqrt{n!2^{n}\sqrt{\pi}}}\underset{\mathbb{r}}{% % tcimacro{\tint } % % beginexpansion { \textstyle\int } % endexpansion } f\left ( x\right ) \chi_{\left [ -m , m\right ] } \left ( x\right ) h_{n}\left ( x\right ) dx\end{aligned}\ ] ] so we must have@xmath60 } \left ( x\right ) e^{x^{2}% } \frac{d^{n}}{dx^{n}}e^{-x^{2}}dx\text{.}\label{line b_n}%\ ] ] now the second step of the proof of theorem [ thmbumps ] claims that the gaussian s derivatives may be approximated by divided backward differences@xmath61 in the @xmath1 norm .
we ll use the `` big oh '' notation : for a real function @xmath62 the statement `` @xmath63 as @xmath64 '' means there exist @xmath65 and @xmath66 such that @xmath67 for @xmath68 .
[ proplpdivdiff]for each @xmath69 and @xmath70@xmath71 in appendix [ secimpconv ] the pointwise formula is derived:@xmath72 where all of the @xmath73 are between @xmath74 and @xmath75 .
therefore the proposition holds with @xmath76 since @xmath77 is integrable for each @xmath78 .
this is not perfectly obvious because we do nt have explicit formulae for the @xmath73 .
but the tails of @xmath79 vanish exponentially , the continuity of @xmath79 guarantees a finite maximum on the bounded interval between the tails , and @xmath80 . continuing the derivation of the coefficients @xmath23 we now have for sufficiently small @xmath81@xmath82 e^{-\left ( x - kt\right ) ^{2}}\label{line f approxi}%\ ] ] in the last equality we just switched the order of summation ( see @xcite , section 2.4 for an overview of such tricks ) . combining @xmath83 and @xmath84 we have [ algbump]for any @xmath7 and any @xmath8 there exist @xmath10 and @xmath85 such that for any @xmath81 with @xmath86@xmath87 for some choice of @xmath11 dependent on @xmath14 and @xmath32 .
if @xmath53 is integrable , then one choice of coefficients is@xmath88 if @xmath53 is not integrable , replace @xmath25 in the above formula with @xmath56 } $ ] where @xmath57 is chosen large enough that @xmath89 } \right\| _ { 2}<\epsilon$ ] . [ remunifg]the approximation in theorem [ algbump ] also holds on @xmath90 $ ] with the uniform norm since the hermite expansion is uniformly convergent on @xmath91 $ ] ( see @xcite , @xcite ) and the finite difference formula s error term from appendix [ secimpconv ] converges to 0 uniformly as @xmath92 .
the stone - weierstrass theorem does not apply in this situation because linear combinations of gaussians with a single variance do not form an algebra .
[ remdense]as a consequence of theorem [ algbump ] for any @xmath8 the closed linear span of @xmath93 is @xmath1 .
it is even sufficient to replace @xmath94 with @xmath95 .
let s explore some concrete examples in applying theorem [ algbump ] .
choose an interesting function with discontinuities and some support negative:@xmath96 } \left ( x\right ) : = \left\ { \begin{array } [ c]{l}% \left ( x-1\right ) ^{2}\\ 0 \end{array } \right .
\begin{array } [ c]{l}% \text{for } x\in\left [ -1,2\right ] \\
\text{otherwise}% \end{array}\ ] ] and observe graphically : @xmath97 \left ( x\right ) $ } } % % { } { 200.png}{\special { language " scientific word " ; type " graphic " ; % maintain - aspect - ratio true ; display " pict " ; valid_file " f " ; % width 1.7253 in ; height 1.1917 in ; depth 0 in ; original - width 8.8643 in ; % original - height 6.1039 in ; cropleft " 0 " ; croptop " 1 " ; cropright " 1 " ; % cropbottom " 0 " ; filename ' 200.png';file-properties " xnpeu";}}}% % beginexpansion { \parbox[b]{1.7253in}{\begin{center } \includegraphics [ natheight=6.103900 in , natwidth=8.864300 in , height=1.1917 in , width=1.7253 in ] % { i200.png}% \\ $ f\left ( x\right ) : = \left ( x-1\right ) ^2\chi_{\left [ -1,2\right ] } \left ( x\right ) $ \end{center}}}% % endexpansion% % tcimacro{\frame{itbpfu}{1.6094in}{1.1122in}{0in}{\qcb{hermite series $ n=20$}% % } { } { 210.png}{\special { language " scientific word " ; type " graphic " ; % maintain - aspect - ratio true ; display " usedef " ; valid_file " f " ; % width 1.6094 in ; height 1.1122 in ; depth 0 in ; original - width 8.5833 in ; % original - height 5.9171 in ; cropleft " 0 " ; croptop " 1 " ; cropright " 1 " ; % cropbottom " 0 " ; filename ' 210.png';file-properties " xnpeu";}}}% % beginexpansion { \parbox[b]{1.6094in}{\begin{center } \includegraphics [ natheight=5.917100 in , natwidth=8.583300 in , height=1.1122 in , width=1.6094 in ] % { i210.png}% \\ hermite series $ n=20 $ \end{center}}}%
% endexpansion% % tcimacro{\frame{itbpfu}{1.6025in}{1.1087in}{0in}{\qcb{hermite $ n=40$}}% % { } { 220.png}{\special { language " scientific word " ; type " graphic " ; % maintain - aspect - ratio true ; display " usedef " ; valid_file " f " ; % width 1.6025 in ; height 1.1087 in ; depth 0 in ; original - width 8.4371 in ; % original - height 5.8228 in ; cropleft " 0 " ; croptop " 1 " ; cropright " 1 " ; % cropbottom " 0 " ; filename ' 220.png';file-properties " xnpeu";}}}% % beginexpansion { \parbox[b]{1.6025in}{\begin{center } \includegraphics [ natheight=5.822800 in , natwidth=8.437100 in , height=1.1087 in , width=1.6025 in ] % { i220.png}% \\ hermite
$ n=40 $ \end{center}}}% % endexpansion $ ] @xmath98{c}% % theorem \ref{algbump}\\ %
$ n=20 $ , $ t=.05$% % \end{tabular } % } } { } { 230.png}{\special { language " scientific word " ; type " graphic " ; % maintain - aspect - ratio true ; display " usedef " ; valid_file " f " ; % width 1.6155 in ; height 1.1147 in ; depth 0 in ; original - width 9.052 in ; % original - height 6.2292 in ; cropleft " 0 " ; croptop " 1 " ; cropright " 1 " ; % cropbottom " 0 " ; filename ' 230.png';file-properties " xnpeu";}}}% % beginexpansion { \parbox[b]{1.6155in}{\begin{center } \includegraphics [ natheight=6.229200 in , natwidth=9.052000 in , height=1.1147 in , width=1.6155 in ] % { i230.png}% \\% \begin{tabular } [ c]{c}% theorem \ref{algbump}\\ $ n=20 $ , $ t=.05$% \end{tabular } \end{center}}}% % endexpansion% % tcimacro{\frame{itbpfu}{1.6172in}{1.1156in}{0in}{\qcb{{}% % \begin{tabular } % [ c]{c}% % theorem \ref{algbump}\\ % $ n=20 $ , $ t=.01$% % \end{tabular } % } } { } { 240.png}{\special { language " scientific word " ; type " graphic " ; % maintain - aspect - ratio true ; display " usedef " ; valid_file " f " ; % width 1.6172 in ; height 1.1156 in ; depth 0 in ; original - width 8.6983 in ; % original - height 5.9897 in ; cropleft " 0 " ; croptop " 1 " ; cropright " 1 " ; % cropbottom " 0 " ; filename ' 240.png';file-properties " xnpeu";}}}% % beginexpansion { \parbox[b]{1.6172in}{\begin{center } \includegraphics [ natheight=5.989700 in , natwidth=8.698300 in , height=1.1156 in , width=1.6172 in ] % { i240.png}% \\ { } % \begin{tabular } [ c]{c}% theorem \ref{algbump}\\ $ n=20 $ , $ t=.01$% \end{tabular } \end{center}}}% % endexpansion% % tcimacro{\frame{itbpfu}{1.6163in}{1.1147in}{0in}{\qcb{{}% % \begin{tabular } % [ c]{c}% % theorem \ref{algbump}\\ % $ n=40 $ , $ t=.01$% % \end{tabular } % } } { } { 250.png}{\special { language " scientific word " ; type " graphic " ; % maintain - aspect - ratio true ; display " usedef " ; valid_file " f " ; % width 1.6163 in ; height 1.1147 in ; depth 0 in ; original - width 8.8955 in ; % original - height 6.1246 in ; cropleft " 0 " ; croptop " 1 " ; cropright " 1 " ; % cropbottom " 0 " ; filename ' 250.png';file-properties " xnpeu";}}}% % beginexpansion { \parbox[b]{1.6163in}{\begin{center } \includegraphics [ natheight=6.124600 in , natwidth=8.895500 in , height=1.1147 in , width=1.6163 in ] % { i250.png}% \\ { } % \begin{tabular } [ c]{c}% theorem \ref{algbump}\\ $ n=40 $ , $ t=.01$% \end{tabular } \end{center}}}% % endexpansion\ ] ] the hermite approximation is slowed by discontinuities , but does converge .
the next choice of @xmath25 is continuous but not smooth .
@xmath99 \left ( x\right ) $ } } % % { } { 300.png}{\special { language " scientific word " ; type " graphic " ; % maintain - aspect - ratio true ; display " usedef " ; valid_file " f " ; % width 1.6414 in ; height 1.1277 in ; depth 0 in ; original - width 8.6144 in ; % original - height 5.9067 in ; cropleft " 0 " ; croptop " 1 " ; cropright " 1 " ; % cropbottom " 0 " ; filename ' 300.png';file-properties " xnpeu " ; } } } % % beginexpansion { \parbox[b]{1.6414in}{\begin{center } \includegraphics [ natheight=5.906700 in , natwidth=8.614400 in , height=1.1277 in , width=1.6414 in ] % { i300.png}% \\ $ f\left ( x\right ) : = \left ( \sin x\right ) \chi_{\left [ -\pi,\pi\right ] } \left ( x\right ) $ \end{center } } } % endexpansion% % tcimacro{\frame{itbpfu}{1.6622in}{1.1312in}{0in}{\qcb{% % \begin{tabular } % [ c]{c}% % hermite expansion\\ % $ n=10$% % \end{tabular } % } } { } { 311.png}{\special { language " scientific word " ; type " graphic " ; % maintain - aspect - ratio true ; display " usedef " ; valid_file " f " ; % width 1.6622 in ; height 1.1312 in ; depth 0 in ; original - width 9.4585 in ; % original - height 6.4169 in ; cropleft " 0 " ; croptop " 1 " ; cropright " 1 " ; % cropbottom " 0 " ; filename ' 311.png';file-properties " xnpeu " ; } } } % % beginexpansion { \parbox[b]{1.6622in}{\begin{center } \includegraphics [ natheight=6.416900 in , natwidth=9.458500 in , height=1.1312 in , width=1.6622 in ] % { i311.png}% \\% \begin{tabular } [ c]{c}% hermite expansion\\ $ n=10$% \end{tabular } \end{center } } } % endexpansion% % tcimacro{\frame{itbpfu}{1.6302in}{1.132in}{0in}{\qcb{% % \begin{tabular } % [ c]{c}% % hermite expansion\\ % $ n=20$% % \end{tabular } % } } { } { 320.png}{\special { language " scientific word " ; type " graphic " ; % maintain - aspect - ratio true ; display " usedef " ; valid_file " f " ; % width 1.6302 in ; height 1.132 in ; depth 0 in ; original - width 10.0733 in ; % original - height 6.979 in ; cropleft " 0 " ; croptop " 1 " ; cropright " 1 " ; % cropbottom " 0 " ; filename ' 320.png';file-properties " xnpeu " ; } } } % % beginexpansion { \parbox[b]{1.6302in}{\begin{center } \includegraphics [ natheight=6.979000 in , natwidth=10.073300 in , height=1.132 in , width=1.6302 in ] % { i320.png}% \\% \begin{tabular } [ c]{c}% hermite expansion\\ $ n=20$% \end{tabular } \end{center } } } % endexpansion\]]@xmath100{l}% % theorem \ref{algbump}\\ % $ n=10 $ , $ t=.01$% % \end{tabular } % } } { } { 330.png}{\special { language " scientific word " ; type " graphic " ; % maintain - aspect - ratio true ; display " usedef " ; valid_file " f " ; % width 1.6431 in ; height 1.126 in ; depth 0 in ; original - width 9.8018 in ; % original - height 6.698 in ; cropleft " 0 " ; croptop " 1 " ; cropright " 1 " ; % cropbottom " 0 " ; filename ' 330.png';file-properties " xnpeu " ; } } } % % beginexpansion { \parbox[b]{1.6431in}{\begin{center } \includegraphics [ natheight=6.698000 in , natwidth=9.801800 in , height=1.126 in , width=1.6431 in ] % { i330.png}% \\% \begin{tabular } [ c]{l}% theorem \ref{algbump}\\ $ n=10 $ , $ t=.01$% \end{tabular } \end{center } } } % endexpansion% % tcimacro{\frame{itbpfu}{1.6579in}{1.1346in}{0in}{\qcb{% % \begin{tabular } % [ c]{l}% % theorem \ref{algbump}\\ % $ n=20 $ , $ t=.05$% % \end{tabular } % } } { } { 340.png}{\special { language " scientific word " ; type " graphic " ; % maintain - aspect - ratio true ; display " pict " ; valid_file " f " ; % width 1.6579 in ; height 1.1346 in ; depth 0 in ; original - width 10.1149 in ; % original - height 6.9064 in ; cropleft " 0 " ; croptop " 1 " ; cropright " 1 " ; % cropbottom " 0 " ; filename ' 340.png';file-properties " xnpeu " ; } } } % % beginexpansion { \parbox[b]{1.6579in}{\begin{center } \includegraphics [ natheight=6.906400 in , natwidth=10.114900 in , height=1.1346 in , width=1.6579 in ] % { i340.png}% \\% \begin{tabular } [ c]{l}% theorem \ref{algbump}\\ $ n=20 $ , $ t=.05$% \end{tabular } \end{center } } } % endexpansion% % tcimacro{\frame{itbpfu}{1.6475in}{1.1277in}{0in}{\qcb{% % \begin{tabular } %
[ c]{l}% % theorem \ref{algbump}\\ % $ n=20 $ , $ t=.01$% % \end{tabular } % } } { } { 350.png}{\special { language " scientific word " ; type " graphic " ; % maintain - aspect - ratio true ; display " usedef " ; valid_file " f " ; % width 1.6475 in ; height 1.1277 in ; depth 0 in ; original - width 10.1149 in ; % original - height 6.9064 in ; cropleft " 0 " ; croptop " 1 " ; cropright " 1 " ; % cropbottom " 0 " ; filename ' 350.png';file-properties " xnpeu " ; } } } % % beginexpansion { \parbox[b]{1.6475in}{\begin{center } \includegraphics [ natheight=6.906400 in , natwidth=10.114900 in ,
height=1.1277 in , width=1.6475 in ] % { i350.png}% \\% \begin{tabular } [ c]{l}% theorem \ref{algbump}\\ $ n=20 $ , $ t=.01$% \end{tabular } \end{center } } } % endexpansion\ ] ] in section [ secimpconv ] we review a standard technique accelerating this convergence in @xmath32 . in our experiments , though , we ve found the hermite expansion is generally the bottleneck , not the round - off error of the derivative approximations for @xmath0.@xmath101{c}% % hermite expansion\\ % $ n=60$% % \end{tabular } % } } { } { 400.png}{\special { language " scientific word " ; type " graphic " ; % maintain - aspect - ratio true ; display " usedef " ; valid_file " f " ; % width 1.4961 in ; height 1.0222 in ; depth 0 in ; original - width 9.9168 in ; % original - height 6.7706 in ; cropleft " 0 " ; croptop " 1 " ; cropright " 1 " ; % cropbottom " 0 " ; filename ' 400.png';file-properties " xnpeu " ; } } } % % beginexpansion { \parbox[b]{1.4961in}{\begin{center } \includegraphics [ natheight=6.770600 in , natwidth=9.916800 in , height=1.0222 in , width=1.4961 in ] % { i400.png}% \\% \begin{tabular } [ c]{c}% hermite expansion\\ $ n=60$% \end{tabular } \end{center } } } % endexpansion% % tcimacro{\frame{itbpfu}{1.4875in}{1.0188in}{0in}{\qcb{% % \begin{tabular } % [ c]{c}% % hermite expansion\\ % $ n=100$% % \end{tabular } % } } { } { 410.png}{\special { language " scientific word " ; type " graphic " ; % maintain - aspect - ratio true ; display " usedef " ; valid_file " f " ; % width 1.4875 in ; height 1.0188 in ; depth 0 in ; original - width 10.031 in ; % original - height 6.8545 in ; cropleft " 0 " ; croptop " 1 " ; cropright " 1 " ; % cropbottom " 0 " ; filename ' 410.png';file-properties " xnpeu " ; } } } % % beginexpansion { \parbox[b]{1.4875in}{\begin{center } \includegraphics [ natheight=6.854500 in , natwidth=10.031000 in , height=1.0188 in , width=1.4875 in ] % { i410.png}% \\% \begin{tabular } [ c]{c}% hermite expansion\\ $ n=100$% \end{tabular } \end{center } } } % endexpansion% % tcimacro{\frame{itbpfu}{1.5333in}{1.0144in}{0in}{\qcb{% % \begin{tabular } % [ c]{c}% % hermite expansion\\ % $ n=120$% % \end{tabular } % } } { } { 420.png}{\special { language " scientific word " ; type " graphic " ; % maintain - aspect - ratio true ; display " usedef " ; valid_file " f " ; % width 1.5333 in ; height 1.0144 in ; depth 0 in ; original - width 10.1356 in ; % original - height 6.6876 in ; cropleft " 0 " ; croptop " 1 " ; cropright " 1 " ; % cropbottom " 0 " ; filename ' 420.png';file-properties " xnpeu " ; } } } % % beginexpansion { \parbox[b]{1.5333in}{\begin{center } \includegraphics [ natheight=6.687600 in , natwidth=10.135600 in , height=1.0144 in , width=1.5333 in ] % { i420.png}% \\% \begin{tabular } [ c]{c}% hermite expansion\\ $ n=120$% \end{tabular } \end{center } } } % endexpansion\ ] ] we need about 120 terms before visual accuracy is achieved for this simple function .
there is a host of methods in the literature for improving convergence of the hermite expansion , but generally we have better success with functions that are smooth and bounded @xcite .
our last examples in this section illustrate how convergence is faster for functions which are smooth and `` clamped off '' , meaning multiplied by @xmath102 } $ ] whether or not they are positive or symmetric .
theorem [ thmbumps ] promises any @xmath2 function can be approximated @xmath103 .
theorem [ algbump ] gives a formula for the coefficients @xmath23 but this formula is not unique , and in fact is not `` best '' according to the classical continuous least squares technique . in least squares we minimize the error function@xmath104 by setting @xmath105 for @xmath106 and solving for the @xmath23 . these @xmath107 linear equations are called the * normal equations*. the matrix form of this system is @xmath108 where @xmath57 is the matrix@xmath109 _ { j , k=0}^{n}%\ ] ] and@xmath110 _ { j=0}^{n}\text{\qquad and\qquad } \overrightarrow{b}=\left [ \underset{\mathbb{r}}{\int}f\left ( x\right ) e^{-\left ( x - jt\right ) ^{2}}dx\right ] _ { j=0}^{n}%\ ] ] @xmath57 is symmetric and invertible , so we can always solve for the @xmath23 . but
these least squares matrices are notorious for being ill - conditioned when using non - orthogonal approximating functions .
the hilbert matrix is the archetypical example .
the current application is no exception since the matrix entries are very similar for most choices of @xmath14 and @xmath32,@xmath111so round - off error is extreme .
choosing @xmath112 instead of @xmath113 in the graphed example above requires almost 300 significant digits .
for @xmath114 define the norm@xmath115 write @xmath116 @xmath117 to mean @xmath118 .
[ thmlowfreq]for every @xmath114 and @xmath8 there exists @xmath14 @xmath119 and @xmath85 such that for any @xmath81 with @xmath86 @xmath120 for some choice of @xmath121 dependent on @xmath14 and @xmath32 .
we use the fourier transform with convention@xmath122 \left ( s\right ) = \frac{1}{\sqrt{2\pi}}% \underset{\mathbb{r}}{% % tcimacro{\tint } % % beginexpansion { \textstyle\int } % endexpansion } f\left ( x\right ) e^{-isx}dx\text{.}%\ ] ] @xmath123 is a linear isometry of @xmath124 with@xmath125 & = \frac{1}{\sqrt{2\alpha}% } e^{-\frac{s^{2}}{4\alpha}}\text{,}\\ \mathcal{f}\left [ f\left ( x+r\right ) \right ] & = e^{-irs}\mathcal{f}% \left [ f\left ( x\right ) \right ] \text{\qquad and}\\ \mathcal{f}\left [ g\ast h\right ] & = \sqrt{2\pi}\mathcal{f}\left [ g\right ] \mathcal{f}\left [ h\right ] \text{.}%\end{aligned}\ ] ] where @xmath126 is convolution .
let @xmath127 and we now show @xmath128 \left ( x\right ) \in l^{2}$ ] . notice @xmath129
\in l^{2}$ ] and@xmath130 \right\| _ { 1}=c\left\| g^{2}\right\| _ { 1}=c\left\| g\right\| _ { 2}% ^{2}=c\left\| f\right\| _ { 2}^{2}<\infty\end{aligned}\ ] ] for some @xmath131 . here
$ ] is the solution to the diffusion equation for time @xmath32 and initial condition @xmath133 .
( the notation @xmath134 refers to the weierstrass transform . )
the reason for the third equality in the previous calculation is that @xmath135 maintains the @xmath136 integral of any positive initial condition @xmath133 for all time @xmath9 @xcite .
now approximate the real and imaginary parts of @xmath137 with theorem [ algbump ] .
then we get@xmath138 \left ( x\right ) \underset{\epsilon}{\approx}\ \overset{n}{\underset{n=0}{% % tcimacro{\tsum } % % beginexpansion { \textstyle\sum } % endexpansion } } a_{n}e^{-\left ( x - nt\right ) ^{2}}\text{\qquad}a_{n}\in\mathbb{c}%\ ] ] and applying @xmath123 gives@xmath139 hence@xmath140 using the fact that @xmath141 .
this result is surprising , even in the context of this paper , because for instance , series of the form @xmath142 for all @xmath32 and @xmath23 are not dense in @xmath2 and in fact only inhabit a 4-dimensional subspace of the infinite dimensional hilbert space @xcite . on any finite interval @xmath143 $ ] for any @xmath144
the finite linear combinations of sine and cosine functions with frequency lower than @xmath145 are dense in @xmath146 , \mathbb{r}\right ) $ ] . on @xmath143
$ ] the gaussian is bounded and so the norms with or without weight function are equivalent
. apply theorem [ thmlowfreq ] to @xmath147 , \mathbb{r}\right ) $ ] and choose @xmath32 such that @xmath148 to get@xmath149 where @xmath150 \left ( x\right ) \right ] e^{x^{2}}\frac{d^{k}}{dx^{k}}\left ( e^{-x^{2}}\right ) dx\text{.}%\ ] ] applying remark [ remdense ] to this result shows even discrete sets of positive frequencies that approach 0 make the span of the corresponding sine and cosine functions equal to@xmath146 , \mathbb{r}% \right ) $ ] . finally , low - frequency cosines span the even functions : on any finite interval @xmath151 $ ] for any @xmath144 the finite linear combinations of cosine functions with frequency lower than @xmath145 are dense in @xmath152 , \mathbb{r}\right ) $ ] .
let @xmath153 , \mathbb{r}\right ) $ ] and extend it as an even function on @xmath154 $ ] .
now use the previous corollary to write @xmath155 we d like to conclude right now that the @xmath156 or @xmath157 , but that is not true .
however , every function @xmath117 on @xmath154 $ ] may be written uniquely as a sum of even and odd functions@xmath158 and so@xmath159 therefore@xmath160 _ { e}=\overset{n}{\underset{n=0}{% % tcimacro{\tsum } % % beginexpansion { \textstyle\sum } % endexpansion } } a_{n}\cos\left ( ntx\right ) \text{.}%\ ] ] beware this last result ; it s not as strong as fourier approximation .
the coefficients for the sine functions calculated above may be large ; the proposition merely promises the linear combination of the sine terms is small .
using least squares , however , will have vanishing sine coefficients .
the mathematical inspiration for theorem [ thmbumps ] comes from geometrical investigations in infinite dimensional control theory .
we noticed that function translation and vector translation in @xmath161 do not commute .
specifically , `` function translation '' is a flow on the infinite dimensional vector space @xmath1 given by the map @xmath162 where @xmath163 .
`` vector translation '' in the direction of @xmath42 is the flow @xmath164 where @xmath165 .
taking for example @xmath166 and composing @xmath167 and @xmath168 we see @xmath169 since for @xmath170@xmath171 notice however the key fact@xmath172 in finite dimensions the commutator quotient above gives the lie bracket @xmath173 $ ] of the vector fields @xmath174 and @xmath175 which generate the flows @xmath167 and @xmath168 , respectively . a fundamental result in
finite - dimensional control theory states that the reachable set via @xmath174 and @xmath175 is given by the integral surface to the distribution made up of iterated lie brackets starting from @xmath174 and @xmath175 ( chow s theorem , which is an interpretation of frobenius foliation theorem , see @xcite , e.g. ) .
the idea we are exploiting is that iterated lie brackets for our flows @xmath167 and @xmath168 will give successive derivatives of the gaussian , whose span is dense in @xmath13 . consequently , the reachable set via @xmath167 and @xmath168 from @xmath170 should be all of @xmath1 .
that is to say , sums of translates and multiples of one gaussian ( with fixed variance ) can approximate any integrable function .
unfortunately this program does nt automatically work on the infinite dimensional vector space @xmath1 since the function translation flow is not generated by a simple vector field on @xmath13 .
so instead of studying vector fields , we consider flows as primary .
the fundamental results can be rewritten and still hold in the general context of a metric space @xcite .
then other functions besides @xmath76 can be checked to be derivative generating and other flows may be used in place of translation .
e.g. , fourier approximation is achieved using dilation @xmath176 where @xmath177 and @xmath178 .
this gives us a general tool for determining the density of various families of functions .
another opportunity for generalizing the results of this paper presents itself with the observation that hermite expansions are valid for functions defined on @xmath179 or @xmath180 and in spaces of tempered distributions ; and divided differences works in all of these spaces as well .
note also that while the results of section [ secappalgo ] work for uniform approximations of continuous functions on finite intervals ( remark [ remunifg ] ) , this is an open question for low - frequency trigonometric approximations .
the results of this paper can be ported to the language of control theory where we can then conclude the system@xmath181 is bang - bang controllable with controls of the form @xmath182 .
theorem [ algbump ] drives the initial condition @xmath170 to any state in @xmath2 under the system @xmath183 , but may be nowhere near optimal for approximating a function such as @xmath184 , since it uses only gaussians @xmath185 with choices of @xmath186 . finally , interpreting theorem [ thmbumps ] in terms of signal analysis , we see a gaussian filter is a universal synthesizer with arbitrarily short load time
. let @xmath187 .
a gaussian filter is a linear time - invariant system represented by the operator@xmath188 notice if you feed @xmath134 a dirac delta distribution @xmath189 ( an ideal impulse at time @xmath190 ) you get @xmath191 .
then theorem [ thmbumps ] gives for any @xmath7 and any @xmath8 and any @xmath192 there exists @xmath9 and @xmath10 with @xmath193 such that@xmath194 for some choice of @xmath11 .
feed a gaussian filter a linear combination of impulses and we can synthesize any signal and arbitrarily small load time @xmath195 .
the design of physical approximations to an analog gaussian filter are detailed in @xcite , @xcite .
the results in this paper may be much improved with voluminous techniques available from numerical analysis .
e.g. , @xcite gives an algorithm which speeds the calculation of sums of gaussians , and @xcite explores hermite expansion acceleration useful in step 1 of the proof of theorem [ thmbumps ] .
this section is devoted to reviewing methods which improve the error in step 2 , approximating derivatives of the gaussian with finite differences .
we also derive the error formula used in proposition [ proplpdivdiff ] .
above we approximated derivatives with the formula@xmath196{c}% $ \underbrace{\frac{1}{t^{n}}% % tcimacro{\tsum _ { k=0}^{n}}% % beginexpansion { \textstyle\sum_{k=0}^{n } } % endexpansion \left ( -1\right ) ^{n - k}\binom{n}{k}f\left ( x+kt\right ) } $ \\ gives round - off error as $ t\rightarrow0^{+}$% \end{tabular}% \begin{tabular } [ c]{c}%
$ \underset{}{+}$% \end{tabular}% \begin{tabular } [ c]{l}% $ \underbrace{o\left ( t\right ) } $ \\ truncation error \end{tabular } \text{.}\label{linenthder = o(t)}%\ ] ] the nrlund - rice integral may be of interest for extremely large @xmath33 as it avoids the calculation of the binomial coefficient by evaluating a complex integral . in this section , though , we devote our attention to deriving @xmath33-point formulas ; these formulas decrease round - off error by increasing the number of evaluations @xmath197this shrinks the truncation error without sending @xmath64 . in approximating the @xmath78th derivative with an @xmath198 point formula@xmath199
we wish to calculate the coefficients @xmath200 . in the forward difference method , the @xmath201 , but keeping these values general allows us to find the coefficients for the central or backward difference formulas just as easily . the following method for finding the @xmath200
was shown to us by our student jeffrey thornton who rediscovered the formula .
taylor s theorem has@xmath202 for some @xmath203 between @xmath74 and @xmath204 . from this
it follows@xmath205{c}% $ f\left ( x\right ) $ \\ $ tf^{\prime}\left ( x\right ) $ \\ $ \vdots$\\ $ t^{n}f^{\left ( n\right ) } \left ( x\right ) $ \\ $ t^{n+1}$% \end{tabular } \right ] ^{t}\left [ \begin{tabular } [ c]{cccc}% $ 1 $ & $ 1 $ & $ \cdots$ & $ 1$\\ $ k_{0}$ & $ k_{1}$ & $ \cdots$ & $ k_{n}$\\ $ \frac{k_{0}^{2}}{2!}$ & $ \frac{k_{1}^{2}}{2!}$ & $ \cdots$ & $ \frac{k_{n}^{2}% } { 2!}$\\ $ \vdots$ & $ \vdots$ & $ \ddots$ & $ \vdots$\\ $ \frac{k_{0}^{n}}{n!}$ & $ \frac{k_{1}^{n}}{n!}$ & $ \cdots$ & $ \frac{k_{n}^{n}% } { n!}$\\ $ \tfrac{k_{0}^{n+1}f^{\left ( n+1\right ) } \left ( \xi_{0}\right ) } { \left ( n+1\right ) ! } $ & $ \frac{k_{1}^{n+1}f^{\left ( n+1\right ) } \left ( \xi _ { 1}\right ) } { \left ( n+1\right ) ! }
$ & $ \cdots$ & $ \frac{k_{n}% ^{n+1}f^{\left ( n+1\right ) } \left ( \xi_{n}\right ) } { \left ( n+1\right ) ! } $ % \end{tabular } \right ] \left [ \begin{tabular } [ c]{c}% $ c_{0}$\\ $ c_{1}$\\ $ \vdots$\\ $ c_{n}$% \end{tabular } \right]\end{aligned}\ ] ] now pick @xmath206 $ ] as a solution to@xmath207{cccc}% $ 1 $ & $ 1 $ & $ \cdots$ & $ 1$\\ $ k_{0}$ & $ k_{1}$ & $ \cdots$ & $ k_{n}$\\ $ \frac{k_{0}^{2}}{2!}$ & $ \frac{k_{1}^{2}}{2!}$ & $ \cdots$ & $ \frac{k_{n}^{2}% } { 2!}$\\ $ \vdots$ & $ \vdots$ & $ \ddots$ & $ \vdots$\\ $ \frac{k_{0}^{n}}{n!}$ & $ \frac{k_{1}^{n}}{n!}$ & $ \cdots$ & $ \frac{k_{n}^{n}% } { n!}$% \end{tabular } \right ] \left [ \begin{tabular } [ c]{c}% $ c_{0}$\\ $ c_{1}$\\ $ \vdots$\\ $ c_{n}$% \end{tabular } \right ] = \left [ \begin{tabular } [ c]{c}% $ 0$\\ $ \vdots$\\ $ 1$\\ $ \vdots$\\ $ 0$% \end{tabular } \right ] \label{linenumdiffcoeffmatrix}%\ ] ] which is possible since the @xmath208 are different , so the matrix is invertible , as is seen using the vandermonde determinant@xmath209 then we must have@xmath210{c}% $ f\left ( x\right ) $ \\ $ tf^{\prime}\left ( x\right ) $ \\ $ \vdots$\\ $ t^{n}f^{\left ( n\right ) } \left ( x\right ) $ \\ $ t^{n+1}$% \end{tabular } \right ] ^{t}\left [ \begin{tabular } [ c]{l}% $ 0$\\ $ \vdots$\\ $ 1$\quad($k$-th position)\\ $ \vdots$\\ $ 0$\\ $ \frac{1}{\left ( n+1\right ) ! } \overset{n}{\underset{i=0}{% % tcimacro{\tsum } % % beginexpansion { \textstyle\sum } % endexpansion } } c_{i}k_{i}^{n+1}f^{\left ( n+1\right ) } \left ( \xi_{i}\right ) $ % \end{tabular } \right ] \\ & = t^{k}f^{\left ( k\right ) } \left ( x\right ) + \frac{t^{n+1}}{\left ( n+1\right ) ! }
\overset{n}{\underset{i=1}{% % tcimacro{\tsum } % % beginexpansion { \textstyle\sum } % endexpansion } } c_{i}k_{i}^{n+1}f^{\left ( n+1\right ) } \left ( \xi_{i}\right ) \text{.}%\end{aligned}\ ] ] therefore@xmath211 for @xmath200 which satisfy @xmath212 where@xmath213 this @xmath214 formula shows how truncation error may be decreased by increasing @xmath33 without shrinking @xmath32 , thus combatting round - off error at the expense of increased computation of sums .
thornton also points out that the @xmath208 may be chosen as complex values when @xmath25 is analytic ( as is the case with our gaussians ) .
this gives us another opportunity to mitigate round - off error , since a greater quantity of regularly - spaced nodes @xmath208 can be packed into an epsilon ball around zero in the complex plane than on the real line .
as final note we mention there have been numerous advances to the present day in inverting the vandermonde matrix .
we mention only the earliest application to numerical differentiation @xcite which gives a formula in terms of the stirling numbers .
greg leibon , daniel rockmore & gregory chirikjian , a fast hermite transform with applications to protein structure determination , proceedings of the 2007 international workshop on symbolic - numeric computation , acm , new york , ny , pp .
117 - 124 , 2007 .
j. madrenas , m. verleysen , p. thissen , and j. l. voz , a cmos analog circuit for gaussian functions , ieee transactions on circuits and systems - ii : analog and digital signal processing , vol .
43 , no . 1 , 1996 | linear combinations of translations of a single gaussian , @xmath0 , are shown to be dense in @xmath1 . two algorithms for determining the coefficients for the approximations
are given , using orthogonal hermite functions and least squares .
taking the fourier transform of this result shows low - frequency trigonometric series are dense in @xmath2 with gaussian weight function . _
key words : hermite series , gaussian function , low - frequency trigonometric series _
ams subject classifications : 41a30 , 42a32 , 42c10 = 1 |
newly born neutron stars are expected to be quite different from ordinary neutron stars , which are a commonly accepted model of radio pulsars . just after their formation in a gravitational collapse of a massive stellar core , a future neutron star is a hot ( internal temperature @xmath2k ) , lepton rich ( one electron for three nucleons ) object - in view of these specific features it is called a _
protoneutron star_. after some seconds , a protoneutron star transforms into a neutron star , loosing the lepton number excess via emission of neutrinos trapped in dense , hot stellar interior .
the short period of neutrino burst is of paramount importance for the fate of collapsing star .
it is commonly accepted , on the basis of numerical simulations of gravitational collapse of stellar cores , that the prompt - shock mechanism fails to make a successful explosion corresponding to a sn ii .
it is currently believed , that neutrino flow from a protoneutron star can revive the stalled shock .
it is thus clear that the structure and evolution of a protoneutron star , formed as a product of gravitational collapse , are crucial for the success ( or failure ) of the accretion shock revival , and therefore , for a successful sn ii explosion . while the lifetime of a protoneutron star ( seconds ) is negligibly short compared with the lifetime of a neutron star at the radio pulsar stage , it is some three orders of magnitude longer than the dynamical timescale for these objects ( milliseconds ) . in view of
this one can study the bulk evolution of protoneutron stars in the quasistationary approximation , treating convection just as an additional transport process in the stellar interior .
static properties of protoneutron stars , under various assumptions concerning their composition and equation of state of hot , dense interior , were studied by numerous authors ( burrows & lattimer 1986 , takatsuka 1995 , bombaci et al .
1995 , bombaci 1996 , prakash et al .
1997 , gondek et al .
1997 ) .
the formation of a protoneutron star ( pns ) occurs on a dynamical timescale .
it involves compression , with an overshoot of central density , and a hydrodynamical bounce .
pns is expected to begin its life pulsating around its quasistatic equilibrium .
radial pulsations of pns were recently studied in ( gondek et al .
1997 ) .
if the presupernova core was rotating - even very slowly , the pns is expected to be rather rapidly rotating .
this is an inevitable consequence of the conservation of the angular momentum of the collapsing core , which during a fraction of a second shrinks to about one percent of its initial radius .
the effects of rotation on the process of collapse of massive stellar core were studied in mnchmeyer & mller ( 1989 ) , janka & mnchmeyer ( 1989 ) and yamada & sato ( 1994 ) .
rapid rotation can have a strong effect on the structure of a pns .
the effect of rotation is stronger than in the case of cold neutron stars ( ns ) , because a hot pns , due to the thermal and neutrino trapping effects , is a more extended object than a cold configuration of the same baryon ( rest ) mass . rapid uniform rotation of pns was studied recently in ( hashimoto et al .
1995 , goussard et al .
1997 ) , who considered also the consequences of the rotational properties of pns for cold ns , into which a pns transforms due to deleptonization and cooling .
uniform rotation of pns should be considered as a first step in the study of their rotational properties .
even if the presupernova core was uniformly rotating , collapse would generate a significant amount of differential rotation of pns ( during collapse , each collapsing annulus conserves to a very good approximation
its angular momentum with respect to the rotation axis ) .
numerical simulations of collapse of rotating stellar cores indicate a characteristic distribution of angular velocity within a newly formed pns : angular velocity of the matter , which is approximatively constant on coaxial cylinders , decreases with increasing distance from the rotation axis ( mnchmeyer & mller 1989 , janka & mnchmeyer 1989 ) .
evolutionary timescales of a pns are at most seconds ; actually , the lifetime of a pns is shorter than about twenty seconds ( the time needed for the hot stellar interior to deleptonize ) .
the only mechanisms which we could contemplate as the means of a sufficiently rapid transport of angular moment on such a short timescale are related to the superstrong magnetic fields , convection and , maybe , turbulent viscosity .
our knowledge of the quantitative role of these effects , which could rigidify rotation of pns , is at present insufficient to make reliable estimate of rigidification timescale . on the other hand ,
the viscous timescale , related to the transport of neutrinos within the hot , neutrino opaque interior of a pns , is much longer than the pns lifetime . in the present paper
we generalize our previous calculations of the models of rapidly rotating pns ( goussard et al .
1997 ) to the case of differential rotation .
in particular , we study to what extent the presence of differential rotation modifies the conclusions obtained assuming uniform rotation of pns , referring to the limitations imposed on the periods of solitary pulsars formed directly from pns
. existing calculations of rapid differential rotation were restricted to the case of _ cold _ ns ( wilson 1972 , komatsu et al .
1989 ) , and used schematic ( unrealistic ) equations of state of @xmath3 neutron star matter .
our models of hot , differentially rotating pns are constructed using a realistic eos of dense hot matter , taking into account both the effects of high temperature , and of trapped neutrinos .
the model of the dense , hot interior of pns is described in sect .
we begin with a short description of the assumed thermal and lepton structure .
then we present the equation of state of the pns matter , corresponding to various stages of the pns evolution .
we discuss also characteristic timescales , relevant for pns , and justify the assumption of stationarity ( hydrostatic equilibrium ) of rotating pns .
the mathematical formulation of the problem of stationary differential rotation of pns is presented in sect .
3 , where we also consider the problem of stability of differentially rotating pns .
the numerical method of solution of equations of stationary motion is described in sect .
4 , where we also discuss in some detail the problem of precision of our numerical solutions . in sect .
5 we present our models of rapidly , differentially rotating pns , and discuss in particular the mass shedding limits for these models , at various stages of pns evolution .
we study also the relation between differentially rotating pns and uniformly rotating cold ns . finally , sect .
6 contains the discussion of our results and our conclusions .
our models of pns are based on available results of ( mostly 1-d ) simulations of neutron star birth ( burrows & lattimer 1986 , janka & mnchmeyer 1989 , mnchmeyer & mller 1989 , burrows et al .
we consider pns just after it settled into a quasi - stationary state ( we neglect pulsations which could be excited due to hydrodynamical overshoot and bounce ) . as we restrict ourselves to the period after the successful revival of the standing ( accretion ) shock
, we can neglect the effects of accretion . in what follows
, we will consider three situations .
the first one corresponds to a very initial state of a pns , which is then composed of a hot shocked envelope , with entropy per baryon ( in the units of @xmath4 ) @xmath5 up to some baryonic density @xmath6 , and an unshocked core , characterized by @xmath7 above some density @xmath8 .
the layer with @xmath9 constitutes a transition region between the unshocked core and the shocked envelope .
our standard model of a pns will have baryon ( rest ) mass of @xmath10 .
then , the shocked envelope will contain about @xmath11 of the baryon mass .
both core and envelope are neutrino opaque - a neutrino transparent outer layer of the envelope has a negligible mass , and was not included .
we will hereafter call such a protoneutron star configuration an early - type pns ( epns ) .
the second situation ( established after 0.2 - 0.5 s , see burrows & lattimer 1986 , keil et al .
1996 ) will correspond to a thermally homogeneous pns . at this stage ,
the thermal state of pns interior ( below the neutrinosphere ) is characterized by @xmath12 , and some characteristic lepton fraction @xmath13 .
this situation will be referred to as a late - type pns ( lpns ) .
further cooling and deleptonization transforms pns into a ns of the same @xmath14 .
the structure of a sufficiently cool ns ( internal temperature @xmath15 k ) is determined essentially by the @xmath3 eos ; due to strong degeneracy thermal effects in the eos of the dense ns interior can be completely neglected .
this will correspond to a third situation considered by us .
the three situations , described in this paragraph , correspond to quite different eos of the stellar interior , described below .
we will assume a well defined `` neutrinosphere '' which separates a hot , neutrino - opaque interior from colder neutrino transparent envelope .
hot dense matter in the hot , neutrino opaque interior is composed of nucleons ( both free and bound in nuclei ) and leptons ( electrons , positrons , and neutrinos and antineutrinos of all flavors ; for simplicity we do not include muons )
. nucleon component of hot matter will be described by one of the models developed by lattimer & swesty ( 1991 ) , corresponding to the incompressibility of symmetric nuclear matter at saturation density @xmath16mev .
the local state of matter in the hot interior is determined by the baryon ( nucleon ) number density @xmath17 , the net electron fraction @xmath18 , and the net fraction of trapped electron - neutrinos , @xmath19 .
an initial stage of pns evolution is characterized by a significant trapped lepton number @xmath20 . under these conditions , both electrons and electron neutrinos
are strongly degenerate .
the deleptonization , due to the @xmath21 diffusion outward ( driven by the gradient of the chemical potential of @xmath21 ) and subsequent emission from their neutrinosphere , implies a strong decrease of @xmath13 on a timescale of seconds ( sawyer & soni 1979 , prakash et al .
1997 ) . after the deleptonization ( which for simplicity is assumed to be complete , with @xmath22 ) , there is no trapped lepton number , so that neutrinos trapped within the hot interior do not influence the beta equilibrium of nucleons , electrons and positrons . the neutrino diffusion outward ( towards the neutrinosphere ) is then driven by the temperature gradient .
the neutrinosphere is determined by the condition that optical thickness of the layer above it , for electron neutrinos with mean energy corresponding to @xmath23 at the neutrinosphere , be unity .
the effect of high @xmath24 on the eos of the shocked envelope of a pns can be seen by comparing the pressure - mass density dependence in the low and high entropy cases ( fig .
1 ) . for @xmath25
the pressure increases by more than an order of magnitude at @xmath26 , as compared to the case with @xmath27 ; the relative effect of @xmath24 on @xmath28 decreases with decreasing density . above @xmath29 , the effect of trapped lepton number on the eos of shocked envelope with @xmath25 is much smaller than that of @xmath24 .
actually , our models of epns were constructed only with a @xmath22 shocked envelope .
this was done mostly for the sake of simplicity , and also to counterbalance some oversimplifications of our model with constant trapped lepton fraction , which would clearly overestimate the role of neutrinos ( in numerical simulations , @xmath13 decreases outward , c.f .
, keil et al . 1996 ) . however , we think that a more detailed treatment of the shocked envelope would not change our main conclusions . on one hand , the layer with @xmath30 , in which the assumption of @xmath22 could lead to a significant underestimating of pressure ( fig .
1 ) , contains a very small fraction of the stellar mass . on the other hand
, the inclusion of @xmath31 would further stiffen the eos of the shocked envelope , and could only make our conclusions concerning the role of the shocked envelope even stronger ( see sect . 5.1 ) .
the eos of pns evolves with time , due to the deleptonization , which changes the composition of the matter , and due to temperature change .
however , significant changes of both @xmath23 and @xmath13 within the hot , neutrino opaque interior occur on a timescale of seconds , if only diffusive processes are operating .
the evolutionary timescales can be shortened , if convective transport is included .
still , these evolutionary timescales are orders of magnitude longer than the dynamical timescale , governing the readjustment of pressure and gravity forces .
our treatment of pns is based on the assumption , that convective motions do not influence directly the bulk mechanical equilibrium of rotating pns . for the convective motions to be not important dynamically
, convective matter velocities should be significantly smaller than the sound velocity within the neutrino opaque pns interior .
velocity of sound , @xmath32 , can be calculated from the eos of the pns matter using @xmath33 where @xmath34 is the mass - energy density , and the adiabatic index @xmath35 is defined by @xmath36 in the case of cold ns matter the appropriate @xmath35 is that calculated at a fixed @xmath37 ( gondek et al .
1997 ) .
in fig.2a we plot the values of @xmath38 versus matter density , within the shocked envelope of a pns , and compare these values with those corresponding to different physical conditions in this density interval .
the sound velocity in the shocked envelope ( @xmath25 ) is significantly higher than that corresponding to the matter at @xmath27 .
strongest effect of high @xmath24 is seen for @xmath39 .
trapped lepton number increases @xmath32 , but its effect is relatively small as compared to that of @xmath25 for @xmath40 .
sound velocity in the shocked envelope is significantly higher than the velocities of convective motions at the early stage of evolution of a protoneutron star obtained in numerical calculations ( keil et al .
this conclusion is even stronger in the case of the unshocked core ( fig .
the effect of finite entropy on the value of @xmath32 is there negligible , because of the strong degeneracy of the matter , while trapped lepton number slightly decreases it .
a very important timescale is that of viscous dissipation of differential rotation . in the absence of convection and turbulence ,
shear viscosity is due to the diffusion of neutrinos .
we will estimate the neutrino viscosity under the conditions prevailing in the dense hot core of pns , using the general scheme of van den horn & van weert ( 1981a , b , 1984 ) for the calculation of transport coefficients of neutrinos trapped in dense , hot matter , combined with neutrino mean free paths as quoted in burrows & lattimer ( 1986 ) .
we obtain three expressions , corresponding to three different physical regimes for dense hot matter with trapped neutrinos .
the earliest epoch corresponds to strongly degenerate electron neutrinos . in the dense pns core ,
nucleons are degenerate , and neutrino viscosity in this regime ( referred to as @xmath41 ) , resulting from transport of degenerate electron neutrinos , is @xmath42 where @xmath43 ( our formula is in good agreement with the results obtained by goodwin & pethick ( 1982 ) , who used the methods of the theory of fermi liquids ) .
after deleptonization , neutrinos become non - degenerate , and the neutrino viscosity in degenerate nucleon gas becomes somewhat lower , @xmath44 finally , at subnuclear densities we can have non - degenerate neutrinos trapped in a dense hot nucleon gas , which corresponds to @xmath45 in all cases the viscous timescales within the stellar interior , @xmath46 , with @xmath47 , are much longer than the lifetime of a pns .
both magnetic field effects and local turbulence could shorten viscous timescales , but they are expected to be still much longer than the dynamical timescale .
the above discussion shows , that evolutionary timescales within a pns , resulting from the transport processes in the stellar interior , are significantly longer than the dynamical timescale for pns . in view of this , we are able to decouple the pns evolution from its dynamics .
we can then treat pns rotation in a stationary approximation , with a well defined eos of the pns matter .
the problem of the calculation of the stationary state of differentially rotating cold neutron stars , within the framework of general relativity , was considered previously by some authors ( komatsu et al .
1989 , wilson 1972 , 1973 ) , who restricted themselves , however , to polytropic ( and hence unrealistic ) eos of dense matter . here
we will extend the methods used at @xmath3 to the case of hot pns with realistic eos .
we will use the notation and formalism developed in bonazzola et al .
( 1993 ) , hereafter referred to as bgsm .
as we already mentioned in paper i , one of the problems introduced by finite temperature and differential rotation , if one does not make any further assumption on the equilibrium configuration , is that the equation of stationary motion does not have a first integral ( bardeen 1972 ) .
this equation reads ( eq .
( 5.8 ) of bgsm ) : + @xmath48 where @xmath28 is the pressure , @xmath34 the mass - energy density , @xmath49 the lapse function appearing in the space - time metric , @xmath50 is the lorentz factor due to rotation , @xmath51 is the angular velocity and @xmath52 where @xmath53 is the fluid 4-velocity .
this equation is the general relativistic equivalent of a well - known newtonian formula ( see e.g. tassoul 1978 ) .
it is straightforward to show that a _ sufficient _ condition for ( [ statmot ] ) to be integrable is @xmath54 eq.([statmot ] ) can then be rewritten @xmath55
= -f \partial_{i } \omega~ , \ ] ] where the heat function @xmath56 is defined by @xmath57 which is not to be confused in non - isentropic cases with the logarithm of the dimensionless relativistic enthalpy per baryon @xmath58 $ ] ( see paper i ) . then ,
as in bgsm ( 5.1 ) , we apply the schwarz theorem and find that either @xmath59 or @xmath60 this corresponds to the well - known result that , assuming barotropic equilibrium ( @xmath61 ) , the rotation of the star is either uniform or cylindrical .
being interested in differential rotation , we restrict ourselves to the case ( [ fom ] ) .
the rotation law is obtained by solving eq .
( 5.17 ) of bgsm ( we correct here a misprint of this paper ) : @xmath62 where @xmath63 , @xmath64 are metric potentials and @xmath65 is the shift vector ( see bgsm for a precise definition ) .
one then obtains the first integral of motion in the form : @xmath66 this enables us to calculate @xmath56 in the whole star and thus the other thermodynamic quantities , using the @xmath56-tabulated eos of the form @xmath67 , @xmath68 , @xmath69 ( see 4.2 ) . finally ,
in the case of general relativistic thermal equilibrium , one can show , in the context of the eckart ( 1940 ) or carter ( 1983 ) theory of heat diffusion , that @xmath70 where @xmath71 is the redshifted temperature which would be measured by a distant observer , corresponds to thermal equilibrium ( no heat flux ) .
such a @xmath23-profile leads to a simple form for the first integral of motion : @xmath72 where @xmath73 is the baryon chemical potential .
the @xmath23-profile is then determined from @xmath74 which allows us to use the previous formulation ( [ firstint ] ) of the first integral of motion .
there is obviously an infinite number of degrees of freedom in the choice of the @xmath75-law .
however , we wish to have a law which : ( 1 ) approximately reproduces the @xmath51-profile obtained in 2d - simulations , i.e. essentially decreases toward the edge of the star ; ( 2 ) is a `` simple '' law , in order to limit the number of free parameters .
we thus chose the rotation law used by komatsu et al .
( 1989 ) : @xmath76 where @xmath77 is the central angular velocity and @xmath78 is a free parameter , with the dimension of the square of a length ( denoted @xmath63 in komatsu et al . 1989 ) .
this simple rotation law allows for a rapid solution of ( [ solvf ] ) since it becomes then a cubic equation . in the newtonian limit
, the rotation law ( [ fkomat ] ) yields the angular velocity @xmath79 we see that @xmath80 is thus a characteristic scale of spatial variation of @xmath51 : when @xmath80 is small the rotation profile is very steep , leading to a constant specific angular momentum distribution , and when it is large the rotation is almost uniform .
numerical experiments have shown that the general relativistic profile has the same qualitative properties as the newtonian one .
note that this newtonian profile is also the equivalent , for cylindrical rotation , to the one used by mnchmeyer & mller ( 1989 ) and janka & mnchmeyer ( 1989 ) at the beginning of their calculations with spherical rotation .
it is also the one used by mller and eriguchi ( 1985 ) in their calculations of `` fizzling '' white dwarfs . differentially rotating self - gravitating bodies are subject to various dynamical instabilities , which may deeply affect their structure ( see e.g. zahn 1993 ) . in what follows ,
we study to what extent our models of hot pns are stable regarding the two main axisymmetric instabilities due to the differential character of rotation , appearing in barotropic configurations : the rayleigh and the shear instability .
we then briefly discuss the case of secular instabilities linked to the spontaneous symmetry breaking phenomenon .
it is straightforward to show that the newtonian @xmath51-profile ( [ onew ] ) satisfies the classical rayleigh stability criterion , i.e. : @xmath81 where @xmath82 is the specific angular momentum of baryonic matter .
this criterion yields also a _ sufficient _
condition for stability in the case of cylindrical rotation laws , i.e. @xmath83 and one can show that the general relativistic @xmath51-profile obtained with the law ( [ fkomat ] ) fulfils this last condition _ in the isentropic case _
( komatsu et al . , 1989 , 3.1 ) .
for the full general relativistic formulation in non - isentropic cases , we checked that all our models verified ( [ rayleigh ] ) . as this stability analysis is a local one in which the gravity does not enter , we infer that ( [ rayleigh ] ) is a good criterion even in the general relativistic case .
the case of the shear instability is much more complicated , and we will restrict ourselves to crude estimates . our configurations , in which the reynolds number ( which compares turbulent to molecular viscosity ) is of the order of @xmath84 ( see section 2.2 ) , are very likely to be unstable ( see zahn 1993 ) .
in principle , this instability could be blocked by a stable entropy or specific angular momentum stratification inside the star . however , even if the entropy and specific angular momentum gradients of our models were all stable ( apart from a thin outer layer of the star ) , these gradients are too weak to stabilize the flow .
indeed , the richardson number @xmath85 , which compares the efficiency of the stabilizing @xmath24 and @xmath82-gradients to destabilizing effect of shear , can be written in the newtonian case as @xmath86 \times \nonumber \\ & & \times \left(\frac{{\rm d}u}{{\rm d}r}\right)^{-2}\end{aligned}\ ] ] where @xmath87 is the gravitational acceleration , @xmath35 the adiabatic index , @xmath88 the mass density and @xmath89 the vorticity . in view of the fact , that our models are almost adiabatic , i.e. that @xmath90 in the whole star , and that as @xmath78 decreases the @xmath82-distribution tends to a @xmath91 one ( giving thus a marginally rayleigh - stable stratification ) , it was not a surprise to find that almost all our models verified @xmath92 ( at least in some part of the star ) , which is a sufficient condition for instability . however , the global effect of the shear instability is to suppress the phenomenon that causes it , namely , differential rotation . as the maximum angular momentum of pns decreases with decreasing differential rotation ( see 5 ) ,
the introduction of this instability can only decrease the maximum angular velocity of such stars .
a rapidly rotating pns can be affected by secular instabilities , driven by the gravitational radiation reaction ( grr instability ) or by viscous dissipation .
however , detailed calculations performed for hot uniformly rotating ns suggest that , for temperatures exceeding @xmath93 , the grr instability can only slightly decrease the maximum rotation frequency ( cutler et al .
1990 , ipser & lindblom 1991 , lindblom 1995 , yoshida & eriguchi 1995 , zdunik 1996 ) . if the instability is driven by viscous dissipation , bonazzola et al .
( 1995 ) have shown for cold uniformly rotating ns that only very stiff equations of state are likely to allow for spontaneous symmetry breaking .
as our models are differentially rotating and hot , no previous study applies to our epns and lpns . to estimate to
what extent our models could be plagued by spontaneous symmetry breaking , we used the _
newtonian _ criterion of stability @xmath94 , where @xmath23 is the kinetic energy and @xmath95 the gravitational energy of the star .
as our pns models are slowly rotating compared to cold ns , we were not surprised to find that @xmath96 for the epns models . for lpns , we found some models for which @xmath97 , which , considering the approximate criterion we used , can not be viewed as conclusive of symmetry breaking .
furthermore , inclusion of symmetry breaking effects can only decrease the maximum angular momentum of pns and thus , their maximal rotational velocity , which would only strengthen our conclusions .
we used the same code based on spectral methods as in paper i , modified to take into account differential rotation .
we refer the reader to our previous paper for a brief description of the code and tests of precision , which we made in the case of uniform rotation . here
we briefly outline the different or new features of the code .
the global relative error , evaluated by the means of the virial identity test ( see bgsm ) , is @xmath98 .
this is by two orders of magnitude better than the precision reached in paper i. such an increase of precision was possible due to new interpolation procedure used for the eos ( see next section ) .
we used two grids in @xmath99 for the star and one in the @xmath100-variable for the exterior , of @xmath101 chebychev coefficients , with @xmath102 in the interior and @xmath103 in the exterior , and one grid in @xmath104 with @xmath105 .
we checked that taking a greater @xmath101 or @xmath106 does not change the results by more than a few @xmath107 at most , which stays within the global precision reached .
let us stress that such a low number of grid points is sufficient to reach high accuracy within a spectral method ( this would not be possible in the case of a finite difference scheme ) .
the temperature drop at the `` neutrinosphere '' is not always located on the border of the internal grid in @xmath99 , which could influence the precision ( remember that the spectral methods are very sensitive to discontinuities ) .
we checked that , in fact , even when the `` neutrinosphere '' is far from the border of the grid , the global physical properties of the star did not change much for the late - type models of pns ( we found also relative variations of a few @xmath107 at most ) .
for early - type models of pns , the results were more sensitive to the location of the `` neutrinosphere '' , and we were obliged to position the grids quite precisely to reach the former precision .
the precision of the axisymmetric code is very sensitive to the thermodynamic consistency of the interpolated eos .
in particular , the following thermodynamical relation must be fulfilled to ensure a good precision of the code , @xmath108 if the above relation is only approximately fulfilled by the eos ( as is the case when various thermodynamical quantities are independently interpolated ) the equation of stationary motion can not be strictly satisfied , neither .
this influences greatly the value of the virial parameter and thus , the precision of the calculations . in order to have a consistent eos ( in the sense that ( [ eoscons ] ) is verified ) , we used an original procedure , adapted from a method derived by swesty ( 1996 ) .
once the temperature profile is chosen , the eos can be tabulated .
the heat function @xmath56 is calculated by interpolating the quantity @xmath109 in the integrand in eq.(4 ) as a function of @xmath28 using cubic splines and integrating this interpolating polynomial .
the quantity @xmath56 can then be chosen as the new interpolation variable : as this is a quantity given directly by the first integral of motion ( see 3 ) this ensures a rapid calculation of the thermodynamical quantities .
futhermore , following swesty , we use the fact that @xmath110 is known and interpolate @xmath28 and @xmath110 by means of a hermite interpolation scheme of degree 3 . the energy density is obtained at each point by derivating the interpolant and using ( [ eoscons ] ) .
other thermodynamical quantities can be interpolated independently since they do not enter this relation . in doing so
, we ensure that ( [ eoscons ] ) is valid for the whole interpolation polynomial and not only at the interpolation points . with such an interpolation procedure , we were able to lower the virial error parameter by two orders of magnitude .
early - type pns , with a high - entropy , bloated envelope , are short living structures
. their lifetime might be as short as 100 - 200 ms ( if one includes convective motions , see keil et al .
1996 ) , but this is still a long time on a dynamical timescale .
as early - type models of pns have not been investigated in paper i , we study here the static properties of these objets .
the gravitational and baryonic mass of static models versus central baryonic density are displayed in fig .
[ fig : statmass ] .
let us notice , that we have restricted ourselves to the range of masses , consistent with assumed thermal structure of pns .
as one can see , there is a minimal mass for such equilibria , as for cold ones , but it is of the order of @xmath111 , much higher than the standard minimal mass of cold ns ( @xmath112 ) .
this high minimal mass of epns is due to the fact that thermal pressure at the base of the neutrinosphere tends to blow - off the envelope of the star , and thus additional mass compared to the cold case is needed to preserve equilibrium .
this effect was already present in the models of paper i , but we were at that time interested in the maximal mass of pns and did not investigate this problem .
for late - type pns , the minimal mass is increased at most up to @xmath113 ( gondek et al . , 1997 ) .
[ cols= " < , < , < , < , < , < , < , < " , ] it should be stressed , that this qualitative conclusion is generally valid for thermally homogeneous lpns , for all considered eos of stellar interior ( isentropic with @xmath114 , isothermal , with no trapped lepton number @xmath22 , and with maximum trapped lepton fraction @xmath115 ) , and that the actual value of @xmath116 is roughly independent of the assumed eos . for @xmath117
, we can consider configurations having an angular momentum greater than the maximal angular momentum of cold ns .
such pns will necessarily eject a part of @xmath118 and @xmath14 during their evolution toward cold , uniformly rotating ns .
this will probably lead to a maximally rotating ( i.e. , rotating at its mass shedding limit ) ns surrounded by an ejected matter . for @xmath119 , evolution of pns into a cold ns
leads to the final rotation frequency which is lower than the maximal one for such a cold ns , but the difference decreases with decreasing @xmath80 .
let us consider now the case of early - type pns .
being aware of the particular behaviour of @xmath118 depicted in the previous paragraph , we scanned the whole domain of @xmath120 for the maximal angular momentum of lpns . in this domain ,
the maximal angular momentum configurations were never keplerian .
we found that @xmath121 , which is roughly independent of @xmath78 .
as the crucial factor determining the `` minimal mass limit '' is the effective gravity at the base of the shocked envelope for large @xmath51-contrast , it is not surprising to find that our results are roughly _ independent _ of the parameter @xmath78 ( thus , in some way , of the assumed rotation law ) , and depends _ only _ on the angular momentum of the star .
these limits on the total angular momentum of the early - type pns leads to limits on the period of cold ns of the same baryon mass of @xmath122 ms .
these results are relevant for ns which originate in a type ii supernova and evolve _ without significant accretion _ from a hot , differentially rotating lpns to a cold , uniformly rotating ns .
if the baryonic mass of such a pns is @xmath1 , then the maximum frequency of the resulting ns can not exceed @xmath123 hz .
this evolutionary argument thus imposes more stringent limits on the rotation rate of ns than standard calculations ( cook et al .
1994 , salgado et al .
in the present paper we studied models of rapidly differentially rotating protoneutron stars . using realistic equations of state of the protoneutron star interior , we calculated , using general relativistic equations of stationary motion , families of configurations corresponding to the main stages of evolution of protoneutron stars .
the earliest stage corresponds to two - component objects , with a low entropy unshocked core and a bloated , high - entropy envelope .
the next stage is that of a lepton rich protoneutron star , with a relatively uniform distribution of entropy per baryon within the hot neutrino - opaque core , containing degenerate neutrinos .
finally , the latest stage is that a deleptonized hot neutrino - opaque core , which contains non - degenerate neutrinos .
we considered mostly the case of a protoneutron star of the `` canonical baryon mass '' @xmath0 , which roughly corresponds to measured gravitational masses of binary pulsars .
configurations of stationary motion were calculated assuming axial symmetry , and neglecting effects of meridional circulation and convection .
such an approximation was valid , because the velocities of meridional circulation , as well as those connected with convective motion , were much smaller than the velocity of sound in the protoneutron star interior .
on the other hand , they were also much smaller than the linear equatorial velocities considered . while convective motions could be assumed to have no direct dynamical effect , they influenced the pns structure through changes in composition and temperature , through transport of heat and of lepton number , leading to the evolution of the equation of state of stellar interior .
our assumption that the local temperature was a function of local density was sufficient for the equations of stationary motion to be integrable ; stationary state of differential rotation was then that of a barotropic equilibrium .
the approximation of stationarity was crucial for the simplicity of calculations .
the shear viscosity , resulting from neutrino diffusion , implies angular momentum transport and heating of the stellar interior .
however , the estimates of the characteristic timescales , under various regimes prevailing within the hot neutrino opaque core , imply that these effects can be neglected on the timescale of the pns lifetime .
clearly , the timescale of angular momentum and energy transport can actually be significantly shortened by convective and turbulent motions within the stellar interior .
also , the presence of magnetic field , and subsequent amplifying of it due to the differential character of rotation ( winding - up of magnetic field lines ) , could contribute to the angular momentum transport and to rigidifying of the rotational motion .
however , even in this case characteristic timescales referring to the global structure of rotating star may be expected to be of the order of seconds . clearly , all these effects which are expected to be operating in a differentially rotating protoneutron star , deserve a separate study .
our models describe rotating pns after revival and lift - off of the shock , i.e. , after accretion onto pns became insignificant for mechanical equilibrium . basically , we considered two limiting scenarios of the ` liberation ' of pns from accretion .
the first case corresponds to a relatively early switching - off of accretion , so that rotating pns has still a pronounced two - component thermal structure , with an unshocked low entropy core and a shocked high entropy envelope .
the second case is that of a relatively late lifting - off of the shock , when the accretion stage was sufficiently long to allow for mixing of the low and high entropy matter , leading to a roughly isentropic pns interior .
let us notice , however , that the second scenario is , in view of very high accretion rate , quite risky for the pns survival .
huge accretion rates ( of the order of @xmath124s ) might result in a high probability of exceeding the maximum allowable mass for neutron stars , and therefore could easily lead to formation of a black hole . for a `` canonical baryon mass '' of @xmath0 , and within the considered family of differentially rotating configurations , we find a set of configurations with angular momentum exceeding the maximum one that can be accommodated by rigidly rotating cold neutron stars of the same baryon mass .
we expect , that the evolution of such differentially rotating protoneutron stars into cold , uniformly rotating neutron stars will be accompanied by ejection of matter from their equators .
such a mass shedding configuration may be expected to be susceptible to some additional instabilities , which - if breaking axial symmetry - could contribute to a further angular momentum loss via emission of gravitational radiation .
assuming that the initial differentially rotating configuration had a characteristic two - component structure , with a high entropy envelope , containing some half of the baryon mass of the protoneutron star , we get ( for a standard baryon mass of @xmath0 ) a minimum period of resulting cold neutron star @xmath125ms .
shorter rotation periods of cold neutron star of the baryon mass of @xmath1 ( such as the periods measured for the most rapid millisecond pulsars like psr 0034 - 0534 , psr 1937 + 21 , psr 1957 + 20 , see e.g. bailes & lorimer 1995 ) could then be reached only by a spin - up via accretion of a significant fraction of solar mass from an accretion disk .
we are very grateful to j - p .
zahn and s. talon for introducing us to the difficult subject of instabilities in differentially rotating fluids .
this research was partially supported by the jumelage program `` astronomie france - pologne '' of cnrs / pan and by the kbn grant no .
p304 014 07 .
the numerical computations have been performed on silicon graphics workstations , purchased thanks to the support of the spm department of the cnrs and the institut national des sciences de lunivers .
bailes m. , lorimer d. , 1995 , in `` millisecond pulsars : a decade of surprise '' , asp conf . ser .
78 , fruchter a.s . ,
tavani m. & bacher d.c .
bardeen j.m . , 1972 , in `` black holes - les astres occlus '' , les houches , ed .
de witt c. & de witt b.s . ,
gordon and breach sci . pub .
, new york bombaci i. , prakash m. , prakash m. , ellis p.j .
, lattimer j.m . ,
brown g.e . , 1995 , nucl .
phys . a , 583 , 623 bombaci i. , 1996 , a&a 305 , 871 bonazzola s. , gourgoulhon e. , salgado m. , marck j.a .
, 1993 , a&a 278 , 421 bonazzola , s. , frieben , j. , gourgoulhon , e. , 1995 , apj 460 , 379 burrows a. , lattimer j.m . , 1986 ,
apj , 307 , 178 burrows a. , hayes j. , fryxell b.a . , 1995 , apj , 450 , 830 carter b. , 1983 , in `` a random walk in relativity and cosmology '' , dadhich n. , krisna rao j. , narlikar j.v . & vishveshwara c.v .
eds . , bombay cook g.b . ,
shapiro s.l . ,
teukolsky s.a . , 1992 ,
apj , 398 , 203 cook g.b . ,
shapiro s.l .
, teukolsky s.a . , 1994 , apj , 424 , 823 cutler c. , lindblom l. , splinter r.j . , 1990 , apj 363 , 603 eckart c. , 1940 , phys .
, 38 , 919 gondek d. , haensel p. , zdunik j.l . , 1997 ,
a&a 325 , 217 goodwin b.t .
, pethick c.j . , 1982 ,
apj , 253 , 816 goussard j - o . ,
haensel p. , zdunik j.l .
, 1997 , a&a 321 , 822 ( paper i ) hashimoto m - a , oyamatsu k. , eriguchi y. , 1995 , apj 436 , 257 van den horn l.j . , van weert c.g . , 1981a , apj 251 , l97 van den horn l.j .
, van weert c.g .
, 1981b , phys .
83a , 88 van den horn l.j . , van weert c.g . , 1984 ,
a&a 136 , 74 ipser j.r .
, lindblom l. , 1991 , apj 373 , 213 janka h.t . , mnchmeyer r. , 1989 , a&a , 226 , 69 keil w. , janka h.t . , mller e. , 1996 , apj , 473 , l111 komatsu h. , eriguchi y. , hachisu i. , 1989 , mnras 239 , 153 lattimer j.m .
, swesty f.d , 1991 , nucl .
a535 , 331 lindblom l. , 1995 , apj 438 , 265 mnchmeyer r. , mller e. , in `` timing neutron stars '' , ed .
gelman h. & van den heuvel e.p.j . , kluwer acad . pub .
mller e. , eriguchi y. , 1985 , a&a , 152 , 325 prakash m. , bombaci i. , prakash m. , ellis p.j . , lattimer j.m .
, knorren r. , phys .
reports , 280 , 1 salgado m. , bonazzola s. , gourgoulhon e. , haensel p. , 1994 ,
a&a , 291 , 155 sawyer r.f . ,
soni a. , 1979 , apj 230 , 859 shapiro s.l .
, teukolsky s.a . , nakamura t. , 1990 , apj , 357 , l17 swesty f.d . , 1996 , j. of comp .
phys . , 127 , 118 takatsuka t. , 1995 , nucl .
a588 , 365 tassoul j.l . ,
1978 , `` theory of rotating stars '' , princeton university press , princeton wilson j.r .
, 1972 , apj , 176 , 195 wilson j.r . , 1973 ,
, 20 , 1082 yamada s. , sato k. , 1994 , apj , 434 , 268 yoshida s. , eriguchi y. , 1995 , apj 438 , 830 zahn j - p . , 1993 , in `` astrophysical fluid dynamics '' , les houches , ed .
zahn j - p . & zinn - justin j. , elsevier sci . pub .
zdunik , j.l . , 1996 , a&a 308 , 828 | models of differentially rotating protoneutron stars are calculated , using realistic equations of state of dense hot matter .
various conditions within the stellar interior , corresponding to different stages of protoneutron star evolution , are considered .
numerical calculations are performed within the approximation of stationary equilibrium , using general relativistic equations of stationary motion of differentially rotating , axially symmetric stars and using a numerical code based on spectral methods .
families of differentially rotating models of a given baryon mass are calculated , using a two - parameter formula describing the angular velocity profile within a rotating protoneutron star .
apart from the usual `` mass shedding limit '' , we introduce an additional `` minimal mass limit '' for differentially rotating protoneutron stars resulting from a type ii supernovae
. maximum angular momentum , which can be accommodated by a protoneutron star within these limits is calculated , for various thermal conditions in stellar interior , for a baryon mass of @xmath0 . in the case of a thermally homogeneous ( isentropic or isothermal ) neutrino - opaque interior
this maximum angular momentum turns out to be somewhat higher than that of a cold neutron star of the same baryon mass , rotating uniformly at the mass shedding angular velocity .
however , if the protoneutron star has a thermal structure characteristic of initial state , with a low entropy ( unshocked ) core , and a high entropy ( shocked ) outer half of baryon mass , the maximum angular momentum is significantly lower .
this leads to a minimum period of uniform rotation of cold neutron stars of baryon mass @xmath1 , formed directly ( i.e. without a subsequent significant accretion of mass ) from protoneutron stars with shocked envelope , of about 1.7 ms and strengthens the hypothesis that millisecond pulsars are accretion accelerated neutron stars . |
we thank the geo collaboration for the construction of geo600 and the quantum interferometry group for the development of the squeezed light source and its technology .
the authors are also grateful for support from the science and tech .
facilities council ( stfc ) , the univ . of glasgow in the uk , the max planck society , the bundesm .
f. bildung und forschung ( bmbf ) , the volkswagen stiftung , the cluster of excellence quest , the max planck research school ( imprs ) , the state of nds .
in germany , and dfg grant sfb / tr 7 . | we report on the first long - term application of squeezed vacuum states of light to improve the shot - noise - limited sensitivity of a gravitational - wave observatory .
in particular , squeezed vacuum was applied to the german / british detector geo600 during a period of three months from june to august 2011 , when geo600 was performing an observational run together with the french / italian virgo detector . in a second period squeezing application continued for about 11 months from november 2011 to october 2012 . during this time ,
squeezed vacuum was applied for 90.2% ( 205.2 days total ) of the time that science - quality data was acquired with geo600 .
sensitivity increase from squeezed vacuum application was observed broad - band above 400hz .
the time average of gain in sensitivity was 26% ( 2.0db ) , determined in the frequency band from 3.7khz to 4.0khz .
this corresponds to a factor of two increase in observed volume of the universe , for sources in the khz region ( e.g. supernovae , magnetars ) .
we introduce three new techniques to enable stable long - term application of squeezed light , and show that the glitch - rate of the detector did not increase from squeezing application . squeezed
vacuum states of light have arrived as a permanent application , capable of increasing the astrophysical reach of gravitational - wave detectors .
gravitational waves were predicted by albert einstein as a consequence of his general theory of relativity @xcite , but have not been directly measured to date . while a number of large - scale laser - interferometers such as ligo @xcite , virgo @xcite , tama @xcite , and geo600 @xcite , were ( or are , in the case of geo600 ) in operation , their sensitivities were not yet sufficient to make the first direct detection of gravitational waves ( gw ) .
one of the sensitivity limits of these instruments originates from the zero - point fluctuation of the quantum vacuum state of light .
while this noise can in principle be reduced with respect to the signal size by the use of higher laser power , this method is ultimately limited by thermal deformation of optical components and other laser power related instabilities @xcite .
it was proposed by caves @xcite that quantum noise in the shot - noise regime can also be reduced with the injection of squeezed vacuum states into the otherwise open output port of an interferometer .
it took several years until squeezed - light sources became available and to demonstrate this principle on table - top interferometers @xcite .
even more time was required to develop squeezed - light sources that made squeezed states at audio frequencies available , which is the frequency band relevant for the km - scale interferometers @xcite .
squeezing above the audio band was demonstrated in a prototype gw - interferometer @xcite with suspended mirrors , and the first demonstration of quantum - enhancement in one of the gravitational - wave observatories was done in 2010 in the geo600 detector @xcite .
squeezing of the ligo observatory in hanford , wa , was then demonstrated in late 2011 @xcite . while all of these experiments were one - time demonstrations of squeezed vacuum applied on time scales of minutes , in this work we report on the first continuous ( and ongoing ) application of squeezed light to a gravitational - wave interferometer , namely geo600 , to enhance the shot - noise limited sensitivity .
the geo detector is currently the only operational laser - interferometric gw detector worldwide and is acquiring calibrated measurement data in the so - called _ astrowatch _ mode .
data taken in this mode covers about 2/3 of the total time , when regular work as part of an incremental upgrade program called _ geo - hf _
@xcite does not take place .
the search for gravitational - wave signals in the astrowatch data will be triggered only by the occurrence of external ( e.g. electromagnetic or neutrino ) signals , as expected e.g. from a supernova in our galaxy .
geo600 @xcite is a michelson interferometer with folded 600 m long arms , using the advanced technique of dual recycling @xcite ( power- and signal recycling ) .
figure [ geosetup ] shows a simplified layout of geo600 , together with the squeezed - light source @xcite .
the squeezed states are injected into the main interferometer via a faraday isolator and must be mode - matched and aligned to the main interferometer beam .
in addition , two servo systems ( denoted in fig . [ geosetup ] as ` phase locked loop ' and ` squeezing phase feedback ' ) are necessary to assure the squeezing is at the correct optical laser frequency and detected interferometer output quadrature . through the implementation of three new techniques for alignment and squeezing phase control
, we achieve a stable and long - term application of squeezed light to a gw observatory .
figure [ strain_fig ] shows typical strain spectral densities of geo600 , comparing the non - squeezed noise floor with the squeezed noise floor , at the end of august 2012 .
shot - noise reduction is observed at frequencies higher than about 400hz .
we apply an automatic alignment system for a squeezed vacuum field for the first time .
because all of the geo600 core optics are suspended as pendula to reduce seismic perturbations , the alignment of the squeezed field into the interferometer must continuously follow the spatially drifting geo600 output field .
an error signal for this control loop is derived by sensing the optical beat between the 15.2mhz frequency shifted squeezed vacuum control field @xcite and the 14.9mhz michelson interferometer sidebands leaving the interferometer s dark port @xcite , as illustrated in fig .
[ geosetup ] .
the servo feedback signals are sent to a 2-axis piezo actuated mirror , placed in the squeezing injection path , suppressing relative motions up to a frequency of 3hz .
furthermore , we introduce a new control signal for the phase relation between the injected squeezed field and the interferometer output field .
we derive an error signal using the detected light transmitted through the output mode - cleaner ( omc ) where the squeezing coherent control sidebands @xcite beat with the spatially filtered geo600 carrier light ( tem00-mode ) .
the reduction of higher order modes in the filtered geo output beam , as well as the fact that all of the output tem00-mode carrier light is available to use , results in a superior error signal to shot noise ratio , as compared to previously used techniques .
although the detected squeezing level is currently primarily limited by optical loss @xcite , employing this new control signal already improves the strain sensitivity of geo 600 . as optical losses
are reduced , which is foreseen during the geo - hf upgrade , phase noise will play a more significant role in limiting the detected squeezing level , thus making the benefits of this technique more important .
the new control signal will also be of particular interest to the third generation laser interferometeric gw - detectors @xcite where the goal is to detect up to 10db of squeezing .
finally , the coherent control scheme for controlling the squeezer phase is extended to include a noise locking scheme @xcite to compensate for slow drifts of the optimal squeezing angle .
the implementation of this noise lock is similar to the one described in @xcite , but the combination with the coherent control scheme is used for the first time .
the demodulation phase of the squeezing phase control signal is electronically modulated at a frequency of 11.6hz . in order to maximize the detected
squeezing level , as measured in a frequency band between 3.6khz and 5.4khz of the h(t ) data stream , a control loop with unity gain at about 10mhz corrects for drifts of the optimal demodulation phase over long timescales .
the unity gain frequency of the coherent control loop is about 1khz .
the three new techniques described above were implemented in stages during two long - term geo600 science data taking periods . from june 3 to september 5 , 2011 , geo600 took part in an observational run , named s6e / vsr4 , together with the virgo detector . during the 93.3 days of this run , no detector commissioning work was performed . as a result ,
science - quality data were acquired for 84.3% ( 78.8 days ) of the total time , with squeezing applied for 53.1% ( 41.8 days ) of the science time .
the average squeezing level observed in this period was 16% ( 1.3db ) . in the second period lasting from november 17 , 2011 to october 15 , 2012 , the geo600 observatory took data in regular astrowatch mode .
compared to the s6e / vsr4 run , we achieved a much higher squeezing application time , as well as an improved observed average squeezing level .
the astrowatch time obtained in this period of 333 days was 68.2% ( 227.4 days ) with squeezed vacuum being applied for 90.2% ( 205.2 days ) of the astrowatch time with an average gain in detector sensitivity of 26% ( 2.0db ) .
for the last two weeks of this period we achieved the highest duty factor so far : squeezing was applied to 99.66% of the astrowatch science - quality data , demonstrating the success of the new control techniques . in order to characterise the effect of the squeezing application we look at the frequency dependent increase in sensitivity , as well as the stationarity of the data .
we use the omega data analysis pipeline @xcite for these investigations , which calculates the amplitude spectral density of 64-second data intervals , overlapping by 8 seconds , starting from the beginning of each science data segment . for the squeezing level estimation
we compare this amplitude spectrum with the expected un - squeezed shot - noise amplitude spectrum , evaluating for the frequency range from 3692hz to 3957hz .
this range is chosen as it is clean of spectral lines and other noise contributions , simplifying the long term calculation of the squeezing .
however , the suppression is broadly the same in the 2khz to 6khz regime ( see also figure [ strain_fig ] ) . figure [ hist_fig ] shows the result for the period from november 17 , 2011 to october 15 , 2012 .
the average gain in sensitivity from squeezed vacuum application in this period was 26% ( 2.0db ) .
this improvement in sensitivity corresponds to an increase of the observable volume of the universe by a factor of two for gravitational - wave sources in this frequency band . to investigate the noise behavior on short time scales ( i.e. 1 second and below )
the omega pipeline uses the 64s long spectra to whiten the data , and then performs a wavelet analysis to search for transients , which are thresholded and clustered into triggers .
these transients , named glitches for our purpose , degrade our ability to detect potential gravitational - wave signals with similar time - frequency parameters .
glitches occur in all gravitational - wave detectors and can have many different origins ( i.e. optical , electrical , etc . , see e.g. @xcite ) .
any new technique added to a gravitational - wave detector should not increase the glitch rate , which would compromise the potential utility of the technology .
figure [ glitch_fig ] shows histograms of signal - to - noise ratios for glitches detected during two 24 hour reference periods , one each for squeezing and no squeezing applied , constituting the first investigation of this kind .
neither the rate nor outlier amplitude of the glitches was observed to increase for the interval with squeezing applied , compared with the time without the application of squeezing .
this data without the application of squeezing was created intentionally to examine the effect of the squeezer on the glitch population in the detector data .
no other deliberate attempt was made to set aside data without squeezing , but investigation of a similar data stretch from months earlier when squeezing was unavailable for technical reasons yielded consistent results .
we will continue to apply squeezed vacuum as much as possible during the next years , while also trying to increase the level of observed squeezing .
currently , the main limitation in the level of observed squeezing comes from optical losses in the output - mode cleaner .
we conclude that we have demonstrated the application of squeezed vacuum states over a period of more than one year , enhancing the shot - noise limited sensitivity of a laser - interferometric gravitational - wave observatory .
squeezed vacuum states are now regularly used at geo600 . |
investigation of the anisotropy of the relic cmb radiation ensures a basic channel of the information about the early universe .
the continuing progress in the technology of deep - sky galactic surveys has resulted in impressive knowledge on the _ large scale structure _ in the universe as well as its evolution back to high redshifts .
both @xmath0 and lss observational data complement on supercluster scales and thus disclose the cosmological model .
geometrical experiments ( snia , deuterium ) provide additional independent tests hinting on a complex nature of the dark matter in the universe .
the cosmological model gets parameters .
there is a persistent evidence that * the underlying space is flat , * the hubble parameter is close to @xmath1 which is consistent with the age of old globular clusters if @xmath2 , * the structure in the universe has developed in the process of gravitational instability from small gaussian perturbations of matter density with a scale invariant fundamental spectrum .
the situation in cosmology is currently dominated by observations , however the accuracy of present observational data is still too low to determine a unique set of all cosmological parameters .
the main physical issue a nature and composition of dark matter is still under discussion .
a great hope of cosmologists is related with the development of ground and space based @xmath0 and far universe observations which will help to delimit the cosmological model up to a few per cent in the nearest future .
the seeds of the visible large scale structure in the universe are the _ cosmological density perturbations _ which grow due to gravitational instability in the late period of the universe expasion when density is dominated by cold non - relativistic matter ( @xmath3 ) .
these primordial cdps must have been created at the inflationary big bang epoch , as after the end of inflation the universe was hot ( radiationally dominated ) and thus gravitationally stable against small perturbations of matter density and gravitaty field .
the cdps existed in the hot universe as longwave gravitating sound waves propagating across the relativistic matter with a constant amplitude .
the cdps started growing only after the equality epoch when the matter pressure decayed .
the required fundamental cdp for galaxy clusters to be formed by now , has therefore been predicted at the level @xmath4 which was finally confirmed by cobe ( bennett et al . , 1996 ) and boomerang ( jaffe et al . , 2000 ) on this same level but at larger scales ( that in turn appeared to be in a successful consistency with another famous prediction known as the _ harrison - zeldovich _ scale invariant perturbation spectrum ) . at the same time , the cdp amplitude at mpc scale should have been enhanced as indicated by @xmath5-forest and early galaxy formation .
this can be achieved by introducing weakly interacting massive particles ( _ cold dark matter _ ) : contrary to baryionic damping , the density contrast of cold collisionless particles grows logarithmically in the hot universe boosting the power spectrum at galactic scales ( @xmath6 ) .
such an optimistic situation in understanding the structure formation has produced a strong impetus for the observational and theoretical cosmology extending dramatically by continuing progress in the improved technology of deep - sky surveys and cmb temperature detections .
the ultimate goal is to reconstruct the model parameters and cdp power spectrum from mpc up to the horizon scale , the scope straighforwardly related to the high energy physics thus capable of being observationally tested now .
three points should be emphasised in connection with the problem of lss formation : theoretical , model , and observational .
the first point means that lss formation in the universe is as fundamental problem as the creation of the universe as a whole : both features , the small cdps and friedmannian background model ( the _ cosmological principle _ ) , were produced in a unique process of inflation in the very early universe .
the theory works at very high energies ( @xmath7 gev ) whereas the observations occur in a low - energy limit ( @xmath8 ev ) . to provide a fair comparison in such a situation
we need a model to know how the perturbations evolved during the entire history of the universe .
therefore , any confrontation in cosmology between theory and observations appears to be model dependent . to determine the model we need priorities and parameters .
the former assumes gravitational instability as the principal mechanism of cdp dynamics at large scales , and spatially ramdom gaussian primordial perturbations ( both assumptions stay in good agreement with observations ) . the latter assumes knowledge of the current time when the structure is observed ( @xmath9 ) , the composition and abundance of cosmic matter ( @xmath10 ) , dark energy ( @xmath11 , quintessence , relic scalar field ) , _ cosmic gravitational waves _ ( t / s ) , the number of species of massive neutrinos and relativistic particles , the parameters of the perturbation spectra ( @xmath12 , @xmath13 , @xmath14 ) and biasing coefficients ( @xmath15 ) , the optical depth at reheating , etc .
nowadays cosmologists venture the following approach : if a family of the dark matter models is postulated as fairly simple ( with a few model parameters ) then the recovering of both the cdp power spectrum and the cosmological parameters can be provided by a given set of observational data .
below , we discuss the model under such a conventional probability sense .
there is no principal restrictions on this way : any theory can be tested to the limit if we have enough data .
the more data are available the less uncertainties remain in the theory and more parameters can be determined . actually , we are now in the beginning of data collection .
cosmologists have started the model restoration exercise taking simple theories and confronting them with the observational data available .
this development progresses with an increasing number of model parameters .
theory goes from simplicity , however nature appears complex .
until recently there were two basic theories claiming to approach the corner stones of the lss formation : inflation and defects . while being very much different in their grounds on galaxy seeds
the linear gaussian scalar perturbations in one case and the non - linear non - gaussian cosmic defects ( strings , monopoles , textures ) in other case both models presented the fundamental inevitable perturbations produced in the very early universe : the parametrically amplified quantum vacuum fluctuations of the inflaton and the topological defects left after phase transitions in the early universe , respectively .
however , a simple defect models normalised by the cmb fluctuations have failed to meet the lss formation ( watson , 1997 ) .
the reason is that the non - linear matter perturbations generate all three types of the metric fluctuations ( scalar , vortex and tensor ) which all contribute to the sachs - wolfe @xmath0 anisotropy on large angular scale , so the resulting s - mode amplitude appears insufficient to develop the observed galaxy distribution . by
now only the inflation theories ( adiabatic gaussian density perturbations ) with cold dark matter have got through ordeals of fitting the lss and @xmath0 requirements .
the principal quest here is the predicted gaussian nature of small cdps , which faces a satisfactory consistency with the real distribution of galaxies on scales @xmath16 mpc ( e.g. juszkiewicz and bouchet , 1996 ) .
the only obstacle to test reliably this important feature of the cdp seeds is the restricted depth of the available galaxy surveys .
deep galaxy surveys would also be highly welcome for clarifying another challenge of the modern cosmology : the fractal model attacking persistently the cosmological principle .
the point is that large voids seen in the field galaxy and cluster distributions extend up to scales @xmath17 mpc which is close to catalogues sizes , thus leaving a room for discussions about the value of the homogeneity scale ( sylos - labini et al . , 1997 ) .
nevertheless , it sould be emphasized that the fractal challenge is rather an issue for optical galaxies , and not for the distribution of the total mass in the universe .
the latter should be pretty homogeneous on scales larger than @xmath18 mpc to fit beautiful hubble diagrams , to say nothing on the uniform microwave and x - ray backgrounds evidencing the cosmic homogeneity at larger scales .
thus , we consider only models backed by inflationary theories .
the main tool for gaussian perturbations is the second moment of their spatial distribution related to the power spectrum : @xmath19 the dimensionless cdp spectrum @xmath20 has a simple meaning of the variance of density contrast in the scale @xmath21 ( the wave number ) within the scale band @xmath22 , it is evidently additive ( @xmath23 ) . before passing to the discussion on the spectrum observational reconstruction
let me sketch briefly the situation with cosmological parameters .
it seems that the longstanding debate on @xmath24 is approaching to its end and we are going to learn the value of the hubble constant soon . today
, two methods are very promising : measuring cepheids in distant galaxies and the supernovae type ia .
the current agreement stands for the value of @xmath25 between 0.6 and 0.7 with accuracy @xmath26 , which marginally excludes the matter dominated models with a negligible cosmological constant ( such models would require @xmath27 ) regarding the low limit for the age of the universe coming from globular clusters .
another issue is the matter abundance in the universe . at the first glance the situation is a following : again we have two groups of experiment resulting in different conclusions .
however , the agreement is possible . the first experiment deals with megaparsec scales galaxy halos , groups and x - ray clusters .
an assumption on the hydrostatic equilibrium within cluster cores yields a low dynamical mass responsible for the shape of gravitational potential on mpc scale : @xmath28 .
another important observation is a large fraction of baryons inside x - ray clusters reaching somehow @xmath29 within scale @xmath30 mpc : @xmath31 which is also consistent with a low matter density involved dynamically in small scales : if @xmath32 due to the deuterium measurements , @xmath33 at the dynamical scale ( @xmath34 mpc ) .
however , some experiments dealing with lss ( @xmath35 ) hints at a matter domination ( @xmath36 ) .
there are few arguments for it ( still more model dependent ones in comparison with the small - scale arguments ) : * the existence of substructures in the majority of galaxy clusters evidencing that the clusters are just forming systems , which is possible only in the universe dynamically close to the critical density of matter ; * the large coherence bulk velocities obviously of the cosmological origin , allowing a principal reconstruction of the total density contrast ( and , in some cases , the consistency with a model @xmath37 and the galaxy biasing factor @xmath38 ) ; * the weak gravitational lensing confirming high dynamical mass abundance around some x - ray clusters ; * the lensing argument on the fraction of splitting quasars ( still much dependent on the model assumptions ) ; * the high abundance of @xmath5 absorption systems ( the @xmath5 clouds ) ; * the evolutionary argument on the galaxy clusters number density ( still under discussion ) .
the low evolution argument about the dependence of cluster abundance on redshifts , which for many years was considered as a basic argument in favour of a low density universe , is not any more as strong as it seemed before .
it is the enacs identification of the nearby galaxy clusters ( by the dispersion velocities of their optical galaxies , see mazure et al .
( 1996 ) ) that has shown a previous underestimation of the number density of nearby abell clusters . at the moment we may state the consistency of the cluster number density evolution curve with a model @xmath37 .
a breakthrough in the problem of the model geometry restoration is being currently done while building up classical hubble diagrams ( redshifts _ vs _ apparent magnitudes ) for distant supernovae of type ia ( perlmutter at al . , 1998 ) .
contrary to galaxies , such sources look amazingly standard candles which is well supported by the distance measurements to nearby supernovae .
tested by distant supernovae ia , the deviations of the hubble diagram from the linear law hint upon the real geometry of the universe
. currently , the predictions are close to a half - to - half matter / vacuum universe ( @xmath39 ) .
however , a careful analysis of the systematic effects is required to make this method really trustable . while reconstructing cosmological parameters from the cmb temperature fluctuations we see the strongest effect coming from the locations and amplitudes of the acoustic peaks ( the sakharov oscillations ) .
the results of boomerang and maxima-1 have specified the first and second acoustic peaks which favours a spatially flat universe with high abundance of baryons : @xmath40 ( jaffe et al . , 2000 ) . however
, better understanding of the cosmological model will come after carefull analysis of the 2d and 3d acoustic peaks which is a matter of the nearest future .
finally , a possible reconciliation between the dm experiments on small and large scales can be the following : some fraction of dark matter in the universe is distributed on large scales and is not involved in galaxy halos and groups . how can it be arranged ?
today we have purely theoretical ideas on such a possibility .
the most frequently discussed are the models with non - zero @xmath41-term @xmath42 , _ mixed _ dark matter ( cold+hot , with the hot particles like massive neutrinos with a few ev rest mass and the corresponding density parameter @xmath43 ) , and a combination of both ( the @xmath41mdm models ) . in all cases cdm particles form a dynamic structure on mpc scales , while on large scales there is an additional contribution coming from light neutrinos or / and vacuum density ( the @xmath41-term affects the cosmological expansion rate and effectively shifts the power spectrum of density perturbations towards large scales ) .
a sceptical point concerning these and other cosmology models which are considered today as possible candidates for the real universe is as follows : all of them are multi - parameter ( non - minimal ) models ; the more parameters are involved , the better the situation is with the data confrontation .
does the latter tell us that we miss something important in our discussion on the formation of the universe structure
i can only conclude here saying than none of the models under discussion meets all the observational tests .
say , regarding previous examples , for @xmath44 models one can expect a large fraction of old ( relaxed ) galaxy clusters and lensed quasars , whereas the mdm models require @xmath45 and too a small abundance of x - ray clusters and high - redshift quasars . not to speak that none of the models considered now can explain a sharp turn of the spectrum slope at @xmath46 indicated by galaxy clusters distribution in space
probably , the dark matter has a more complex nature than we think , existing in the form of a relic scalar field left after inflation , or varing vacuun density , or some other form of interacting relic particles which requires special investigation .
in such a situation the observational verifications become extremely important .
the principal test here is the lss evolution in the early universe .
the cosmological models of lss formation discussed today are aimed to fit the observational data at @xmath47 .
thus we can not distinguish between the models without going into their evolution at medium and high redshifts where the models demonstrate their essential difference .
two main experiments promote a snowball progress in the reconstruction of the cdp spectrum , which was impossible in previous years : @xmath48 and direct investigation of the evolution and hierarchy of lsss .
the reason for stimulating such a progress is that these two experiments confront each other and overlap : the @xmath0 investigations go nowadays to small comoving scales up to @xmath49 mpc ( recall the corresponding angular scale in arcmin @xmath50 ) , and , at the same time , we observe a developed structure of clusters , filaments , voids , and superclusters reaching the scales @xmath51 mpc .
any reasonable assumption on the `` formation '' of large voids and superclusters in gaussian perturbation theories inevitably leads to @xmath0 predictions at @xmath52 capable of current detection .
it is a puzzle that namely this scale specifies the horizon at the decoupling era and therefore the angular scale of the first acoustic peak .
the existence of the acoustic peaks was predicted by theory long ago .
now , the time for observations has come : the first peak is detected .
the time is close when the rest peaks will be found which allow to precisely formulate and delimit the theory .
we are aware of the cosmological temperature anisotropy on large and small scales , and have some information on the whole spectrum of the cmb fluctuations .
fortunately , the small angular scales ( @xmath53 ) can be effectively tested from the earth s surface .
a hope is that the terrestrial instruments together with balloon ones as well as the map and planck surveyor satellites , will provide a sensitivity advance sufficient for the cosmological model reconstruction .
meanwhile , the situation with the cdp spectrum looks rather dramatic . at large scales
( @xmath54 mpc ) the fundamental spectrum is small in amplitude and consistent with the hz slope : @xmath55 however , at smaller scales ( @xmath56 mpc ) the power should be boosted as we observe there rich structures in spatial distribution of galaxies , clusters , @xmath5 systems , and distant sources like quasars .
the latter is especially important .
we live in the period of the decay of quasar and star formation activities ( boyle and terlevich , 1997 ) .
we thus have a unique opportunity to observe these numerous early sources tracing the past dynamics of lss formation .
this would be extremely informative as the lss perturbation amplitude , being still less than unity today at @xmath57 mpc , was ever lower in the past , which would predict a strong inverse evolution of such huge systems as superclusters and voids forming due to the gravitational instability .
it seems that quasars , the active galactic nuclei of distant galaxies , form the lss at medium redshifts ( @xmath58 ) which is provided by their correlation function and the existence of large qso groups recalling in properties ( the comoving size and abundance ) the local superclusters ( komberg et al . ,
actually , distant bright quasars may originate in merging galaxies in protoclusters , and thus can trace the sites of enhanced matter density at medium and high redshifts analogous to how galaxy clusters trace them in the near space .
the dynamical formation of these early lsss suggests that the spectral amplitude at superclusters scale ( @xmath51 mpc ) should be comparable and pretty close to that at cluster scale ( @xmath59 mpc ) , i.e. the cdp spectrum is nearly flat between those scales ( komberg and lukash , 1994 ) : @xmath60 this estimate for the spectrum shape is also backed by the local observations of galaxy and galaxy cluster distributions ( guzzo , 1991 , peacock , 1996 , einasto et al . , 1997 ) .
a strong break in the spectrum slope from the hz asymptotic ( 3 ) to the flat part ( 4 ) should have happened at supercluster scale ( @xmath51 mpc ) which is obviously a real feature of the fundamental cdp spectrum .
this signature of the god in the primordial spectrum demands its explanation in physics of the early universe .
i can not help mentioning another connection between the very early universe and the primordial perturbations .
this is a possibility to have high abundance of cosmic gravitational waves contributing to large - scale cmb anisotropy .
there are at least two reasons for such discussion .
the first one is theoretical .
inflation theory is not discriminative to any of the perturbation modes if inflation occurs at gut energies ( lukash and mikheeva , 1996 , 2000 ) : both s ( cdp ) and t ( cgw ) modes can be produced with similar amplitudes and thus comparable contribution to the cmb anisotropy , @xmath61 the second reason comes from observations .
if the scalar perturbation spectrum is blue ( @xmath62 ) then a non - zero t / s@xmath63 is required to reconcile the cobe @xmath0 measurement with the galaxy cluster abundance ( lukash et al . , 2000 ) .
the problem of t / s is fundamental but can be treated at the moment only theoretically .
a serious discussion on the observational detection of t / s could be launched after cmb polarization measurements , which would require the instrumental sensitivity @xmath64 currently marginally reachable .
as never before , cosmologists are close to recovering the real model of our universe and the post - recombination cdp spectrum directly from observations , both @xmath0 and lss , and to creating an exciting link to the physics of the very early universe .
we are going to gain the data from the advanced ground and space based cmb explorers as well as large surveys of spatial distribution of galaxies , galaxy clusters and quasars , to delimit the cosmological model with unprecedented precision .
the list of current conclusions may be incomplete : * dark energy is not curvature ( @xmath65 ) ; * quintessence is consistent with @xmath41-term ; * defects are not responsible for lss formation ; * the slope of the fundamental cdp spectrum is consistent with harrison - zeldovich one ( @xmath66 ) ; * distant supernovae ia indicate the current acceleration of the universe ( @xmath67 ) ; * @xmath0 and lss data hint on low @xmath68 ; * baryonic dark matter is indicated by the @xmath0 acoustic peaks ( @xmath40 ) ; * adiabatic @xmath41cdm models has survived ( @xmath69 ) ; * the best - fit @xmath41mdm models predict half - to - half matter / energy domination ( @xmath70 ) ( novosyadlyj et al . , 2000 ) .
in view of current cosmological discussion we can happily state a common observational agreement on four important points : * @xmath71 ( we live in a flat friedmann universe ) ; * @xmath72 ( the spectrum of density perturbations is scale free ) ; * @xmath73 ( the hubble constant is close to @xmath1 ) ; * the observed spectrum of density perturbations is best fitted by @xmath41cdm ( @xmath74 ) and @xmath41mdm ( @xmath75 ) models ; nevertheless , the discussion about baryonic and matter abundances in the universe is up on the agenda .
the critical tests in building up the true cosmology are the following : * direct detection of the neutrino rest masses and other forms of dark matter ; * determination of the locations and amplitudes of the acoustic peaks in @xmath0 ; * detection and investigation of the polarization of @xmath0 ; * further progress in data analysis of the formation and evolution of large scale structure in the universe ; * development of the geometrical tests ( supernovae ia , @xmath25 , light elements abundance ) . summarizing , we can state a prominent progress : the detalization and status of some parts of cosmology remind one an enginearing science level . still the fundamental cosmological issues waiting their solutions remain the following : * the problem of dark matter ( multicomponent , baryonic , ... ) ; * the problem of @xmath41-term ( value , the period of domination , ... ) ; * the break in the cdp spectrum slope at scale @xmath51 mpc which requires _ new _ physical explanation ( scale , physics of dark matter , why lss and cmb scales are close to each other ( @xmath76 ) ) ; * the intensity of cosmic gravity waves ( t / s , energy scale of cosmic inflation ) .
the work was supported in part by intas ( 97 - 1192 ) .
the author is grateful to the organizing committee for the hospitality . | the current status of the theoretical and observational cosmology is reviewed . |
this paper is a sequel to work of dynkin @xcite on subroot lattices of root lattices and to work of carter @xcite on presentations of weyl group elements as products of arbitrary reflections ( that means , at all possible roots , not only at roots of a fixed root basis as in @xcite ) .
a _ root lattice _ is a @xmath3-lattice @xmath1 together with a scalar product @xmath4 on the underlying real vector space and a finite set @xmath5 of _ roots _ such that @xmath6 is a generating set of @xmath1 as a @xmath3-lattice , the reflection @xmath7 at a root @xmath8 maps @xmath6 to itself , @xmath9 for any two roots @xmath8 and @xmath10 , and @xmath11 for any root @xmath8 ( definition [ t2.2 ] ) .
then the group @xmath12 is the weyl group of the root lattice .
the set @xmath6 of roots is the _ root system_. root lattices turn up in the theory of semisimple lie algebras .
a standard reference is ( * ? ? ?
the irreducible root lattices are classified and form the series @xmath13 and the exceptional cases @xmath14 and @xmath15 . section [ s2 ] recalls their classification , standard models and dynkin diagrams . because the condition @xmath11 for any root @xmath8 is unnatural in the context of a generalization of root lattices , we will consider the slightly more general notion without this condition and call it _ p.n .
root lattice_. here _
_ stands for _ possibly nonreduced_. the classification of the irreducible p.n .
root lattices contains besides the irreducible root lattices only the series @xmath16 .
dynkin @xcite classified all isomorphism classes of pairs @xmath17 where @xmath1 is an irreducible root lattice and @xmath2 is a subroot lattice ( definition [ t3.1 ] ) .
crucial is an algorithm which was rediscovered by borel and de siebenthal @xcite and which allows to construct by a sequence of two types of steps subroot lattices of a root lattice , by adding roots to and dropping roots from a given set of roots .
we call the steps ( bdds1 ) and ( bdds2 ) .
in theorem [ t3.1 ] we recall dynkin s classification and give the easy extension to the series @xmath16 . in theorem [ t3.8 ]
we carry out the algorithm explicitly for the exceptional cases and construct thus for these cases for any isomorphism class of pairs @xmath17 a representative .
this allows to calculate also the quotient groups @xmath0 and to show the following .
define the numbers @xmath18 [ t1.1 ] ( part of theorem [ t3.8 ] ) @xmath19 here the inequalities @xmath20 are quite obvious .
the inverse inequalities require the knowledge of the groups @xmath0 and an explicit execution of the algorithm with the steps ( bdds1 ) and ( bdds2 ) .
the tables 3.13.6 contain dynkin s classification , the groups @xmath0 and the numbers @xmath21 . for the series ,
the execution of the algorithm is less important than for the exceptional cases , as for the series one can associate graphs @xmath22 to sets @xmath23 of roots which are helpful for understanding such sets and the subroot lattices which they generate ( definition [ t3.5 ] and lemma [ t3.6 ] ) .
section [ s4 ] gives a proof of the following basic fact which seems to have been unnoticed up to now and which may be of some independent interest .
[ t1.2 ] ( theorem [ t4.1 ] ) let @xmath24 be a p.n .
root lattice .
let @xmath23 be any set of roots which generates the lattice @xmath1 as a @xmath3-module .
then @xmath25 contains a @xmath3-basis of @xmath1 .
the proof for the series is easy , it uses the graphs @xmath22 ( lemma [ t4.2 ] ) .
the proof for the exceptional cases is a case discussion .
thanks to the results in section [ s3 ] , it can be reduced to a discussion of just a few cases , but they require some detailed work ( lemma [ t4.3 ] and lemma [ t4.4 ] ) . theorem [ t4.1 ] is crucial for the proof of theorem [ t6.2 ] . by definition , any weyl group element @xmath26 can be written as a product of reflections at roots , @xmath27 then the tuple @xmath28 is called a _ presentation _ of @xmath26 , @xmath29 is its length , and the subroot lattice @xmath30 is called the subroot lattice of this presentation ( definition [ t5.1 ] ) . the _ length _
@xmath31 is the minimum of the lengths of all presentations of @xmath26 .
a presentation is _ reduced _ if its length is @xmath31 .
carter ( * ? ? ?
* lemma 2 and lemma 3 ) proved the following .
[ t1.3 ] ( part of lemma [ t5.2 ] ) a presentation @xmath28 of a weyl group element @xmath26 is reduced if and only if @xmath32 are linearly independent .
and then the subroot lattice @xmath2 of the presentation satisfies @xmath33 [ t1.4 ] ( definition [ t5.3 ] ) a weyl group element @xmath26 is a _ quasi coxeter element _ if it has a reduced presentation whose subroot lattice is the full lattice @xmath1 .
it is a _ strict quasi coxeter element _ if the subroot lattice of any reduced presentation is the full lattice @xmath1 . in the homogeneous cases , quasi coxeter elements and strict quasi coxeter elements
agree . in these cases ,
the definition of quasi coxeter elements is due to voigt ( * ? ? ?
3.2.1 ) . in the inhomogeneous cases ,
definition [ t1.4 ] is new .
theorem [ t5.6 ] gives the classification of the quasi coxeter elements and the strict quasi coxeter elements for all irreducible p.n .
root lattices . in the homogeneous cases
this is an easy consequence of the results of carter @xcite .
but in the inhomogeneous cases and especially in the case @xmath34 , there is some additional work to do ( lemma [ t5.7 ] ) .
any weyl group element can be written as a product of strict quasi coxeter elements for a suitable orthogonal sum of irreducible subroot lattices .
the tables 7 - 11 in @xcite give for any conjugacy class only one subroot lattice @xmath2 and only one presentation as a strict quasi coxeter element for this subroot lattice .
different presentations of one weyl group element @xmath26 may have different subroot lattices @xmath2 and @xmath35 .
only @xmath36 is clear , due to lemma [ t1.3 ] .
theorem [ t5.10 ] complements @xcite and gives for the exceptional cases all subroot lattices of presentations and all presentations as quasi coxeter elements for these subroot lattices . here
the case @xmath34 is more difficult than the cases @xmath37 and @xmath38 .
the theorems [ t5.10 ] , [ t5.6 ] and [ t3.1 ] together allow to recover the complete classification of all conjugacy classes of weyl group elements for the irreducible p.n .
root lattices in @xcite and provide additional information .
section [ s6 ] studies nonreduced presentations of weyl group elements .
define the numbers @xmath39 it is easy to see @xmath40 .
[ t1.5 ] ( theorem [ t6.2 ] ) @xmath41 the proof builds on theorem [ t4.1 ] and some additional arguments especially for the cases @xmath42 and @xmath34 . theorem [ t6.2 ] has an application to _ extended affine root lattices _ in section [ s7 ] .
they had been defined by k. saito ( * ? ? ?
* ( 1.2 ) and ( 1.3 ) ) , see also @xcite@xcite and definition [ t7.1 ] .
one simply replaces in the definition of a p.n .
root lattice the scalar product by a positive semidefinite bilinear form @xmath43 .
then the quotient @xmath44 becomes in a natural way a p.n .
root lattice .
any element @xmath45 induces an element @xmath46 .
presentations and quasi coxeter elements in an extended affine root lattice are defined as in a p.n .
root lattice .
the simple lemma [ t7.4 ] gives for a quasi coxeter element @xmath45 the inequalities @xmath47 gives a constraint on the elements @xmath48 which are induced by quasi coxeter elements .
theorem [ t6.2 ] says @xmath49 , and theorem [ t5.10 ] allows to calculate this number .
this section recalls some basic facts on root systems .
a standard reference is ( * ? ? ?
vi ) . though we follow the more recent notations and call _ root systems _ what is call there _ reduced root systems_. we call _ p.n . root systems _ ( _ p.n . _ for _ possibly nonreduced _ ) what is called there _ root systems_. we include the p.n .
root lattices because the condition below , which distinguishes root systems , is not necessarily preserved if one goes from an _ extended affine root lattice _
( see section [ s7 ] ) to a quotient lattice .
[ t2.1 ] ( i ) a free @xmath3-module @xmath1 of rank @xmath50 is called a _
lattice_. then @xmath51 , @xmath52 and @xmath53 .
let @xmath1 be a lattice and @xmath54 be scalar product on @xmath55 .
for @xmath56 and @xmath57 define @xmath58 then @xmath59 is a reflection
. two reflections @xmath60 and @xmath61 satisfy @xmath62 [ t2.2 ] ( a ) a _ p.n .
root lattice _ is a triple @xmath24 where @xmath1 is a lattice , @xmath4 is a scalar product , and @xmath5 is a finite set such that the following properties hold . @xmath63
the elements of @xmath6 are the _ roots _ , and @xmath6 is a _ p.n .
root system_. the finite group @xmath64 is the _ weyl group_.
\(b ) a _ root lattice _ is a p.n .
root lattice @xmath24 which satisfies additionally the condition : @xmath65 then @xmath6 is a _ root system_.
\(c ) ( lemma ) the orthogonal sum of several ( p.n . )
root lattices is ( in a most natural way ) a ( p.n . )
root lattice .
\(d ) a ( p.n . )
root lattice is _ irreducible _ if it is not isomorphic to the orthogonal sum of several ( p.n . )
root lattices .
the classification of p.n . root lattices and of root lattices is as follows .
again , a standard reference is ( * ? ? ?
vi ) . [ t2.3 ] ( a ) any ( p.n . )
root lattice is either irreducible or isomorphic to an orthogonal sum of several irreducible ( p.n . )
root lattices .
\(b ) if @xmath24 is an irreducible ( p.n . )
root lattice then also @xmath66 for any @xmath67 is an irreducible ( p.n . )
root lattice .
two irreducible ( p.n . )
root lattices are of the same _ type _ if they differ up to isomorphism only by such a scalar @xmath68 .
\(c ) the types of irreducible p.n .
root lattices are given by 5 series and 5 exceptional ones with the following names , @xmath69 all except @xmath16 are root lattices .
\(d ) the following list presents one irreducible p.n .
root lattice of each type .
always @xmath70 for some @xmath71 . here
@xmath54 is the restriction to @xmath55 of the standard scalar product on @xmath72 , and @xmath73 is the standard on - basis of @xmath72 .
@xmath74 @xmath75 [ t2.4 ] ( i ) the p.n .
root lattices above have roots of the following lengths , @xmath76 the root lattices of types @xmath77 have only roots of one length and are therefore called _ homogeneous_. the root lattices of types @xmath78 and @xmath15 have roots of two lengths , _ short _ and _ long _ roots .
root lattices @xmath16 have roots of three lengths , short , long and _ extra long _ roots .
\(ii ) in the tables 3.1 3.4 the symbols @xmath79 will denote root lattices with roots of lengths as above . there
we will also consider a few root systems with other lengths , and a few other names for some of the root lattices above : @xmath80 in the table 3.5 the roots in the root systems of type @xmath81 have lengths @xmath82 and @xmath83 . in the table 3.6 ,
roots in @xmath84 and @xmath85 have length @xmath86 , roots in @xmath87 have length @xmath83 .
\(iii ) the weyl group @xmath88 of the root lattice above of type @xmath89 acts on the basis @xmath90 of @xmath91 by permutations , @xmath92 , and @xmath93 maps @xmath94 to @xmath95 . the weyl groups of the p.n .
root lattices above of the types @xmath96 and @xmath16 coincide and act on the basis @xmath97 of @xmath98 by _ signed permutations _ , @xmath99 and @xmath100 maps @xmath94 to @xmath101 .
the weyl group of the root lattice above of type @xmath102 is the subgroup of index 2 given by the condition @xmath103 .
\(iv ) let @xmath24 be an irreducible root lattice . to any subset @xmath104 with @xmath105
we associate a _
generalized dynkin diagram _ as follows .
it is a graph with @xmath106 vertices , labelled @xmath107 . between vertices
@xmath108 and @xmath109 with @xmath110 there is no edge or an edge with additional information as follows . @xmath111 the corresponding pictures are depicted below .
other cases will not be considered .
if @xmath25 is a @xmath3-basis of @xmath1 and the diagram is connected , then the diagram encodes up to a common scalar the intersection numbers @xmath112 , and thus it determines the irreducible root system .
\(v ) the following list gives for each of the root lattices in theorem [ t2.3 ] ( d ) a _ root basis _ @xmath113 ( a @xmath3-basis of @xmath1 with additional properties @xcite ) and an additional root @xmath114 ( which is minus the maximal root with respect to the root basis ) . the diagram for the root basis is called _ dynkin diagram _ , the diagram for @xmath115 is called _ extended dynkin diagram_. the roots @xmath115 satisfy a linear relation .
for the cases @xmath116 , it is given in lemma [ t3.7 ] ( c ) .
in the case of @xmath117 , @xmath118 @xmath119 the following table gives the extended dynkin diagrams .
the dynkin diagram encodes up to a common scalar the intersection numbers @xmath112 of the basis @xmath113 , and thus it determines the irreducible root system .
the subroot lattices of an irreducible root lattice can be determined up to isomorphism by a recipe due to dynkin @xcite and borel and de siebenthal @xcite . in @xcite
also a list and additional information is given . in this section
, we will review the recipe and go through it , and thus we will recover the list .
we will extend the list in two ways .
first , we consider also the p.n .
subroot lattices of the p.n .
root lattices of type @xmath16 .
second , we will calculate for any ( isomorphism class of a ) pair @xmath17 where @xmath2 is a ( p.n . )
subroot lattice of an irreducible ( p.n . )
root lattice @xmath1 the quotient group @xmath0
. this will be helpful in section [ s4 ] and crucial in the sections [ s6 ] and [ s7 ] .
[ t3.1 ] let @xmath24 be a ( p.n . )
root lattice .
\(a ) a ( p.n . )
root lattice @xmath120 is a _ ( p.n . ) subroot lattice _ of @xmath24 if @xmath121 and @xmath122 is the restriction of @xmath54 to @xmath2 and @xmath123 .
a notation : because @xmath122 and @xmath124 are determined by @xmath2 , we will talk of the subroot lattice @xmath2 .
\(b ) a ( p.n . )
root lattice @xmath120 is the ( p.n . )
root lattice of a _
subroot system _ if @xmath121 and @xmath122 is the restriction of @xmath54 to @xmath2 and @xmath125 .
\(c ) the index of a subroot lattice @xmath2 is @xmath126\in{{\mathbb z}}_{\geq 1}$ ] .
[ t3.2 ] let @xmath24 be a ( p.n . )
root lattice .
\(i ) let @xmath121 be a @xmath3-sublattice .
define @xmath122 as the restriction of @xmath54 to @xmath2 .
define @xmath127 .
then @xmath120 is a ( p.n . )
subroot lattice if and only if it is a ( p.n . ) root lattice , and this holds if and only if @xmath2 is generated by @xmath124 as a @xmath3-module : holds for @xmath124 , and @xmath128 holds for @xmath129 because of and .
this gives .
\(ii ) if @xmath130 is any nonempty subset , then the data @xmath131 satisfy the conditions in ( i ) and are a ( p.n . )
subroot lattice .
( iii ) any ( p.n . ) subroot lattice is the root lattice of a subroot system .
if @xmath24 is a homogeneous root lattice also the inverse holds .
but if @xmath24 contains orthogonal summands which are of types @xmath132 or @xmath15 , then there are subroot lattices @xmath133 such that the subsets @xmath134 of short roots give rise to root lattices @xmath135 of subroot systems @xmath136 , such that these root lattices are not subroot lattices .
we will not work much with them , but in @xcite they are used .
\(iv ) if one erases from any of the extended dynkin diagrams one vertex , one obtains a disjoint union of dynkin diagrams .
this leads to the following recipe with two kinds of steps with which one obtains easily subroot lattices of a root lattice .
it is due to @xcite and @xcite , therefore we call the steps ( bdds1 ) and ( bdds2 )
. start with a root lattice @xmath24 .
choose a root basis @xmath23 , that is a @xmath3-basis of @xmath1 consisting of roots such that its generalized dynkin diagram ( defined in remark [ t2.4 ] ( iv ) ) is a disjoint union of dynkin diagrams .
@xmath1 decomposes uniquely into an orthogonal sum of irreducible subroot lattices , which are called the _
summands _ of @xmath1 .
choose one summand @xmath2 of @xmath1 , add to @xmath25 the unique root @xmath137 in @xmath124 which gives together with the roots in @xmath138 an extended dynkin diagram ( it is a linear combination of the roots in @xmath139 and delete from @xmath140 an arbitrary root in @xmath138 .
the new set @xmath141 defines a subroot lattice @xmath142 of @xmath1 of the same rank as @xmath1 .
choose one summand @xmath2 of @xmath1 and delete from @xmath25 an arbitrary root in @xmath138 .
the new set @xmath141 defines a subroot lattice @xmath142 of @xmath1 with @xmath143 . in both cases
@xmath144 is a root basis of @xmath142 .
therefore one can repeat the steps .
the change in the dynkin diagrams is easy to see . in the step ( bdds1 )
one extends one component to its extended version and then erases one vertex . in the step ( bdds2 )
one simply erases one vertex .
the following theorem is mainly due to dynkin @xcite , the recipe in part ( a ) is also in @xcite .
the only new ( though rather trivial ) part is the discussion of the cases @xmath16 . that will follow from lemma [ t3.6 ] below . [ t3.3 ] ( a ) let @xmath24 be a root lattice
. any subroot lattice is obtained by the choice of a suitable root basis of @xmath1 and by a suitable sequence of the steps ( bdds1 ) and ( bdds2 ) .
\(b ) the first columns of the tables 3.1 3.6 list all isomorphism classes of pairs @xmath145 where @xmath24 is an irreducible ( p.n . )
root lattice with the lengths of the roots as in theorem [ t2.3 ] ( d ) and where @xmath2 is a subroot lattice .
the tables give the name for the type of @xmath2 , where additionally the lengths of the roots of the summands of @xmath2 are taken into account .
the symbols @xmath146 from remark [ t2.4 ] ( ii ) are used .
the new notations @xmath147'$ ] and @xmath147''$ ] are explained in ( d ) below .
\(c ) with one class of exceptions , the following holds .
if @xmath145 and @xmath148 are isomorphic pairs as in ( b ) , then a weyl group element @xmath149 with @xmath150 exists .
the class of exceptions are the sublattices of @xmath102 of types @xmath151 with all @xmath152 odd . for each of those types there are two conjugacy classes with respect to @xmath153 .
\(d ) the tables 3.3 and 3.4 contain pairs @xmath154'$ ] and @xmath154''$ ] with @xmath155 for @xmath156 and with @xmath157 for @xmath38 . here
@xmath154'$ ] and @xmath154''$ ] denote ( classes in the sense of ( b ) of ) subroot lattices which are isomorphic if one forgets the embedding into @xmath1 .
but for a subroot lattice @xmath121 of type @xmath154'$ ] and a subroot lattice @xmath158 of type @xmath154''$ ] , the pairs @xmath17 and @xmath159 are not isomorphic .
this is an implication of the following properties : a subroot lattice @xmath160 of type @xmath161 for @xmath156 and of type @xmath162 for @xmath38 with @xmath163 exists , but no subroot lattice @xmath164 of type @xmath161 for @xmath156 and of type @xmath162 for @xmath38 with @xmath165 exists .
the informations in the following tables 3.1 3.6 are treated in theorem [ t3.3 ] , lemma [ t3.7 ] and theorem [ t3.8 ] .
always @xmath1 is one of the p.n .
subroot lattices in theorem [ t2.3 ] ( d ) , and @xmath2 is a p.n
. subroot lattice of the type indicated . in the tables
3.23.6 it is @xmath166 . here
the roots @xmath167 for @xmath168 ( in the cases @xmath169 ) are defined in lemma [ t3.7 ] .
the quotient @xmath0 is given up to isomorphism . here
@xmath170 for @xmath171 .
for @xmath172 see theorem [ t3.8 ] . for the symbols
@xmath146 see remark [ t2.4 ] ( ii ) . here
@xmath173 , @xmath174 , @xmath175 , @xmath176 in the cases @xmath177 , @xmath178 in the cases @xmath179 , @xmath180 in the case @xmath89 , @xmath181 in the cases @xmath42 and @xmath102 , @xmath182 in the cases @xmath183 and @xmath16 , @xmath184 @xmath185 @xmath186 @xmath187 ' & 8 & 1,3 & { { \mathbb z}}\times{{\mathbb z}}_2 & 2 \\ { } [ a_5+a_1 ] '' & - & 2 & { { \mathbb z } } & 1 \\ d_6 & - & 1 & { { \mathbb z } } & 1 \\ d_4 + 2a_1 & 8 & 1,6 & { { \mathbb z}}\times{{\mathbb z}}_2 & 2 \\ a_3 + 3a_1 & 8 & 1,4 & { { \mathbb z}}\times{{\mathbb z}}_2 & 2 \\ 3a_2 & 8 & 2,5 & { { \mathbb z}}\times{{\mathbb z}}_3 & 2 \\ 2a_3 & 8 & 3,4 & { { \mathbb z}}\times{{\mathbb z}}_2 & 2 \\ a_6 & - & 3 & { { \mathbb z } } & 1 \\ 6a_1 & 8,9 & 1,4,6 & { { \mathbb z}}\times{{\mathbb z}}^2_2 & 3 \\ \hline \end{array}\end{aligned}\ ] ] @xmath188 ' & - & 1,3 & { { \mathbb z}}^2 & 2 \\ { [ } a_5 ] '' & - & 1,2 & { { \mathbb z}}^2 & 2 \\ d_4+a_1 & - & 1,6 & { { \mathbb z}}^2 & 2 \\ a_3+a_2 & - & 3,5 & { { \mathbb z}}^2 & 2 \\ 5a_1 & 8 & 1,4,6 & { { \mathbb z}}^2\times{{\mathbb z}}_2 & 3 \\ a_2 + 3a_1 & - & 4,6 & { { \mathbb z}}^2 & 2 \\ { [ } a_3 + 2a_1 ] ' & 8 & 1,3,4 & { { \mathbb z}}^2\times{{\mathbb z}}_2 & 3 \\ { [ } a_3 + 2a_1 ] '' & - & 1,4 & { { \mathbb z}}^2 & 2 \\ \hline d_4 & - & 1,6,7 & { { \mathbb z}}^3 & 3 \\ a_4 & - & 1,2,3 & { { \mathbb z}}^3 & 3 \end{array } \hspace*{0.5 cm } \begin{array}{l|l|l|l|l } { [ } a_3+a_1 ] ' & - & 1,3,4 & { { \mathbb z}}^3 & 3 \\ { [ } a_3+a_1 ] '' & - & 1,2,4 & { { \mathbb z}}^3 & 3 \\ 2a_2 & - & 1,2,5 &
{ { \mathbb z}}^3 & 3 \\ a_2 + 2a_1 & - & 2,4,5 & { { \mathbb z}}^3 & 3 \\ { [ } 4a_1 ] ' & 8 & 1,3,4,6 & { { \mathbb z}}^3\times{{\mathbb z}}_2 & 4 \\ { [ } 4a_1 ] '' & - & 2,4,6 & { { \mathbb z}}^3 & 3 \\ \hline a_3 & - & 1,2,3,4 & { { \mathbb z}}^4 & 4 \\ a_2+a_1 & - & 1,2,3,5 & { { \mathbb z}}^4 & 4 \\ { [ } 3a_1 ] ' & - & 1,3,4,6 & { { \mathbb z}}^4 & 4 \\ { [ } 3a_1 ] '' & - & 1,2,4,6 & { { \mathbb z}}^4 & 4 \\ \hline a_2 & - & 1, .. ,5 & { { \mathbb z}}^5 & 5 \\ 2a_1 & - & 1, .. ,4,6 & { { \mathbb z}}^5 & 5 \\ \hline a_1 & - & 1, .. ,6 & { { \mathbb z}}^6 & 6 \end{array}\end{aligned}\ ] ] @xmath189 @xmath190 ' & - & 3 & { { \mathbb z } } & 1\\ { } [ a_7 ] '' & 9 & 1,2 & { { \mathbb z}}\times{{\mathbb z}}_2 & 2 \\ \hline \end{array}\end{aligned}\ ] ] @xmath191 ' & - & 3,5 & { { \mathbb z}}^2 & 2\\ { } [ 2a_3 ] '' & 9 & 1,2,6 & { { \mathbb z}}^2\times{{\mathbb z}}_2 & 3\\ d_5+a_1 & - & 1,7 & { { \mathbb z}}^2 & 2\\ a_3 + 3a_1 & 9 & 1,4,6 & { { \mathbb z}}^2\times{{\mathbb z}}_2 & 3\\ d_4+a_2 & - & 1,6 & { { \mathbb z}}^2 & 2 \end{array } \hspace*{0.5 cm } \begin{array}{l|l|l|l|l } 6a_1 & 9,10 & 1,4,6,8 & { { \mathbb z}}^2\times{{\mathbb z}}_2 ^ 2 & 4\\ a_2 + 4a_1 & 9 & 4,6,8 & { { \mathbb z}}^2\times{{\mathbb z}}_2 & 3\\ a_4 + 2a_1 & - & 1,4 & { { \mathbb z}}^2 & 2\\ a_6 & - & 1,3 & { { \mathbb z}}^2 & 2\\ a_3\ ! + \ !
a_2\ ! + \ !
a_1 & - & 4,8 & { { \mathbb z}}^2 & 2\\ { } [ a_5+a_1 ] ' & - & 2,3 & { { \mathbb z}}^2 & 2\\ { } [ a_5+a_1 ] '' & 9 & 1,2,4 & { { \mathbb z}}^2\times{{\mathbb z}}_2 & 3\\ a_4+a_2 & - & 3,4 & { { \mathbb z}}^2 & 2\\ 2a_2 + 2a_1 & - & 4,6 & { { \mathbb z}}^2 & 2 \\ \hline \end{array}\end{aligned}\ ] ] @xmath192 ' & - & 2,3,5 & { { \mathbb z}}^3 & 3\\ { } [ a_3 + 2a_1 ] '' & 9 & 1,2,4,6 & { { \mathbb z}}^3\times{{\mathbb z}}_2 & 4\\ a_3+a_2 & - & 3,5,8 & { { \mathbb z}}^3 & 3\\ a_5 & - & 1,2,8 & { { \mathbb z}}^3 & 3\\ 5a_1 & 9 & 1,4,6,8 & { { \mathbb z}}^3\times{{\mathbb z}}_2 & 4\\ a_4+a_1 & - & 1,3,4 & { { \mathbb z}}^3 & 3\\ d_4+a_1 & - & 1,6,7 & { { \mathbb z}}^3 & 3\\ a_2 + 3a_1 & - & 1,4,6 & { { \mathbb z}}^3 & 3\\ 2a_2+a_1 & - & 2,3,6 & { { \mathbb z}}^3 & 3 \\ \hline \end{array } \hspace*{0.5 cm } \begin{array}{l|l|l|l|l } d_4 & - & 1,6,7,8 & { { \mathbb z}}^4 & 4\\ { } [ 4a_1 ] ' & - & 2,3,5,7 & { { \mathbb z}}^4 & 4\\ { } [ 4a_1 ] '' & 9 & 1,2,4,6,8 & { { \mathbb z}}^4\times{{\mathbb z}}_2 & 5\\ a_2\ ! + \ !
2a_1 & - & 1,2,4,6 & { { \mathbb z}}^4 & 4\\ 2a_2 & - & 1,2,3,6 & { { \mathbb z}}^4 & 4\\ a_3+a_1 & - & 1,2,3,5 & { { \mathbb z}}^4 & 4\\ a_4 & - & 1,2,3,4 & { { \mathbb z}}^4 & 4\\ \hline a_3 & - & 1,2,3,4,5 & { { \mathbb z}}^5 & 5\\ a_2+a_1 & - & 1,2,3,4,6 & { { \mathbb z}}^5 & 5\\ 3a_1 & - & 1,2,3,5,7 & { { \mathbb z}}^5 & 5 \\ \hline \end{array}\end{aligned}\ ] ] @xmath193 @xmath194 @xmath195 [ t3.4 ] ( i ) let @xmath24 be an irreducible root lattice .
theorem [ t3.3 ] ( a)+(b)+(d ) tells the following .
there is an almost 1:1 correspondence between the set of isomorphism classes of pairs @xmath145 with @xmath2 a subroot lattice and the set of unions of dynkin diagrams which are obtained by iterations of the graphical versions of the steps ( bdds1 ) and ( bdds2 ) in remark [ t3.2 ] ( iv ) , namely ( bdds1 ) : go from one dynkin diagram to the extended dynkin diagram and erase an arbitrary vertex .
( bdds2 ) : erase an arbitrary vertex .
the only exceptions are the pairs @xmath154'$ ] and @xmath154''$ ] discussed in theorem [ t3.3 ] ( d ) .
they have the same dynkin diagrams .
\(ii ) in the table 11 in ( * ? ? ?
ii , 5 ) there are two misprints .
@xmath196 has to be replaced by @xmath197 . and
one of the two @xmath198 has to be replaced by @xmath199 . in the cases of the series @xmath200 and @xmath102
, one can see the subroot lattices also in a different way , by associating a graph to a generating set @xmath201 of a subroot lattice @xmath2 .
this works also in the case of the series @xmath16 and will give the proof of the statements in theorem [ t3.3 ] for @xmath16 .
the graphs are defined as follows .
[ t3.5 ] let @xmath24 be a p.n .
root lattice in theorem [ t2.3 ] ( d ) of one of the types @xmath202 .
let @xmath130 be a nonempty subset .
it defines a p.n .
subroot lattice @xmath203 .
a graph @xmath22 with or without markings of the vertices and with one or two types of edges is defined as follows .
\(a ) @xmath1 of type @xmath89 : the graph @xmath22 has @xmath204 vertices which are labelled @xmath205 .
it has @xmath106 edges .
a root @xmath206 with @xmath207 gives an edge between the vertices @xmath208 and @xmath209 .
so , if @xmath210 and @xmath211 are in @xmath25 , there are two edges between the vertices @xmath208 and @xmath209 .
the same applies in the cases ( b ) and ( c ) .
\(b ) @xmath1 of type @xmath16 : the graph @xmath22 has @xmath212 vertices which are labelled @xmath213 .
any root @xmath214 in @xmath25 leads to a marking of the vertex @xmath208 which is called a _
short _ marking ( and which may be represented by a circle around the vertex ) .
any root @xmath215 in @xmath25 leads to a marking of the vertex @xmath208 which is called a _
long _ marking ( and which may be represented by a square around the vertex ) .
so , depending on how many of the roots @xmath214 and @xmath216 are in @xmath25 , the vertex @xmath208 has between 0 and 4 markings .
any root @xmath217 gives a normal edge between the vertices @xmath208 and @xmath209 .
any root @xmath218 gives a dotted edge between the vertices @xmath208 and @xmath209 .
so , between the vertices @xmath208 and @xmath209 there are between 0 and 4 edges .
\(c ) @xmath1 of type @xmath183 or @xmath42 or @xmath102 : the graph is defined as in the case of type @xmath16 .
( in the case of @xmath183 there are no long markings , in the case of @xmath42 there are no short markings , in the case of @xmath102 there are no markings at all ) .
the following lemma is obvious .
[ t3.6 ] consider the same data as in definition [ t3.5 ] .
the orthogonal irreducible summands of the subroot lattice @xmath2 can be read off from the graph @xmath22 as follows .
each of the following subgraphs yields a summand , which is generated by the roots which contribute via markings or edges to this subgraph .
a component of @xmath22 which has no markings and in which any cycle has an even number of dotted edges yields a summand of type @xmath219 . here
@xmath220 is the number of vertices of the component .
( an isolated vertex with no markings yields thus the summand @xmath221 ) .
the union of all components of @xmath22 which contain a vertex with a short marking yields a summand of type @xmath222 if @xmath1 is of type @xmath183 and a summand of type @xmath223 if @xmath1 is of type @xmath16 . here
@xmath29 is the number of vertices of the union of these components .
if this union is empty , we write @xmath224 if @xmath1 is of type @xmath183 and @xmath225 if @xmath1 is of type @xmath16 . a component of @xmath22 which does not contain a vertex with a short marking , but which contains a cycle with an odd number of dotted edges or which contains a vertex with a long marking yields a summand of type @xmath226 if @xmath1 is of type @xmath42 or @xmath16 .
here @xmath29 is the number of vertices of this component . a component of @xmath22 which does not contain a vertex with a marking , but which contains a cycle with an odd number of dotted edges yields a summand of type @xmath227 if @xmath1 is of type @xmath183 or @xmath102 .
here @xmath29 is the number of vertices of this component . the statements in theorem [ t3.3 ] for the cases @xmath228 and @xmath102 follow easily from this lemma and from the structure of the weyl group , which was described in remark [ t2.4 ] ( iii ) .
for the calculation of the quotients @xmath0 in theorem [ t3.8 ] , we need in the cases @xmath116 a concrete subroot lattice @xmath2 for each isomorphism class of pairs @xmath17 .
this is found in lemma [ t3.7 ] by carrying out the recipe with the steps ( bdds1 ) and ( bdds2 ) .
[ t3.7 ] let @xmath1 be an irreducible root lattice in theorem [ t2.3 ] ( d ) of one of the types @xmath116 .
additionally to the roots @xmath115 which are defined in remark [ t2.4 ] ( v ) , the following roots are considered .
@xmath229 \(a ) in the cases @xmath230 and @xmath34 , the generalized dynkin diagrams which take into account the roots @xmath115 and the roots above look as follows . ( the edges which are not
horizontal or vertical will be irrelevant except for the dotted edge between @xmath231 and @xmath232 in the dynkin diagram of @xmath38 .
it will be used once , in the construction of a subroot lattice of type @xmath233 . )
\(b ) the second and third column in the tables 3.2 3.6 encode a realization of the recipe in remark [ t3.2 ] ( iv ) , in the following way .
suppose in the tables 3.2 3.6 in the line for one subroot lattice , @xmath234 is given as the sequence @xmath235 of numbers and @xmath236 is given as the sequence @xmath237 of numbers , with @xmath238 .
one carries out @xmath239 steps ( bdds1 ) : in the @xmath29-th step one adds the root @xmath240 and then erases the root @xmath241 .
afterwards one carries out @xmath242 steps ( bdds2 ) : one erases the roots @xmath243 .
this leads to a subroot lattice of the type indicated in the first column .
\(c ) the roots @xmath167 for @xmath244 are linear combinations of the roots @xmath113 .
the linear relations are as follows .
@xmath245 * proof : * the proof of the lemma is tedious as there are many cases , but the parts ( a ) and ( c ) and most of part ( b ) are completely elementary . in part ( b ) one has to check not only that the result has the correct dynkin diagram , but also that the steps ( bdds1 ) work , i.e. that one has after adding a root an extended dynkin diagram and that then a root of this extended dynkin diagram is erased .
the details are left to the reader .
the only nontrivial part concerns the subroot lattices of types @xmath154'$ ] and @xmath154''$ ] in the cases @xmath156 and @xmath38 .
one sees that in the cases of @xmath246 , @xmath247 , the constructed subroot lattices of types @xmath154'$ ] are contained in the subroot lattice @xmath248 of type @xmath219 .
the constructed subroot lattices of type @xmath154''$ ] contain in the case @xmath156 the roots @xmath249 and in the case @xmath38 the roots @xmath250 .
the following claim shows that the constructed subroot lattices @xmath154''$ ] are not contained in subroot lattices of type @xmath161 respectively @xmath162 .
\(i ) let @xmath1 be the lattice of type @xmath156 in theorem [ t2.3 ] ( d ) .
there is no subroot lattice @xmath2 of type @xmath161 with @xmath251 .
\(ii ) let @xmath1 be the lattice of type @xmath38 in theorem [ t2.3 ] ( d ) .
there is no subroot lattice @xmath2 of type @xmath162 with @xmath252 .
\(i ) suppose that @xmath2 is a subroot lattice of type @xmath161 with @xmath251 .
these three roots generate a subroot lattice @xmath253 of type @xmath254 . by theorem [ t3.3 ] ,
up to isomorphism there is only one pair of type @xmath255 .
therefore a root @xmath256 with @xmath257 exists . but now observe @xmath258 . neither the roots in @xmath259 of type @xmath260 nor the roots in @xmath259 of type @xmath261 can serve as a root @xmath8 .
contradiction .
\(ii ) analogously to ( i ) .
( @xmath262 ) this finishes the proof of lemma [ t3.7 ] . @xmath262
[ t3.8 ] ( a ) let @xmath24 be an irreducible p.n .
root lattice , and let @xmath2 be a p.n .
subroot lattice .
the third column in table 3.1 and the fourth column in the tables 3.2 3.6 gives the isomorphism class of the quotient group @xmath0 .
\(b ) let @xmath24 be a p.n .
root lattice , and let @xmath2 be a p.n .
subroot lattice .
define the numbers @xmath263 then @xmath264 the numbers are additive , i.e. if @xmath265 and @xmath266 and @xmath267 then @xmath268 the last column of the tables 3.1 3.6 gives the numbers @xmath21 for the pairs with @xmath1 irreducible .
minimal sequences of the steps ( bdds1 ) and ( bdds2 ) for the cases with @xmath1 of type @xmath116 are given in the second and third column of the tables 3 - 2 3.6 ( see lemma [ t3.7 ] ) .
* proof : * ( a ) first we treat the cases @xmath228 and @xmath102 .
let @xmath25 and @xmath22 be as in definition [ t3.5 ] and lemma [ t3.6 ] , and let @xmath121 be the corresponding subroot lattice .
let @xmath269 ( with @xmath270 ) be the decomposition into subgraphs @xmath271 with the properties in lemma [ t3.6 ] , let @xmath272 be the subroot lattice which corresponds to the subgraph @xmath271 , and let @xmath273 then @xmath274 ( with @xmath275 for @xmath89 and @xmath276 else ) and @xmath277 the following table lists in the second and fourth line the isomorphism classes of the quotients in the first column .
@xmath278 finally , in the cases @xmath42 and @xmath102 , for any @xmath279 @xmath280 therefore @xmath0 has the isomorphism type claimed in the table 3.1 .
now we treat the cases @xmath116 .
let @xmath121 be one of the subroot lattices constructed in lemma [ t3.7 ] using the data in the tables 3.2 3.6 .
let @xmath281 respectively @xmath282 with @xmath238 be the roots in one line in the second respectively third column of these tables . then @xmath283 the denominator of the right hand side is a @xmath3-lattice of rank @xmath239 ( because @xmath284 ) and is generated by parts of those relations in lemma [ t3.7 ] ( c ) which express the roots @xmath281 as linear combinations of the roots @xmath113 .
we give one example : @xmath17 of type @xmath285 , then @xmath286 and @xmath287 .
the relation for @xmath288 gives the element @xmath289 of @xmath290 , the relation for @xmath232 gives the element @xmath291 .
therefore here @xmath292 the calculations for all other cases @xmath17 are analogous .
they are tedious as there are many cases , but elementary .
\(b ) the additivity of the numbers @xmath293 is obvious .
therefore it is sufficient to prove for irreducible @xmath1 .
the last column of the tables 3.1 3.6 can be read off from the second to last column immediately .
the first of the inequalities @xmath294 is obvious .
the second inequality follows simply from the fact that in each step ( bdds1 ) or ( bdds2 ) , one root is erased . in the cases @xmath228 and @xmath102 , one easily constructs the p.n .
subroot lattices in @xmath21 steps .
therefore then @xmath295 , and equalities hold in . in the cases
@xmath116 one observes @xmath296 in the tables 3.2 3.6 . in lemma [ t3.7 ]
@xmath297 steps of type ( bdsd1 ) and ( bdsd2 ) are used .
therefore @xmath295 , and equalities hold in . @xmath262
the purpose of this section is to prove the following theorem .
it is crucial in the proof of theorem [ t6.2 ] .
but it may be also of some independent interest .
the proof in the cases @xmath202 is an almost trivial application of the graphs in definition [ t3.5 ] and lemma [ t3.6 ] .
the proof in the cases @xmath116 is more involved .
[ t4.1 ] let @xmath24 be a p.n .
root lattice .
let @xmath23 be any set of roots which generates the lattice @xmath1 as a @xmath3-module .
then @xmath25 contains a @xmath3-basis of @xmath1 . in the case of vector spaces instead of @xmath3-modules ,
the analogous statement is trivial . for @xmath3-modules
it is not true in general , that any generating set contains a basis .
for example the lattice @xmath298 with standard basis @xmath299 has the set @xmath300 as generating set , but any two of these elements generate a proper sublattice .
the rest of this section is devoted to the proof of theorem [ t4.1 ] .
it is obviously sufficient to prove it in the cases where @xmath1 is an irreducible p.n .
root lattice . in the cases
@xmath202 , theorem [ t4.1 ] is an immediate consequence of the following lemma .
the lemma follows directly from lemma [ t3.6 ] .
[ t4.2 ] let @xmath24 be a p.n .
root lattice in theorem [ t2.3 ] ( d ) of one of the types @xmath202 .
let @xmath301 be a nonempty subset .
the properties whether @xmath25 is a generating set of @xmath1 or a @xmath3-basis of @xmath1 , will be characterized by properties of the graph @xmath22 from definition [ t3.5 ] .
\(a ) @xmath1 of type @xmath89 : @xmath25 generates @xmath1 as a @xmath3-module @xmath302 @xmath22 is connected .
@xmath25 is a @xmath3-basis of @xmath1 @xmath302 @xmath22 is a tree .
\(b ) @xmath1 of type @xmath183 : @xmath25 generates @xmath1 as a @xmath3-module @xmath302 each component of @xmath22 contains at least one vertex with a ( automatically short ) marking .
@xmath25 is a @xmath3-basis of @xmath1 @xmath302 each component of @xmath22 is a tree and contains exactly one marked vertex , and the vertex has only one ( automatically short ) marking .
\(c ) @xmath1 of type @xmath42 : @xmath25 generates @xmath1 as a @xmath3-module @xmath302 @xmath22 is connected , and it contains a marking ( automatically long ) or a cycle with an odd number of dotted edges .
@xmath25 is a @xmath3-basis of @xmath1 @xmath302 either @xmath22 is a tree and contains exactly one marked vertex and the marking is simple , or @xmath22 is connected and contains no marking , but it contains exactly one cycle and the cycle has an odd number of dotted lines .
\(d ) @xmath1 of type @xmath16 : @xmath25 generates @xmath1 as a @xmath3-module @xmath302 each component of @xmath22 contains at least one vertex with a short marking .
@xmath25 is a @xmath3-basis of @xmath1 @xmath302 each component of @xmath22 is a tree and contains exactly one marked vertex , and there is only one marking , and the marking is short .
\(e ) @xmath1 of type @xmath102 : @xmath25 generates @xmath1 as a @xmath3-module @xmath302 @xmath22 is connected , and it contains a cycle with an odd number of dotted edges .
@xmath25 is a @xmath3-basis of @xmath1 @xmath302 @xmath22 is connected and contains exactly one cycle , and the cycle has an odd number of dotted lines .
it rests to prove theorem [ t4.1 ] in the cases @xmath303 .
let @xmath1 be the root lattice of one of these types in theorem [ t2.3 ] ( d ) .
in each of these cases it is sufficient to consider a generating set @xmath304 with @xmath204 elements ( where @xmath212 is the rank of the root lattice ) .
the cases of bigger generating sets can be reduced to the case of such a set by an easy inductive argument .
there is an up to the sign unique linear combination @xmath305 it has to be shown that an index @xmath209 with @xmath306 exists .
denote by @xmath307 , @xmath308 , the subroot lattice generated by @xmath309 .
then @xmath310 if this holds for some @xmath208 then by induction on the rank of the lattice one can conclude that @xmath309 contains a @xmath3-basis of this subroot lattice . then this @xmath3-basis together with @xmath311 forms a @xmath3-basis of @xmath1 .
thus suppose that all @xmath312 .
then @xmath313=|\lambda_i|,\quad l_i+{{\mathbb z}}\alpha_i = l,\quad k_2(l , l_i)=1.\ ] ] a priori , there are 20 possible cases in the tables 3.2 3.6 , @xmath314 in the cases @xmath315 , the only possible values for @xmath316 are in @xmath317 .
the condition @xmath318 tells that at least one @xmath319 with @xmath320 exists .
the following more complicated argument gives the same conclusion in the case @xmath38 .
assume in the case @xmath38 that all @xmath316 are in @xmath321 .
define the decomposition @xmath322 of @xmath323 into two disjoint subset .
because of @xmath318 , both are nonempty .
define the subroot lattices @xmath324 then @xmath325 , and @xmath326 is a primitive vector in @xmath327 , and @xmath328 is a primitive vector in @xmath329 . but @xmath330 therefore the order of the torsion part of @xmath331 is divisible by 5 . but table 3.4 contains only one type of subroot lattices with this property , the type @xmath332 .
therefore @xmath333 for some index @xmath334 , and @xmath335 , which is impossible . therefore the assumption above that all @xmath316 are in @xmath321 was wrong . in the cases @xmath315 the type of the subroot lattices @xmath336 with @xmath337=3 $ ] is unique , in the case @xmath38 there are two possibilities , @xmath338 by renumbering the roots , we can assume @xmath339=3 $ ] .
the case @xmath340 : the roots @xmath341 and @xmath342 generate an @xmath84 lattice and thus are long . therefore @xmath343 is short .
at least one of @xmath341 and @xmath342 is not orthogonal to @xmath343 . that root and @xmath343 form a @xmath3-basis of @xmath1 .
the cases @xmath344 and the case @xmath345 : the sublattice @xmath346 ( with @xmath3-basis @xmath347 ) contains one orthogonal summand @xmath327 of type @xmath84 .
suppose that @xmath341 and @xmath342 form a @xmath3-basis of this lattice @xmath327 .
then @xmath348 is a subroot lattice with @xmath349 and @xmath350 and @xmath351 . because of @xmath352 and all @xmath353 and @xmath354 ,
the root @xmath355 is not in @xmath356 , so the lattice @xmath329 has rank 3 .
the sum @xmath357 is in the sum @xmath358 of the other orthogonal summands of @xmath359 and in the rank one @xmath3-lattice @xmath360 .
in fact , it is a generator of this rank one lattice : this is equivalent to @xmath361 where @xmath362 if @xmath363 , then @xmath364 were in the root lattice @xmath329 .
but then also @xmath365 were in the root lattice @xmath359 , thus @xmath68 would divide @xmath366 .
then @xmath367 which is not true .
the root lattice @xmath327 is either of type @xmath368 or of type @xmath369 . in both cases ,
the following lemma gives the claim .
[ t4.3 ] in both cases , at least one of @xmath370 and @xmath371 is equal to @xmath372 .
* proof : * ( a ) the case @xmath329 of type @xmath368 : embed @xmath373 as usual into a euclidean space @xmath374 with on - basis @xmath375 such that @xmath376 and @xmath377 .
a generator of the rank one @xmath3-lattice @xmath378 is obviously @xmath379 .
thus @xmath380 lemma [ t4.2 ] ( a ) applies to the graph @xmath381 .
the graph is a tree , as @xmath382 is a @xmath3-basis of @xmath329 .
at least two of the four vertices @xmath375 are leaves ,
so at least one of the three vertices @xmath383 is a leaf .
let @xmath384 with @xmath385 give the edge which contains this vertex .
the coefficient @xmath82 of this vertex must be equal to @xmath386 because the other two terms in the sum @xmath387 have no contribution to the coefficient of this vertex .
\(b ) the case @xmath329 of type @xmath369 : embed @xmath373 as usual into a euclidean space @xmath388 with on - basis @xmath383 such that @xmath389 and @xmath377 .
a generator of the rank one @xmath3-lattice @xmath360 is obviously @xmath390 .
thus @xmath391 lemma [ t4.2 ] applies to the graph @xmath381 . as @xmath341 and @xmath342 are long roots , the graph is a tree with one marked vertex .
then at least one vertex @xmath392 of the 3 vertices @xmath383 has no marking and is a leaf .
then @xmath306 for the same reason as in ( a ) . @xmath262 the case @xmath393
: we can choose a root basis @xmath394 of @xmath1 and an additional root @xmath288 such that they give rise to the extended dynkin diagram in remark [ t2.4 ] ( v ) and such that the subroot lattice @xmath359 is generated by @xmath395 .
further , we can embed @xmath396 into a euclidean space @xmath397 with on - basis @xmath398 such that @xmath399 part ( a ) of the following lemma tells how @xmath259 can be expressed using the @xmath398 .
part ( b ) solves the case @xmath400 and finishes the proof of theorem [ t4.1 ] .
[ t4.4 ] ( a ) @xmath401 ( b ) above , at least one of the @xmath402 with @xmath403 is equal to @xmath372 .
* proof : * ( a ) follows from and the relation @xmath404 see lemma [ t3.7 ] ( c ) . as @xmath405 contains @xmath406 , it contains the combination of @xmath407 on the right hand side of .
as the weyl group @xmath408 consists of all permutations of @xmath398 , the root system @xmath405 contains the right hand side of . counting the size of the right hand side
, one finds @xmath409 thus equality holds in .
\(b ) the roots @xmath410 form a @xmath3-basis of @xmath359 . the root @xmath411 must be a root in @xmath412 , so it must be @xmath413 the graph @xmath414 is a tree by lemma [ t4.2 ] ( a ) .
at least one of the roots @xmath415 is a leaf in this graph : let @xmath384 be the only edge which contains this leaf . then @xmath416 contains the only contribution to the leaf , in the right hand side of the following formula , @xmath417 thus then @xmath418 none of the roots @xmath415 is a leaf in the graph : then there are two or three leafs , and they form a subset of the set @xmath419 . for one of these leafs , the number of vertices on the path from this leaf to the branching vertex ( in the case of three leafs ) or to the unique inner vertex which is in @xmath419 , is maximal .
then the first two edges within this path , which starts at the leaf , have the coefficients @xmath402 with values @xmath420
. so @xmath372 arises .
@xmath262 this finishes the proof of theorem [ t4.1]@xmath262
carter studied and classified the conjugacy classes of the elements of the weyl groups of the irreducible root lattices . here
we will review a part of his results and extend them .
crucial are the ( in the inhomogeneous cases new ) notions of quasi coxeter elements and strict quasi coxeter elements .
the control of these elements reduces the classification of conjugacy classes of weyl group elements to the control of subroot lattices in section [ s3 ] . but first some definitions will be given .
[ t5.1 ] let @xmath24 be a p.n .
root lattice with weyl group @xmath153 .
\(a ) for any element @xmath149 any tuple @xmath421 with @xmath422 and @xmath423 is a _ presentation of @xmath26_. its _ length _ is @xmath422 .
the length @xmath424 of @xmath26 is the minimum of the lengths of all presentations . a presentation with @xmath425 is called _
reduced_. the subroot lattice of a presentation @xmath28 is @xmath426 .
the _ index _ of the presentation is the index @xmath126\in{{\mathbb z}}_{\geq 1}$ ] of the subroot lattice @xmath2 .
\(b ) an element @xmath26 is of _ maximal length _ if @xmath427 the rank of the root lattice .
\(c ) for any element @xmath149 and any @xmath428 define @xmath429 and analogously @xmath430 @xmath431 .
of course @xmath432 .
[ t5.2 ] let @xmath24 be a p.n .
root lattice with weyl group @xmath153 .
\(a ) ( * ? ? ?
* lemmata 2 and 3 ) a presentation @xmath28 of an element @xmath149 is reduced if and only if @xmath433 are linearly independent ( in @xmath434 ) .
the subroot lattice @xmath121 of a reduced presentation satisfies @xmath435 so , the subroot lattices of all reduced presentations of @xmath26 generate the same subspace of @xmath436 , and it is @xmath437 .
\(b ) ( * ? ? ?
* satz 3.2 ) ( * ? ? ?
* satz 3.2.3 ) if @xmath24 is a homogeneous root lattice , then all reduced presentations of one element @xmath149 have the same index .
the following definition of a quasi coxeter element is in the homogeneous cases due to voigt ( * ? ? ?
3.2.1 ) and in the inhomogeneous cases new . [ t5.3 ]
let @xmath24 be a p.n .
root lattice of rank @xmath50 with weyl group @xmath153 .
\(a ) an element @xmath149 is a _ quasi coxeter element _ if a reduced presentation of @xmath26 exists whose subroot lattice is the full root lattice @xmath1 .
of course then it is of _ maximal length _
@xmath438 .
\(b ) an element @xmath149 is a _ strict quasi coxeter element _ if the subroot lattice of any reduced presentation is the full root lattice @xmath1 .
of course then it is a quasi coxeter element .
[ t5.4 ] ( i ) an element @xmath26 in the weyl group of a p.n .
root lattice has many presentations .
often there are several presentations such that the isomorphisms classes of their subroot lattices are different . in the homogeneous cases
at least their indices are equal .
but in the inhomogeneous cases , even their indices can differ .
\(ii ) in a homogeneous root lattice , lemma [ t5.2 ] ( b ) implies that there the notions quasi coxeter element and strict quasi coxeter element coincide .
but in any irreducible inhomogeneous root lattice , there are quasi coxeter elements which are not strict quasi coxeter elements .
see theorem [ t5.6 ] .
\(iii ) of course , if @xmath28 is a reduced presentation of a weyl group element @xmath26 , then @xmath26 is a quasi coxeter element in the subroot lattice @xmath2 of this presentation . and
of course , any weyl group element has a reduced presentation such that it is a strict quasi coxeter element in the subroot lattice @xmath2 of this presentation .
\(iv ) let @xmath439 be the decomposition of a p.n .
root lattice into an orthogonal sum of irreducible p.n .
root lattices , and let @xmath149 be a [ strict ] quasi coxeter element .
then it decomposes into a product @xmath440 of commuting elements @xmath441 , and @xmath442 is a [ strict ] quasi coxeter element in @xmath272 .
\(v ) recall that a coxeter element in an irreducible root lattice is an element @xmath149 which has a presentation @xmath443 such that @xmath347 form a root basis . because their dynkin diagram is a tree , lemma 1 in ( * ? ? ?
6 ) implies that the products of @xmath444 in any order are conjugate . as all root bases are conjugate , all coxeter elements are conjugate .
obviously the coxeter elements are quasi coxeter elements .
it turns out that they are even strict quasi coxeter elements , see theorem [ t5.6 ] .
\(vi ) carter s work @xcite on the classification of weyl group elements gives in a direct way the classification of the quasi coxeter elements in the irreducible homogeneous root lattices and in a less direct way the classification of the strict quasi coxeter elements in the irreducible inhomogeneous root lattices . in theorem [ t5.6 ]
these classifications will be given , and also the classification of the quasi coxeter elements in the irreducible inhomogeneous root lattices .
\(vii ) recall the description of the weyl group @xmath153 in remark [ t2.4 ] ( iii ) for the root lattices of the types @xmath13 in theorem [ t2.3 ] ( d ) : @xmath92 , @xmath445 .
a signed permutation in @xmath446 will be called _ positive _ if the number of sign changes in it is even , it will be called _ negative _ if the number of sign changes in it is odd .
the subgroup @xmath447 consists of the positive signed permutations .
a signed cycle will be written as @xmath448 with @xmath449 and @xmath450 , @xmath451 with @xmath452 for @xmath110 .
it maps @xmath453 to @xmath454 for @xmath455 and @xmath456 to @xmath457 .
it is positive if @xmath458 and negative if @xmath459 .
its _ support _ is defined to be @xmath460 .
any signed permutation is up to the order a unique product of signed cycles ( = cyclic permutations ) such that their supports are disjoint and the union of the supports is @xmath461 .
they are called the signed cycles of the permutation . here
cycles of length one are used .
for example @xmath462 and @xmath463 .
[ t5.5 ] ( i ) carter classified in @xcite the conjugacy classes of weyl group elements for all irreducible root lattices .
a crucial point was the proof that any element @xmath26 can be written as a product @xmath464 where @xmath465 and @xmath466 are involutions with @xmath467 ( proposition 38 and corollary ( ii ) in @xcite ) . by ( * ? ? ? * lemma 5 ) , any involution has a reduced presentation @xmath28 which consists of pairwise orthogonal roots .
the composition of two such reduced presentations of two involutions @xmath465 and @xmath466 with @xmath467 is a reduced presentation of @xmath464 .
its generalized dynkin diagram is a graph whose cycles ( if any exist ) have all even length . in (
* theorem a ) all graphs are classified which have the following properties : the graph contains cycles , all cycles have even length , the graph is a generalized dynkin diagram of a presentation of an element @xmath464 with @xmath465 and @xmath466 as above , the subroot lattice of the presentation is the full lattice , and @xmath26 is not contained in the weyl group of a subroot system .
the graphs are labelled @xmath468 and also @xmath469 if @xmath212 is even and @xmath470 .
in fact , the graphs in @xcite are simplified by not distinguishing normal and dotted edges .
the generalized dynkin diagrams are obtained from the graphs in @xcite by replacing some edges by dotted edges such that any cycle obtains an odd number of dotted edges .
this is possible .
it turns out that the graphs @xmath468 correspond to conjugacy classes of weyl group elements , and that these include the elements with presentations giving rise to the graphs @xmath469
( @xmath212 even ) and @xmath470 .
these weyl group elements are strict quasi coxeter elements , because they are not contained in the weyl group of a subroot lattice .
they are not coxeter elements @xcite .
they and the coxeter elements are the only strict quasi coxeter elements ( theorem [ t5.6 ] below ) .
\(ii ) recall that the coxeter elements in @xmath88 are the cycles of length @xmath204 in @xmath471 , the coxeter elements in @xmath472 are the negative cycles of length @xmath212 in @xmath473 , and the coxeter elements in @xmath474 are the products of two negative cycles of lengths 1 and @xmath475 .
the products of two negative cycles of lengths @xmath29 and @xmath476 for @xmath477 $ ] form the conjugacy class @xmath478 in @xmath474 . in the second column in the tables 5.1 and 5.2 , @xmath479 and @xmath15 denote the conjugacy classes of the coxeter elements .
the root lattice of type @xmath34 contains subroot lattices of types @xmath480 and @xmath481 . in @xmath482
the symbols @xmath483 denote the conjugacy classes in @xmath482 of the coxeter elements in @xmath484 and @xmath485 and of the quasi coxeter elements of type @xmath486 in @xmath487 .
the coxeter elements of the subroot lattice of type @xmath84 in @xmath15 give rise to a conjugacy class in @xmath488 denoted by @xmath84 .
theorem [ t5.6 ] gives the classification of the quasi coxeter elements and the strict quasi coxeter elements for the irreducible p.n .
root lattices .
a good part of it is due to @xcite .
[ t5.6 ] let @xmath489 be one of the irreducible p.n .
root lattices in theorem [ t2.3 ] ( d ) .
the following tables 5.1 and 5.2 list the conjugacy classes of the strict quasi coxeter elements and in the inhomogeneous cases the conjugacy classes of the quasi coxeter elements .
see the remarks [ t5.5 ] for the notations .
* table 5.1 : * @xmath490})\\ e_6 & e_6 , e_6(a_1),e_6(a_2)\\ e_7 & e_7 , e_7(a_1),e_7(a_2),e_7(a_3),e_7(a_4)\\ e_8 & e_8 , e_8(a_1), ... ,e_8(a_8 ) \end{array}\end{aligned}\ ] ] * table 5.2 : * @xmath491})\\ f_4 & f_4,f_4(a_1 ) & f_4,f_4(a_1 ) , b_4,c_3+a_1,d_4(a_1)\\ g_2 & g_2 & g_2,a_2 \end{array}\end{aligned}\ ] ] thus the quasi coxeter elements of @xmath42 consist of the products of one or two negative cycles . * proof : * it is well known that the coxeter elements are not elements of some proper weyl subgroup
. therefore they are strict quasi coxeter elements .
the other elements listed in the second columns are strict quasi coxeter elements because of the results of carter @xcite discussed in the remarks [ t5.5 ] ( i ) . by the same results ,
any other element @xmath149 is in some proper weyl subgroup . in the homogeneous cases ,
a proper weyl subgroup is the weyl group of a proper subroot lattice .
therefore then @xmath26 is not a strict quasi coxeter element .
this completes the proof of table 5.1 . in the inhomogeneous cases ,
the fact that the second column of table 5.2 lists all strict quasi coxeter elements is a consequence of the third column of table 5.2 , in the following way . in the cases of the root lattices of types @xmath34 and @xmath15 it is obvious that the quasi coxeter elements of types @xmath483 and @xmath84 are not strict quasi coxeter elements . in the cases of the p.n .
root lattices @xmath96 and @xmath16 , observe @xmath492 this shows that all permutations whose signed cycles contain at least two negative cycles are not strict quasi coxeter elements .
@xmath16 has no strict quasi coxeter elements because of @xmath493 .
therefore the only elements in the third column of table 5.2 which are strict quasi coxeter elements are those in the second column .
it rests to prove the third column of table 5.2 .
the root lattice of type @xmath494 : because of @xmath495 the quasi coxeter elements of @xmath102 are also quasi coxeter elements of @xmath42 .
let @xmath496 be a quasi coxeter element of @xmath42 which is not a quasi coxeter element of @xmath102 .
then it has a presentation @xmath443 such that its subroot lattice is the full root lattice @xmath497 and @xmath498 . by lemma [ t4.2 ] ( c)(ii ) and ( e)(ii )
then the graph @xmath22 is a tree and contains exactly one marked vertex .
then @xmath26 is a negative cycle , so a coxeter element of @xmath42 .
root lattices of types @xmath499 and @xmath500 : let @xmath501 be a quasi coxeter element of @xmath183 or @xmath16 , and let @xmath443 be a presentation whose subroot lattice is the full root lattice @xmath502 . by lemma [ t4.2 ] ( b)(ii ) and ( d)(ii ) then the graph @xmath22 for @xmath503 is a union of trees which have each exactly one marking and which is short .
thus @xmath26 is a product of negative cycles .
vice versa , any product @xmath26 of negative cycles has a presentation @xmath443 such that the graph @xmath504 is a union of trees which have each exactly one marking and which is short .
thus @xmath26 is a quasi coxeter element . the root lattice of type @xmath340 : obviously its quasi coxeter elements are the products @xmath505 with @xmath8 short and @xmath10 long and @xmath506 and the products @xmath507 with @xmath341 and @xmath342 short and @xmath508 .
the elements of the first type are the coxeter elements of @xmath15 , the elements of the second type can also be written as products @xmath509 with @xmath510 and @xmath511 long roots and @xmath512 .
they are the coxeter elements of the subroot system of long roots , which is of type @xmath84 .
the root lattice of type @xmath513 : see lemma [ t5.7 ] ( b ) .
the restriction there that in the presentation @xmath514 first the short roots come and then the long roots , is not serious .
one can obtain a presentation with this property from an arbitrary presentation using .
@xmath262 [ t5.7 ] let @xmath24 be the root lattice of type @xmath34 in theorem [ t2.3 ] ( d ) . obviously the short roots form a root system of type @xmath481 , which is called @xmath515 , and the long roots form a root system of type @xmath481 , which is called @xmath481 .
\(a ) let @xmath516 be a @xmath3-basis of @xmath517 such that first the short roots come and then the long roots .
then one of the following cases holds .
all four roots are short .
@xmath518 and @xmath343 are short and @xmath519 is long .
then an element @xmath520 exist such that @xmath521 generate the subroot system of type @xmath522 which is also generated by @xmath523 , and then @xmath524 with @xmath525 and @xmath526 .
@xmath341 and @xmath342 are short and @xmath343 and @xmath519 are long .
then @xmath527 and @xmath528 and @xmath529 .
\(b ) let @xmath520 be a quasi coxeter element , and let @xmath530 be a presentation of @xmath26 whose subroot lattice is the full lattice @xmath517 and such that first the short roots come and then the long roots .
then the cases in ( a ) hold , and @xmath26 is in each case as follows .
@xmath26 is in @xmath531 and is a coxeter element there , or it is in @xmath487 and is a quasi coxeter element of type @xmath486 there .
@xmath26 is in @xmath532 for some subroot lattice @xmath2 of type @xmath480 and is a coxeter element there .
@xmath26 is a coxeter element of type @xmath34 or a strict quasi coxeter element of type @xmath533 .
* proof : * ( a ) the following obvious statements will be used : the root lattice @xmath517 contains the subroot lattice @xmath534 of type @xmath480 with root system @xmath535 it has the same long roots as @xmath536 , but less short roots . for any short root @xmath510 a weyl group element @xmath26
exists such that @xmath537 .
if @xmath511 is a short root with @xmath538 then @xmath539 , and @xmath26 can be chosen such that @xmath540 .
the case that all four roots @xmath541 are long is impossible because they would only generate the subroot lattice @xmath542 .
the case that the three roots @xmath543 are long and @xmath519 is short , is also impossible , because by ( b ) a weyl group element exists such that @xmath544 , and then all four images @xmath545 are in @xmath534 .
thus either two roots are short and two roots are long , or three roots are short and one root is long , or all four roots are short .
consider the case that @xmath543 are short and @xmath519 is long .
then @xmath518 and @xmath343 generate a subroot system of rank 3 of @xmath546 . only the two types @xmath522 and @xmath547 are possible a priori . here
the type @xmath547 is not possible , because then by ( b ) an element @xmath520 exists such that @xmath548 for @xmath549 , and these roots and any long root @xmath550 are in @xmath534 . by theorem [ t3.3 ] ( b ) , any two subroot systems of type @xmath522 of the subroot system @xmath515 are conjugate by an element of @xmath551 .
this shows the first half of part ( ii ) .
obviously @xmath524 with @xmath525 and @xmath526 .
this gives part ( ii ) .
consider the case that @xmath341 and @xmath342 are short and @xmath343 and @xmath519 are long . if @xmath552 then by ( b ) @xmath541 are mapped by a suitable element @xmath520 into the subroot lattice @xmath534 .
therefore @xmath527 .
furthermore @xmath553 because else the four roots would generate a reducible subroot lattice .
an element @xmath520 exists such that @xmath554 if @xmath555 then either @xmath556 for some @xmath557 and some @xmath558 with @xmath559 , or @xmath560 for some @xmath561 and some @xmath562 with @xmath563 .
one sees easily with some case discussion that in both cases @xmath564 do not generate @xmath517 .
\(b ) of course , the cases in ( a ) hold . * the case ( i ) : * @xmath26 is in @xmath551 and is either a coxeter element there or a quasi coxeter element of type @xmath565 . in the first case ,
@xmath26 is conjugate to @xmath566 which is in @xmath531 and which is a coxeter element there . in the second case
, @xmath26 is conjugate to @xmath567 which is in @xmath487 and which is a quasi coxeter element of type @xmath486 there . * the case ( ii ) : * the element @xmath26 is conjugate to @xmath568 for some @xmath569 with @xmath525 and @xmath570 and @xmath571 .
this is in @xmath572 for a subroot lattice @xmath573 of type @xmath480 . in the case
@xmath574 the generalized dynkin diagram of the four roots on the right hand side is ( up to the distinction between dotted and normal edges ) the @xmath480 dynkin diagram , so then the element is a coxeter element in @xmath572 . in the case
@xmath575 , the right hand side is equal to @xmath576 where @xmath209 is determined by @xmath577 .
this is again a coxeter element in @xmath572 . *
the case ( iii ) : * using for @xmath341 and @xmath342 , one can suppose @xmath578 .
after conjugation , one can suppose @xmath579 using for @xmath580 and changing possibly some signs and conjugating possibly again , one can suppose @xmath581 for some @xmath582 . in the case
@xmath583 , the generalized dynkin diagram of the roots @xmath584 is ( up to the distinction between dotted and normal edges ) a dynkin diagram of type @xmath34 .
thus @xmath26 is a coxeter element in @xmath482 . in the case
@xmath585 , the element @xmath26 is conjugate to the product of the two involutions @xmath586 and @xmath587 with admissible diagram of type @xmath533 .
thus it is a quasi coxeter element in @xmath482 of type @xmath533 . in the case
@xmath588 , the element @xmath26 is @xmath589 the generalized dynkin diagram of the roots on the right hand side is ( up to the distinction between dotted and normal edges ) a dynkin diagram of type @xmath34 .
thus @xmath26 is a coxeter element in @xmath482 . in the case
@xmath590 , the element @xmath26 is @xmath591 this is conjugate to the element @xmath592 the generalized dynkin diagram of the roots on the right hand side is ( up to the distinction between dotted and normal edges ) a dynkin diagram of type @xmath34 . thus @xmath26 is a coxeter element in @xmath482 .
@xmath262 [ t5.8 ] ( i ) in the tables 711 in @xcite all conjugacy classes of elements of the weyl groups of the root lattices of types @xmath315 and @xmath38 are listed in the following form . for any conjugacy class one element and one presentation of it as a strict quasi coxeter element is chosen .
the tables show the isomorphism class of the pair of full lattice and subroot lattice and the type of the strict quasi coxeter element .
\(ii ) theorem [ t5.10 ] below gives more information for the root lattices of types @xmath315 and @xmath38 .
for @xmath34 it lists for any conjugacy class all isomorphism classes of pairs of full lattice and subroot lattice , which turns up as subroot lattice of a presentation as a quasi coxeter element , and the type of the quasi coxeter element .
this gives all types of reduced presentations for any element .
for @xmath593 and @xmath38 one can extract the same information from theorem [ t5.10 ] and the tables 3.6 , 3.2 , 3.3 and 3.4 .
\(iii ) for the tables in theorem [ t5.10 ] , the notations in table 5.2 have to be refined : there are three conjugacy classes of quasi coxeter elements in @xmath34 which have also presentations as coxeter elements in subroot lattices of type @xmath480 and @xmath485 respectively as a quasi coxeter element of type @xmath486 in the subroot lattice of type @xmath481 .
the presentations of these elements as quasi coxeter elements in @xmath34 are now called @xmath594 and @xmath595 .
analogously , the presentations as quasi coxeter elements in @xmath488 of those elements which have also presentations as coxeter elements in @xmath84 ( the subroot lattice of long roots ) are denoted by @xmath596 . for @xmath597 , the presentations as quasi coxeter elements in @xmath598 of those elements in @xmath598 which are products of negative cycles of lengths
@xmath599 with @xmath600 are denoted by @xmath601 . the case @xmath602 is also denoted by @xmath222 .
this will be used in the table 5.4 for @xmath34 .
similarly , @xmath603 is used there .
\(iv ) the complete control in theorem [ t5.10 ] on the reduced presentations of the weyl group elements of the irreducible root lattices allows to determine the number @xmath604 in definition [ t5.9 ] .
[ t5.9 ] let @xmath24 be a p.n .
root lattice , and let @xmath26 be a weyl group element . recall @xmath605 from theorem [ t3.8 ] .
define the number @xmath606 this number will be important in section [ s6 ] . because of @xmath607 ( theorem [ t3.8 ] ( b ) ) , @xmath608 equality holds if and only if a reduced presentation with subroot lattice @xmath609 exists .
this is the unique primitive subroot lattice @xmath2 with @xmath610
. often equality holds , often not .
[ t5.10 ] let @xmath24 be one of the irreducible p.n .
root lattices in theorem [ t2.3 ] ( d ) .
\(a ) in the cases @xmath89 , @xmath183 and @xmath16 , @xmath611 \(b ) consider in the cases @xmath42 and @xmath102 a weyl group element @xmath26 which is a product of @xmath239 positive cycles and @xmath612 negative cycles with disjoint supports whose union is @xmath461 ( remark [ t5.4 ] ( vii ) ) .
( in the case of @xmath102 , @xmath612 is even . ) then @xmath613 and then any reduced presentation with subroot lattice @xmath2 with minimal @xmath605 satisfies @xmath614},\\ k_4(l , w ) & = & k_2(l , l_1 ) = n - l(w)+\left[\frac{s+1}{2}\right ] .\label{5.12}\end{aligned}\ ] ] \(c ) in the cases @xmath593 and @xmath38 , for the big majority of the weyl group elements there is only one type of reduced presentations .
that means , the pairs @xmath17 are isomorphic where @xmath2 runs through the subroot lattices of all reduced presentations .
the table 5.3 lists for the ( conjugacy classes of the ) exceptions the different ways to write them as quasi coxeter elements of subroot lattices @xmath2 , and it lists the numbers @xmath604 . for the other elements , @xmath615 for the unique isomorphism class @xmath17 .
all these other elements can be found by replacing in the tables 3.6 , 3.2 , 3.3 and 3.4 @xmath2 by the possible quasi coxeter elements with subroot lattice of type @xmath2 .
see table 5.1 for the possibilities .
( e.g. @xmath616 has to be replaced by the two possibilities @xmath616 and @xmath617 .
) @xmath618 '' & 3\\ e_8 & d_4(a_1)+3a_1 \sim 2a_3+a_1 & 2\\ e_8 & d_5(a_1)+2a_1 \sim d_4+a_3 & 2\\ e_8 & d_6(a_1)+a_1 \sim [ a_7 ] '' & 2\\ e_8 & d_4(a_1)+4a_1 \sim 2a_3 + 2a_1 & 2\\ e_8 & d_4+d_4(a_1 ) \sim d_5(a_1)+a_3 & 1\\ e_8 & d_5+a_3 \sim a_7+a_1 \sim d_6(a_1)+2a_1 & 1\\ e_8 & d_6(a_2)+2a_1 \sim 2d_4 & 2\\ e_8 & e_6(a_1)+a_2 \sim a_8 & 1\\ e_8 & e_7(a_1)+a_1 \sim d_8 & 1\\ e_8 & e_7(a_3)+a_1 \sim d_8(a_2 ) & 1 \end{array}\end{aligned}\ ] ] \(d ) in the case of @xmath34 , the following table 5.4 lists for ( the conjugacy classes of ) all weyl group elements all ways to write them as quasi coxeter elements of subroot lattices .
see remark [ t5.8 ] ( iii ) for the notations .
it also lists the numbers @xmath604 .
@xmath619 @xmath620 * proof : * ( a ) in the cases @xmath621 and @xmath102 , any positive cycle in the weyl group can be written as a product @xmath622 where @xmath623 are roots of the type @xmath260 whose graph @xmath624 is a tree .
the subroot lattice @xmath625 is a primitive sublattice .
any element of @xmath88 is a product of positive cycles with disjoint supports whose union is @xmath461 .
because the supports are disjoint , the sum of the subroot lattices of the presentations above of the positive cycles is also a primitive sublattice .
therefore there holds . in the cases @xmath183 and @xmath16 , any negative cycle can be written as a product @xmath626 such that @xmath622 is a positive cycle with graph a tree , and such that @xmath627 is a short root which gives a marking of one vertex of the tree .
the subroot lattice @xmath628 is the primitive sublattice , which is generated by all the short roots which correspond to the vertices of the tree .
any element of @xmath629 is a product of positive cycles and/or negative cycles with disjoint supports whose union is @xmath461 .
because the supports are disjoint , the sum of the subroot lattices of the presentations above of the positive and/or negative cycles is also a primitive sublattice .
therefore there holds .
\(b ) in the cases @xmath42 and @xmath102 , any pair of negative cycles can be written as a product @xmath630 such that @xmath631 and @xmath632 are positive cycles whose graphs are disjoint trees and such that @xmath633 and @xmath634 with @xmath208 a vertex of one tree and @xmath209 a vertex of the other tree . the subroot lattice @xmath635 is of type @xmath636 respectively @xmath637 . in the case
@xmath42 , any single negative cycle can be written as a product @xmath638 such that @xmath631 is a positive cycle with graph a tree , and such that @xmath639 is a long root of the type @xmath640 which gives a marking of one vertex of the tree .
the subroot lattice @xmath641 is of type @xmath642 .
let @xmath26 be a weyl group element which is a product of @xmath239 positive cycles and @xmath612 negative cycles with disjoint supports whose union is @xmath461 .
one presents the positive cycles as above ( in the proof of ( a ) ) and as many pairs of negative cycles as above . at most one ( none in the case @xmath102 ) single negative cycle is left and is also presented as above .
let @xmath2 be the subroot lattice of the presentation .
table 3.1 shows .
one sees easily that no reduced presentation with smaller @xmath21 exists .
\(c ) the tables 7 , 9 , 10 and 11 in @xcite list all conjugacy classes of elements of the weyl groups of root lattices of the types @xmath593 and @xmath38 . they give in each case one type of presentation as a strict quasi coxeter element .
it is easy to find all presentations as quasi coxeter elements which are not in the list .
one has to find out which elements in the list are given also by these presentations . in most cases
it is sufficient to compare the characteristic polynomials .
a table of characteristic polynomials is table 3 in @xcite .
the only cases where this is not sufficient arise for the @xmath38 root lattice and there for the presentations as quasi coxeter elements of types @xmath643 and @xmath644 . in the first case
the presentations as strict quasi coxeter elements of types @xmath645'$ ] and @xmath645''$ ] have the same characteristic polynomial , in the second case the presentations of types @xmath646'$ ] and @xmath646''$ ] .
because of @xmath647'')\neq\textup{index}([2a_3]')=1,\\ \textup{index}(d_6+a_1)=2=\textup{index}([a_7]'')\neq\textup{index}([a_7]')=1,\end{aligned}\ ] ] lemma [ t5.2 ] ( b ) tells that @xmath643 gives the same conjugacy class as @xmath645''$ ] and that @xmath644 gives the same conjugacy class as @xmath646''$ ] .
\(d ) table 8 in @xcite lists 9 , 8 , 5 , 2 and 1 conjugacy classes of elements of the weyl group of type @xmath34 of lengths 4 , 3 , 2 , 1 respectively 0 .
on the other hand there are 19 , 12 , 6 , 2 and 1 types of presentations of elements as quasi coxeter elements of lengths 4 , 3 , 2 , 1 respectively 0 : @xmath648 for those types of presentations as quasi coxeter elements in the table above which are not in the table 8 in @xcite , one has to find out which conjugacy classes they give . in many cases this
is determined by the characteristic polynomials .
the cases where the characteristic polynomials is not sufficient , can be drawn from lemma 26 in @xcite .
it lists the presentations as strict quasi coxeter elements which give different conjugacy classes , but with the same characteristic polynomials .
of the 8 pairs in lemma 26 in @xcite , only those 4 are relevant here , for which presentations as quasi coxeter elements exist which are not in table 8 in @xcite and which have the same characteristic polynomials . these 4 pairs and their characteristic polynomials are as follows : @xmath649 the equality @xmath650 tells that in table 5.4 @xmath651 the equality @xmath652 tells that in table 5.4 @xmath653 the equalities @xmath654 tell that in table 5.4 @xmath655 the equivalence @xmath656 in table 5.4 holds by definition of @xmath657 .
all other equivalences in table 5.4 follow from comparison of characteristic polynomials.@xmath262 [ t5.11 ] ( i ) from the theorems [ t5.10 ] , [ t5.6 ] and [ t3.3 ] ( respectively the first columns of the tables 3.13.6 ) , one can recover the classification of conjugacy classes of the weyl group elements of the root lattices of types @xmath315 and @xmath38 which is given in the tables 711 in @xcite .
\(ii ) the proof above of theorem [ t5.10 ] had used these tables , but not in a very crucial way .
those few cases where different conjugacy classes have the same characteristic polynomials , can be dealt with by hand .
in fact , informations on them are given in the lemmas 26 and 27 in @xcite .
but theorem [ t5.6 ] on the ( strict ) quasi coxeter elements depends in a crucial way on the results in @xcite .
\(iii ) the characteristic polynomials of the strict quasi coxeter elements in all irreducible root lattices are given in table 3 in @xcite .
\(iv ) in ( * ? ? ?
* ( 2.3.4 ) ) a table similar to table 5.3 for @xmath156 and @xmath38 is given .
but one of the cases for @xmath156 and four of the cases for @xmath38 are missing there . the case for @xmath156 which is missing in ( * ? ? ? * ( 2.3.4 ) ) , is also missing in ( * ? ? ? * ( 3.2.9 ) ) .
[ t5.12 ] there is a strange correspondence . define for any irreducible root lattice @xmath24 the two numbers @xmath658 then @xmath659 as the following table shows .
@xmath660 & \left[\frac{n}{2}\right]-1 & 1 & 4 & 2 & 4 & 8 \\ k_7(l ) & 0 & n-1 & \left[\frac{n}{2}\right ] & \left[\frac{n}{2}\right]-1 & 2 & 4 & 2 & 4 & 8 \end{array}\end{aligned}\ ] ] here @xmath661 is the number of partitions of @xmath212 .
[ t6.1 ] let @xmath24 be a p.n . root lattice , and let @xmath26 be a weyl group element . define the number @xmath662 recall the definition of the number @xmath604 in the same situation .
let @xmath663 be a reduced presentation of an element @xmath26 with subroot lattice @xmath2 such that @xmath605 is minimal , i.e. @xmath664 .
let @xmath665 be roots such that @xmath666 .
then obviously @xmath667 is a presentation with root lattice the full root lattice @xmath1 .
therefore @xmath668 [ t6.2 ] let @xmath24 be a p.n . root lattice , and let @xmath26 be a weyl group element
. then @xmath669 the proof consists in a reduction to the special case in the following lemma and in the proof of the following lemma . the proof of the lemma is given first .
[ t6.3 ] let @xmath24 be a p.n .
root lattice of some rank @xmath212 , and let @xmath26 be a weyl group element of length @xmath475 . then @xmath670 * proof of lemma [ t6.3 ] : * if @xmath24 is reducible with orthogonal summands @xmath671 , then @xmath26 decomposes accordingly into a product of commuting elements @xmath672 , and the numbers @xmath604 and @xmath673 are additive , @xmath674 therefore it is sufficient to prove the lemma and also theorem [ t6.2 ] for the irreducible p.n .
root lattices .
let @xmath24 be an irreducible p.n .
root lattice of rank @xmath212 , and let @xmath26 be a weyl group element with @xmath675 . then @xmath676 . if @xmath677 then by also @xmath678 .
thus it is sufficient to prove @xmath679 .
the cases @xmath680 : suppose @xmath678 , and let @xmath681 be a presentation of @xmath26 whose subroot lattice is the full lattice . by theorem [ t4.1 ] , the set @xmath682 contains a @xmath3-basis of the full lattice @xmath1 .
using , we can suppose that @xmath347 is a @xmath3-basis of @xmath1 .
let @xmath683 be an arbitrary reduced presentation of @xmath26 . then @xmath684 the subroot lattice of the presentation on the left hand side is the full lattice , so it has index one .
by lemma [ t5.2 ] ( b ) , the index of the subroot lattice of the presentation on the right hand side is the same , so it is also one .
thus @xmath685 this shows here @xmath686 the cases @xmath499 and @xmath500 : because of theorem [ t5.10 ] ( a ) , @xmath687 holds anyway .
the cases @xmath494 : @xmath677 holds if and only if a reduced presentation with subroot lattice of type @xmath688 or of type @xmath689 for some @xmath690 exists .
this follows from table 3.1 . in the case
@xmath688 , @xmath26 is a positive cycle of length @xmath212 . in the case
@xmath689 , @xmath26 is a product of a positive cycle of length @xmath29 and of one or two negative cycles such that the sum of their lengths is @xmath476 .
it rests to show that @xmath26 is such an element if @xmath678 .
thus suppose @xmath678 .
let @xmath681 be a presentation of @xmath26 whose subroot lattice is the full lattice @xmath1 . by theorem [ t4.1 ] , the set @xmath682 contains a @xmath3-basis of the full lattice @xmath1 .
using , we can suppose that @xmath691 is a @xmath3-basis of @xmath1 .
thus @xmath692 is a quasi coxeter element , so either one negative cycle or the product of two negative cycles .
if @xmath341 is a long root , multiplying @xmath693 from the left with @xmath694 will turn one of the ( one or two ) negative cycles into a positive cycle .
if @xmath341 is a short root , so @xmath695 , then the type of @xmath696 depends on the position of the vertices @xmath208 and @xmath209 in the supports of the ( one or two ) negative cycles .
if @xmath208 and @xmath209 are in the support of the same negative cycle , then it splits into two cycles , one positive and one negative .
if @xmath208 and @xmath209 are in the supports of different negative cycles , then @xmath696 is a positive cycle of length @xmath212 . in any case
, @xmath26 is of one of the types which satisfy @xmath677 .
the case @xmath340 : by table 3.6 , all subroot lattices of rank 1 are primitive sublattices .
therefore @xmath697 holds anyway . the case @xmath513 : by table 5.4 , the only elements @xmath26 with @xmath698 and @xmath699 are those of type @xmath700 , and the elements of this type satisfy @xmath701 .
it rests to show for them @xmath702 .
suppose that such an element @xmath26 satisfies @xmath678 , and let @xmath703 be a presentation of @xmath26 whose subroot lattice is the full lattice . by theorem [ t4.1 ] , the set @xmath704 contains a @xmath3-basis of the full lattice @xmath1 .
using , we can suppose that @xmath541 is a @xmath3-basis of @xmath1 .
we may suppose @xmath705 .
then @xmath706 because of the right hand side , this is a quasi coxeter element in @xmath482 .
first case , @xmath707 is a long root @xmath708 : then @xmath577 is impossible because else the four roots on the left hand side were linearly dependent .
@xmath709 is impossible because else the element on the left hand side were a coxeter element in @xmath487 , and this is not a quasi coxeter element in @xmath482 .
also @xmath710 is impossible because else the left hand side were an element of type @xmath711 , and this is not a quasi coxeter element in @xmath482 , or the four roots on the left hand side were linearly dependent .
the first case is impossible .
second case , @xmath707 is a short root : by conjugation and renumbering of the @xmath712 we can suppose @xmath713 for some @xmath208 . then @xmath525 is impossible because else the four roots on the left hand side were linearly dependent .
@xmath714 is impossible because else the left hand side were an element of type @xmath715 , and this is not a quasi coxeter element in @xmath482 .
the second case is impossible .
thus @xmath716 , so @xmath702 .
this finishes the proof of the case @xmath34 and the whole proof of lemma [ t6.3].@xmath262 let @xmath24 be an irreducible p.n .
root lattice of some rank @xmath212 . at the beginning of the proof of lemma [ t6.3 ]
it was shown that it is sufficient to prove theorem [ t6.2 ] in this case .
let @xmath26 be a weyl group element , and let @xmath717 be a presentation with subroot lattice the full lattice @xmath1 and with @xmath718 minimal with this property . by theorem [ t4.1
] , the set @xmath719 contains a @xmath3-basis of the full lattice @xmath1 . using
, we can suppose that @xmath347 is a @xmath3-basis of @xmath1 .
the element @xmath720 has length @xmath212 .
thus the element @xmath721 has length @xmath722 . and
it satisfies @xmath723 .
lemma [ t6.3 ] applies .
therefore a reduced presentation @xmath724 of @xmath693 and a root @xmath725 exist such that @xmath726 .
let @xmath2 be the subroot lattice of the presentation @xmath727 of @xmath26 .
as @xmath29 is minimal , @xmath728 . because of @xmath729 , @xmath730
the presentation of @xmath26 in shows @xmath731 .
if @xmath732 then by adding two times @xmath733 to a shortest presentation of @xmath26 with subroot lattice @xmath2 , one obtains also @xmath734 , which contradicts the minimality of @xmath29 .
thus @xmath735 .
induction on @xmath29 gives @xmath736 .
now @xmath737 together with this gives .@xmath262
the number @xmath673 in definition [ t6.1 ] and theorem [ t6.2 ] controls existence of quasi coxeter elements in _ extended affine root systems_. these had been defined by k. saito in ( * ? ? ?
* ( 1.2 ) and ( 1.3 ) ) . in @xcite
the equivalence with an alternative definition in @xcite was shown .
the inequalities in lemma [ t7.4 ] below give constraints on a quasi coxeter element @xmath26 in an extended affine root system in terms of conditions for a nonreduced presentation of the induced element @xmath48 in the weyl group of the associated p.n .
root lattice @xmath44 .
[ t7.1 ] an _ extended affine root lattice _ is a triple @xmath24 where @xmath1 is a lattice , @xmath738 is a symmetric positive semidefinite bilinear form , and @xmath739 is a subset such that the following properties hold . here @xmath740 and @xmath60 are defined as in and .
@xmath741 the elements of @xmath6 are the _ roots _ , and @xmath6 is an _ extended affine root system_. @xmath742 is the _ weyl group of the extended affine root lattice_.
\(ii ) let @xmath24 be an extended affine root lattice . because @xmath54 is positive semidefinite , the radical of @xmath744 is @xmath745 define the radicals @xmath746 and @xmath747 .
the quotient @xmath44 with the induced bilinear form @xmath748 and the induced set of roots @xmath749 is obviously a p.n .
root lattice .
it is called the _ quotient p.n .
root lattice_. any element @xmath750 induces an element @xmath751 . if @xmath45 , then @xmath46 . if @xmath752 induces @xmath753\in l/\operatorname{rad}(l)$ ] , then @xmath754 .
\(iii ) the reducedness property is not required here .
even if @xmath24 satisfies it , it does not necessarily hold for the quotient p.n .
root lattice . that is the reason why in this paper _ p.n .
_ root lattices and not only root lattices are considered .
\(a ) for any element @xmath26 of its weyl group , a _ presentation _
@xmath28 , the _ length of a presentation _ , the _ subroot lattice of a presentation _ , and the _ length _ @xmath31 of the element are defined as in definition [ t5.1 ] ( a ) .
the following simple lemma connects the existence of quasi coxeter elements with the numbers @xmath755 from section [ s6 ] .
theorem [ t6.2 ] says @xmath756 , and theorem [ t5.10 ] allows to calculate this number .
[ t7.4 ] let @xmath24 be an extended affine root lattice of rank @xmath212 with a radical @xmath757 of rank @xmath758 .
let @xmath149 be a quasi coxeter element , and let @xmath48 be the induced element in the weyl group @xmath759 of the quotient p.n .
root lattice .
then @xmath760 * proof : * is a trivial consequence of , i.e. @xmath761 . a presentation of length @xmath212 of @xmath26 whose subroot lattice is the full lattice @xmath1 induces a presentation of length @xmath212 of @xmath48 whose subroot lattice is the full lattice @xmath44 .
this shows .@xmath262 but if @xmath44 is a homogeneous root lattice , then there is a sublattice @xmath121 such that @xmath762 is isomorphic to the quotient root lattice and @xmath24 is equal to @xmath763 .
thus up to isomorphism , @xmath24 is determined by the isomorphism class of the ( homogeneous ) quotient root lattice and by the rank @xmath239 of the radical .
\(ii ) let @xmath24 be an extended affine root lattice of rank @xmath212 with radical @xmath757 of rank @xmath239 .
let @xmath149 be a quasi coxeter element such that @xmath48 has maximal length @xmath764 .
then lemma [ t7.4 ] and theorem [ t6.2 ] give @xmath765 sak b. allison , s. azam , s. berman , y. gao , a. pianzola : extended affine lie algebras and their root systems .
* 603 * ( 1997 ) , 1122 .
s. azam : extended affine root systems .
journal of lie theory * 12 * ( 2002 ) , 515527 .
a. borel , j. de siebenthal : les sous - groupes ferms connexes de rang maximum des groups de lie clos .
* 25.1 * ( 1972 ) 159 .
n. bourbaki : groupes et algbres de lie , chapitres 4 , 5 et 6 .
hermann , paris , 1968 .
carter : conjugacy classes in the weyl group . comp . math .
* 25 * ( 1972 ) , 159 .
dynkin : semisimple subalgebras of semisimple lie algebras .
translations of the ams ( 2 ) * 6 * ( 1957 ) , 111244 .
humphreys : reflection groups and coxeter groups .
cambridge studies in advanced mathematics * 29*. cambridge university press , 1990 .
p. kluitmann , geometrische basen des milnorgitters einer einfach elliptischen singularitt .
diplomarbeit , 116 seiten , bonn , 1983 .
k. saito : extended affine root systems i ( coxeter transformations ) .
rims kyoto * 21 * , ( 1985 ) , 75179 .
e. voigt : ausgezeichnete basen von milnorgittern einfacher singularitten .
doktorarbeit , 150 seiten , bonner mathematische schriften * 160 * , 1985 . | this paper is a sequel to work of dynkin on subroot lattices of root lattices and to work of carter on presentations of weyl group elements as products of reflections .
the quotients @xmath0 are calculated for all irreducible root lattices @xmath1 and all subroot lattices @xmath2 .
the reduced ( i.e. those with minimal number of reflections ) presentations of weyl group elements as products of arbitrary reflections are classified .
also nonreduced presentations are studied .
quasi coxeter elements and strict quasi coxeter elements are defined and classified .
an application to extended affine root lattices is given .
a side result is that any set of roots which generates the root lattice contains a @xmath3-basis of the root lattice . [ section ] [ lemma]definition / lemma [ lemma]lemma / definition [ lemma]theorem [ lemma]proposition [ lemma]corollary [ lemma]conjecture [ lemma]conjectures [ lemma]definition [ lemma ] [ lemma]remark [ lemma]remarks [ lemma]example [ lemma]examples [ lemma]notations |
towards the end of his talk at _ permutation patterns 2015 _ , vince vatter pointed out that four of the permutation classes avoiding two patterns of length four remaining to be enumerated are subclasses of _ generalised grid classes_. we illustrate the value of this observation by using it to determine the generating function of the class @xmath3 and to explore the structure of permutations in @xmath4 . breaking somewhat with tradition , we assume familiarity with the standard terminology and notation used for permutation classes .
see @xcite for a very brief introduction , or the survey article @xcite for a much more thorough exposition .
the concept of a generalised grid class was introduced by vatter in ( * ? ? ?
* section 2 ) .
a generalised grid class is defined by a matrix @xmath5 , whose entries are permutation classes .
the class @xmath6 consists of those permutations @xmath7 whose plot can be partitioned into rectangular cells , one corresponding to each entry of @xmath8 , such that the subpermutation of @xmath7 consisting of any points in the cell corresponding to @xmath9 is a member of the permutation class @xmath9 .
such a partition is known as a _ gridding _ of @xmath7 .
see figure [ figgridclassex ] for an illustration . with a slight abuse of notation , we tend to identify a grid class with its defining matrix
. @xmath10 \plotperm[blue!50!black]{16}{1,9,12,11,13,7,10,14,8,2,3,15,16,6,4,5 } \draw[thick ] ( 0.5,12.5)--(16.5,12.5 ) ; \draw[thick ] ( 3.5,0.5)--(3.5,16.5 ) ; \end{tikzpicture } \vspace{-6pt}\ ] ] we prove ( proposition [ propaing ] ) that @xmath3 is a subclass of the generalised grid class @xmath11 each permutation in @xmath4 consisting of the skew sum of a sequence of plane trees , with an increasing sequence of points above ( _ top points _ ) and an increasing sequence of points to its left ( _ left points _ ) .
see figures [ figgridclassex ] and [ figdexample ] for illustrations . by analysing the structure of certain `` canonical '' griddings ,
we establish ( theorem [ thmagf ] ) that the generating function enumerating @xmath4 is @xmath12 we also investigate the properties of a typical large permutation in @xmath4 and determine ( corollary [ cord ] ) that , asymptotically , one fifth of the points in a permutation in @xmath4 are top points , and that more than half of the permutations in @xmath4 have a gridding with no left points and hence contain no occurrences of @xmath2 . in the rest of this introductory section , we present various definitions , notation and results that we use later . in section [ secth ]
, we investigate the structure of @xmath13 , a subclass of @xmath4 first studied by bloom & burstein @xcite .
then , in section [ sectd ] , we build on our analysis of this subclass to explore the structure of permutations in @xmath4 and prove our main results .
we conclude with some questions . to establish generating functions , we make use of the _ symbolic method _ as presented in flajolet & sedgewick ( * ? ?
* chapters i and iii ) .
we use @xmath14 to denote the atomic class consisting of a single point in a permutation , @xmath15 to represent a possibly empty sequence of elements of @xmath16 and @xmath17 to represent a non - empty sequence of elements of @xmath16 .
we also use @xmath18}$ ] to denote the class of non - empty plane trees whose vertices are drawn from @xmath16 , defined by the structural equation @xmath19 } \;=\ ; \ccc \times \seqbig{{\text{\textsc{tree}}[\ccc]}}.\ ] ] we make repeated use of the fact that any permutation @xmath20 consists of the skew sum of a sequence of plane trees , and thus @xmath21}}$ ] .
this is made evident by constructing the _ hasse graph _ of @xmath22 , the graph corresponding to the hasse diagram of the poset on the points of @xmath22 in which @xmath23 if @xmath24 is to the lower left of @xmath25 .
see the lower cell in figure [ fighexample ] and the lower right cell in figure [ figdexample ] for illustrations .
a previous use of hasse graphs in enumerating permutation classes can be found in @xcite .
we make use of the following terminology .
the uppermost ( and leftmost ) tree in such a skew sum is the _
top tree_. in each tree , the left to right maxima ( the vertices in the leftmost maximal path from the root ) constitute the _ trunk _ , the uppermost point of which is its _
tip_. the _ uppermost branching point _ in the trunk of a tree is the uppermost point in the trunk of degree greater than 2 . to conclude this introductory section , we state three general propositions of analytic combinatorics that we require .
recall that @xmath26f(z)$ ] denotes the coefficient of @xmath27 in @xmath28 .
[ propasympt ] the coefficient of @xmath27 in @xmath29 admits for large @xmath30 the following asymptotic expansion : @xmath31\lambda\+(1-z/\rho)^\alpha \;\sim\ ; \frac{\lambda}{\gamma(-\alpha ) } \+ \rho^n \+ n^{-\alpha-1 } \+ \bigg(\!1 \,+\ , \sum_{k=1}^\infty \frac{e_k}{n^k}\bigg),\ ] ] where @xmath32 _ k , = e^t(1+vt)^-1 - 1/v .
@xmath33 proposition [ propasympt ] enables us to determine the complete asymptotics of @xmath26f(z)$ ] by repeated application to successive terms of the ( puiseux ) expansion for @xmath28 around its dominant singularity .
proposition [ propmoments ] can be used to determine the mean and standard deviation of some parameter , and proposition [ propgaussian ] allows us to establish that a parameter is asymptotically normally distributed .
[ propmoments ] suppose that @xmath34 is the bivariate generating function for some combinatorial class , in which @xmath35 marks size and @xmath36 marks the value of a parameter @xmath37 .
then the mean and variance of @xmath37 for elements of size @xmath30 are given by @xmath38 \;=\ ; \frac{\big[z^n\big]\+ \partial_x a(z , x)\big|_{x=1}}{\big[z^n\big]\+ a(z,1 ) } \qquad \text{and } \qquad \mathbb{v}_n[\xi ] \;=\ ; \frac{\big[z^n\big]\+ \partial^2_x a(z , x)\big|_{x=1}}{\big[z^n\big]\+ a(z,1 ) } \:+\ : \mathbb{e}_n(\xi ) \:-\ : { \mathbb{e}_n(\xi)}^2\ ] ] respectively . [ propgaussian ]
let @xmath39 be a bivariate function , analytic at @xmath40 and with non - negative taylor coefficients , and let @xmath41 be the sequence of random variables with probability generating functions @xmath42f(z ,
x)}{\big[z^n\big]f(z,1 ) } .\ ] ] if the following conditions hold : there exist positive @xmath43 and @xmath44 and functions @xmath45 , @xmath46 and @xmath47 , analytic in @xmath48 , such that @xmath49 for some @xmath50 .
the equation @xmath51 has a unique root , @xmath52 , with @xmath53 , @xmath52 is a simple root , @xmath54 , @xmath55 and @xmath56
. then , as long as its asymptotic variance is non - zero , @xmath41 converges in law to a gaussian distribution .
let us begin by considering a subclass of @xmath4 , the permutation class @xmath13 .
first , we present a characterisation of the class , which we then use to establish its generating function and determine some asymptotic properties .
this class was first investigated by bloom & burstein @xcite .
although they do nt explicitly make use of the notion of a generalised grid class , they prove a result equivalent to our first proposition .
[ proph ] @xmath13 is the generalised grid class @xmath57 .
@xmath58 \plotperm[blue!50!black]{4}{4,2,1,3 } \draw[gray ] ( 0.5,3.5)--(4.5,3.5 ) ; \node [ ] at ( 2.5,-0.5 ) { $ \mathbf{4213}$ } ; \end{tikzpicture } \qquad\quad\qquad \begin{tikzpicture}[scale=0.25 ] \plotperm[blue!50!black]{4}{2,4,1,3 } \draw[gray ] ( 0.5,3.5)--(4.5,3.5 ) ; \node [ ] at ( 2.5,-0.5 ) { $ \mathbf{2413}$ } ; \end{tikzpicture } \qquad\quad\qquad \begin{tikzpicture}[scale=0.25 ] \plotperm[blue!50!black]{4}{2,1,4,3 } \draw[gray ] ( 0.5,3.5)--(4.5,3.5 ) ; \node [ ] at ( 2.5,-0.5 ) { $ \mathbf{2143}$ } ; \end{tikzpicture } \vspace{-9pt}\ ] ] let @xmath59 .
suppose we have a gridding of a permutation @xmath60 , and that @xmath61 .
inspection of figure [ fighbasis ] shows that no more than one point of an occurrence of @xmath62 in @xmath63 can occur in the upper cell , or else it would contain a @xmath64 , and neither can three or more points of @xmath62 occur in the lower cell , or else it would contain a @xmath65 . thus @xmath66 is a subclass of @xmath67 . on the other hand ,
suppose @xmath7 is a permutation in @xmath67 of length @xmath30 , and let @xmath22 be the maximal increasing subsequence of @xmath7 of the form @xmath68 for some @xmath69 .
note that if @xmath70 , then , by the maximality of @xmath22 , the term @xmath71 occurs to the right of @xmath72 in @xmath7 .
we claim that @xmath7 can be gridded in @xmath66 by placing the points of @xmath22 in the upper cell , and any remaining points in the lower cell . clearly , @xmath22 avoids @xmath64 .
suppose there was an occurrence of @xmath65 in the lower cell .
inspection of figure [ fighbasis ] shows that , since @xmath7 avoids elements of @xmath46 , the occurrence of @xmath65 must occur to the left of all the points in the upper cell .
however , in this case , the leftmost two points of this @xmath65 together with @xmath72 and @xmath71 would form a @xmath1 , which is impossible because @xmath7 avoids @xmath1 .
thus @xmath67 is a subclass of @xmath66 .
proposition [ proph ] can also be deduced from atkinson ( * ? ? ?
* theorem 2.2 ) , which specifies the structure of basis elements of juxtapositions of finitely based permutation classes .
bloom & burstein also enumerate @xmath66 .
we present our own proof here so that we can make use of it later when considering class @xmath4 .
[ prophgf ] the generating function for @xmath73 is @xmath74 @xmath58 \path [ fill = blue!15 ] ( 6.8,0.5 ) rectangle ( 7.2,21.5 ) ; \path [ fill = blue!15 ] ( 10.8,0.5 ) rectangle ( 11.2,21.5 ) ; \path [ fill = blue!15 ] ( 11.8,0.5 ) rectangle ( 12.2,21.5 ) ; \path [ fill = blue!15 ] ( 15.8,0.5 ) rectangle ( 16.2,21.5 ) ; \path [ fill = blue!15 ] ( 17.8,0.5 ) rectangle ( 18.2,21.5 ) ; \path [ fill = blue!15 ] ( 20.8,0.5 ) rectangle ( 21.2,21.5 ) ; \draw[thick , dashed ] ( 13.5,0.5)--(13.5,21.5 ) ; \draw[thick , dashed ] ( 16.5,0.5)--(16.5,21.5 ) ; \draw[thick ] ( 20,2)--(17,1)--(19,3 ) ; \draw[thick ] ( 14,4)--(15,5 ) ; \draw[thick ] ( 1,6)--(2,9)--(3,10)--(4,14)--(5,15 ) ; \draw[thick ] ( 10,8)--(1,6)--(13,7 ) ; \draw[thick ] ( 9,11)--(3,10)--(6,12)--(8,13 ) ; \plotperm[blue!50!black]{21}{6,9,10,14,15,12,16,13,11,8,17,18,7,4,5,19,1,20,3,2,21 } \draw[thick ] ( 0.5,15.5)--(21.5,15.5 ) ; \end{tikzpicture } \vspace{-6pt}\ ] ] for each permutation @xmath7 in @xmath66 , we consider the gridding of @xmath7 in which the number of points in the upper cell is minimal ( see figure [ fighexample ] for an illustration ) .
we call this the _ canonical _ gridding of @xmath7 in @xmath66 , and refer to the points in the top cell as the _
top points_. clearly , there is a bijection between permutations in @xmath66 and their canonical griddings .
the minimality of the set of top points implies that they all occur to the right of the first occurrence of a @xmath64 in the lower cell . in other words
, the leftmost top point must be to the right of the ( tip of the ) trunk of the top tree and not adjacent to it .
we use the symbolic method to construct the class , marking top points with @xmath75 for clarity ( although we do nt actually need an extra catalytic variable ) .
the structural equations are as follows : @xmath76 { \ttt^{\textsf{t } } } & \;=\ ; { \text{\textsc{tree}}\big[\zzz \times \seq{t\+\zzz}\big ] } \label{eqtr } \\[5pt ] { \ttt_{\textsf{top } } } & \;=\ ; \ppp \:\;+\;\ : \seqplus{\zzz } \times ( \zzz \times \seqplus{{\ttt^{\textsf{t } } } } ) \times \seqbig{\zzz \times \seq{{\ttt^{\textsf{t } } } } } \label{eqttop } \\[5pt ] \hhh & \;=\ ; { \ttt_{\textsf{top}}}\:\times\ : \seq{{\ttt^{\textsf{t } } } } \label{eqh}\end{aligned}\ ] ] in , we use @xmath77 to denote a tree consisting only of a trunk ( a _ path _ tree ) . in , we use @xmath78 to denote a tree together with any top points that occur to the right of some of its vertices . in figure
[ fighexample ] , the dashed lines illustrate how the top points are associated with the trees . in , we use @xmath79 to denote the top tree , together with its top points .
this is either a path tree ( with no top points ) , or else is constructed ( tip downwards ) from at least one point on the trunk above the uppermost branching point , the uppermost branching point on the trunk , rooting at least one subtree , which may have top points to the right of some of its vertices , and any points on the trunk below the uppermost branching point , each possibly rooting a subtree that may have top points to the right of some of its vertices . finally , in
, the class @xmath66 is constructed from the top tree , @xmath79 , and zero or more @xmath78 trees .
translation into functional equations yields : @xmath80 { t_{\textsf{top}}}(z , t ) & \;=\ ; \seqplusfrac{z } \:\;+\;\ : \seqplusfrac{z } \times \seqplusfrac[z\+]{{t^{\textsf{t}}}(z , t ) } \times \seqfrac{\seqfrac[z]{{t^{\textsf{t}}}(z , t ) } } , \\[6pt ] h(z , t ) & \;=\ ; \seqfrac[{t_{\textsf{top}}}(z , t)]{{t^{\textsf{t}}}(z , t ) } .\end{aligned}\ ] ] solving , expanding , simplifying and rearranging gives @xmath81 the univariate generating function then results from setting @xmath82 . from the generating function
, we can determine the following asymptotic behaviour .
[ corh ] asymptotically , @xmath83 and the number of top points in the canonical gridding of a permutation in @xmath84 is normally distributed with mean @xmath85 and standard deviation @xmath86 .
these results follow in the standard way from singularity analysis and the extraction of moments from the generating function .
the asymptotics of @xmath87 results from the application of proposition [ propasympt ] to the first nonconstant term of the puiseux expansion for @xmath88 around its dominant singularity @xmath89 : @xmath90 the asymptotic mean and standard deviation for the number of top points follow by the application of propositions [ propasympt ] and [ propmoments ] .
it is also readily verified that @xmath91 is amenable to proposition [ propgaussian ] . indeed , in the notation of that proposition , we have @xmath92 b(z , t ) & \;=\ ; \frac{-1}{2\+z\+(t+1-t\+z ) } , \\[5pt ] c(z , t ) & \;=\ ; 1 - ( 2\+t+4)\+z + t\+(t+4)\+z^2 , \end{aligned}\ ] ] and the exponent @xmath93 .
the relevant conditions are then satisfied for any @xmath94 and @xmath95 , with @xmath96 , as expected , being the unique root of @xmath51 whose magnitude is at most @xmath43 .
the number of top points thus exhibits a concentrated gaussian limit distribution .
the first twelve terms of the sequence @xmath87 are 1 , 2 , 6 , 21 , 79 , 311 , 1265 , 5275 , 22431 , 96900 , 424068 , 1876143 .
more values can be found at http://oeis.org/a033321[a033321 ] in oeis @xcite .
let us now turn to a consideration of the structure and enumeration of class @xmath4 .
in the first two propositions , we present a characterisation of the class , which we then use to enumerate the class in our main theorem .
we conclude by determining some asymptotic properties .
@xmath97 \bigcup \\[6pt ] \ggg \;=\ ; \ , { { \scalebox{0.9}{$ \begin{array}{|c|c| } \hline & \!\!\av(\mathbf{21})\!\ !
\\ \hline \!\!\av(\mathbf{21})\!\ ! & \!\!\av(\mathbf{213})\!\ ! \\
\hline \end{array } $ } } } \phantom { \;=\ ; \ggg } \\[12pt ] \bigcup \\[4pt ] \ddd \;=\ ; { \av(\mathbf{4213},\mathbf{2143})}\phantom { \;=\ ; \ddd } \\[4pt ] \bigcup \\[4pt ] \hspace{-.35 in } \phantom{\ , { { \scalebox{0.9}{$ \begin{array}{|c| } \hline \!\!\av(\mathbf{21})\!\ ! \\ \hline \!\!\av(\mathbf{213})\!\ ! \\ \hline \end{array } $ } } } \ , \;=\ ; } \hhh \;=\ ; { \av(\mathbf{4213},\mathbf{2413},\mathbf{2143})}\;=\ ; \ , { { \scalebox{0.9}{$ \begin{array}{|c| } \hline \!\!\av(\mathbf{21})\!\ ! \\
\hline \!\!\av(\mathbf{213})\!\ ! \\
\hline \end{array } $ } } } \phantom { \;=\ ; \hhh } \hspace{-.35 in } \end{array } } \ ] ] [ propaing ] the class @xmath3 is a subclass of @xmath98 , which is itself a subclass of @xmath99 .
these class inclusions are illustrated in figure [ figclassinclusions ] .
suppose @xmath7 is a permutation in @xmath4 of length @xmath30 , and let @xmath22 be the maximal increasing subsequence of @xmath7 of the form @xmath68 for some @xmath69 .
if @xmath70 , then , by the maximality of @xmath22 , the term @xmath71 occurs to the right of @xmath72 in @xmath7 .
now let @xmath100 be the maximal increasing prefix @xmath101 of @xmath102 .
if @xmath103 , then , by the maximality of @xmath100 , we have @xmath104 .
we claim that @xmath7 can be gridded in @xmath105 by placing the points of @xmath22 in the upper right cell , the points of @xmath100 in the lower left cell , and any remaining points in the lower right cell .
clearly , both @xmath22 and @xmath100 avoid @xmath64 .
suppose there was an occurrence of @xmath65 in the lower right cell .
the point @xmath72 ca nt occur to the left of the @xmath65 , since that would form a @xmath0 .
thus @xmath71 occurs to the right of @xmath72 , @xmath103 , and thus @xmath104 .
but that is impossible , because the points @xmath106 , @xmath107 , @xmath72 and @xmath71 would form a @xmath1 .
on the other hand , we claim that there is no gridding of @xmath108 in @xmath105 . from proposition
[ proph ] , we know that such a gridding would require at least one point in the lower left cell .
moreover , only the first point of @xmath63 can occur in that cell without creating a @xmath64 there .
but , if the lower left cell were to contain only the first point of @xmath63 , then its remaining points would form a @xmath65 in the lower right cell .
let us now determine the structure of griddings in @xmath105 of permutations in @xmath4 .
given a gridding of a permutation in @xmath105 , we refer to the points in the lower left cell as the _ left points_. we also make use of the concept of a top or left point _ splitting _ ( part of ) a tree in the lower right cell . a top point splits ( part of ) a tree if it occurs to the right of its leftmost point and to the left of its rightmost point .
analogously , a left point splits ( part of ) a tree if it occurs above its lowermost point and below its uppermost point .
splitting is illustrated in the figures by pale stripes .
[ propagrid ] a gridding of a permutation in @xmath105 avoids @xmath1 if and only if no left point splits a tree below its uppermost non - trunk vertex , and if a left point splits a tree , then no top point splits its trunk
. @xmath58 \path [ fill = blue!15 ] ( 0.5,1.8 ) rectangle ( 5.5,2.2 ) ; \draw[thick ] ( 3.5,4)--(2.25,1)--(4.75,3 ) ; \plotperm[blue!50!black]{5}{2 } \plotpt[blue!50!black]{2.25}{1 } \plotpt[blue!50!black]{3.5}{4 } \plotpt[blue!50!black]{4.75}{3 } \draw[thick ] ( 0.5,4.5)--(5.5,4.5 ) ; \draw[thick ] ( 1.5,0.5)--(1.5,5.5 ) ; \node [ ] at ( 3,-0.7 ) { $ \gamma_1 $ } ; \end{tikzpicture } \qquad\quad\qquad \begin{tikzpicture}[scale=0.25 ] \path [ fill = blue!15 ] ( 0.5,2.3 ) rectangle ( 5.5,2.7 ) ; \path [ fill = blue!15 ] ( 3.3,0.5 ) rectangle ( 3.7,5.5 ) ; \draw[thick ] ( 2.25,1.25)--(4.75,3.75 ) ; \plotperm[blue!50!black]{5}{2.5 } \plotpt[blue!50!black]{2.25}{1.25 } \plotpt[blue!50!black]{3.5}{5 } \plotpt[blue!50!black]{4.75}{3.75 } \draw[thick ] ( 0.5,4.5)--(5.5,4.5 ) ; \draw[thick ] ( 1.5,0.5)--(1.5,5.5 ) ; \node [ ] at ( 3,-0.7 ) { $ \gamma_2 $ } ; \end{tikzpicture } \vspace{-9pt}\ ] ] if a left point @xmath109 splits a tree below its uppermost non - trunk vertex , then @xmath109 and the root , tip and uppermost non - trunk vertex of the tree form a @xmath1 .
similarly , if a left point @xmath109 splits a tree and a top point @xmath75 splits its trunk , then @xmath109 , the root of the tree , @xmath75 and the tip of the tree form a @xmath1 . on the other hand
, there are only two griddings of @xmath1 in @xmath105 , which we denote @xmath110 and @xmath111 as shown in figure [ fig2143gridding ] .
a gridding of a permutation in @xmath105 can only contain an occurrence of @xmath110 if the left point of @xmath110 splits a tree below its uppermost non - trunk vertex . if a gridding of a permutation in @xmath105 contains an occurrence of @xmath111 but no occurrence of @xmath110 , then the left point of @xmath111 splits a single edge in the trunk of a tree and the top point of @xmath111 splits the same edge .
we are now in a position to enumerate @xmath4 .
[ thmagf ] the generating function for @xmath3 is @xmath112 @xmath58 \path [ fill = blue!15 ] ( 0.5,1.8 ) rectangle ( 5.5,2.2 ) ; \draw[thick ] ( 3.5,1)--(4.75,3 ) ; \plotperm[blue!50!black]{5}{2 } \plotpt[blue!50!black]{2.25}{4 } \plotpt[blue!50!black]{3.5}{1 } \plotpt[blue!50!black]{4.75}{3 } \draw[thick ] ( 0.5,4.5)--(5.5,4.5 ) ; \draw[thick ] ( 1.5,0.5)--(1.5,5.5 ) ; \end{tikzpicture } \vspace{-9pt}\ ] ] @xmath58 \path [ fill = blue!15 ] ( 4.8,0.5 ) rectangle ( 5.2,22.5 ) ; \path [ fill = blue!15 ] ( 7.8,0.5 ) rectangle ( 8.2,22.5 ) ; \path [ fill = blue!15 ] ( 11.8,0.5 ) rectangle ( 12.2,22.5 ) ; \path [ fill = blue!15 ] ( 12.8,0.5 ) rectangle ( 13.2,22.5 ) ; \path [ fill = blue!15 ] ( 20.8,0.5 ) rectangle ( 21.2,22.5 ) ; \path [ fill = blue!15 ] ( 0.5,3.8 ) rectangle ( 22.5,4.2 ) ; \path [ fill = blue!15 ] ( 0.5,13.8 ) rectangle ( 22.5,14.2 ) ; \draw[thick ] ( 20,5)--(18,1)--(22,2 ) ; \draw[thick ] ( 10,6)--(11,7)--(14,11 ) ; \draw[thick ] ( 11,7)--(15,8)--(16,10 ) ; \draw[thick ] ( 15,8)--(17,9 ) ; \draw[thick ] ( 7,15)--(6,12)--(9,13 ) ; \plotperm[blue!50!black]{22}{4,14,17,16,18,12,15,19,13,6,7,20,21,11,8,10,9,1,3,5,22,2 } \draw[thick ] ( 0.5,17.5)--(22.5,17.5 ) ; \draw[thick ] ( 2.5,0.5)--(2.5,22.5 ) ; \draw[thick , dashed ] ( 0.5,16.5)-- ( 3.5,16.5)-- ( 3.5,22.5 ) ; \draw[thick , dashed ] ( 0.5,15.5)-- ( 5.5,15.5)-- ( 5.5,22.5 ) ; \draw[thick , dashed ] ( 0.5,11.5)-- ( 9.5,11.5)-- ( 9.5,22.5 ) ; \draw[thick , dashed ] ( 0.5 , 5.5)--(17.5 , 5.5)--(17.5,22.5 ) ; \end{tikzpicture } \vspace{-6pt}\ ] ] for each permutation @xmath7 in @xmath4 , we consider the gridding of @xmath7 in @xmath105 in which the number of left points is minimal , with the remaining points canonically gridded ( as a gridding in @xmath66 ) in the two cells at the right .
see figure [ figdexample ] for an illustration .
we call this the _ canonical _ gridding of @xmath7 in @xmath105 .
clearly , there is a bijection between permutations in @xmath4 and their canonical griddings in @xmath105 .
the minimality of the set of left points implies that the uppermost left point is the first point in an occurrence of @xmath2 .
since the leftmost top point occurs to the right of a @xmath64 in the lower left cell , this @xmath2 must be gridded as in figure [ fig2413gridding ] .
this is the case if and only if the uppermost left point splits a tree and this tree is not the top tree . to enumerate @xmath4 , it thus suffices to enumerate griddings of permutations in @xmath105 satisfying the following four constraints , the first two of which characterise the avoidance of @xmath1 ( proposition [ propagrid ] ) , and the last two of which characterise canonical griddings : no left point splits a tree below its uppermost non - trunk vertex .
if a left point splits a tree , then no top point splits its trunk .
the uppermost left point splits a tree and this tree is not the top tree .
the leftmost top point occurs to the right of the ( tip of the ) trunk of the top tree and not adjacent to it .
let us consider the set of permutations in @xmath4 containing @xmath2 ( i.e. those whose griddings have at least one left point ) . as in the proof of proposition [ prophgf ]
, we use the symbolic method to construct this set , reusing the definitions of @xmath77 , @xmath78 and @xmath79 in in the proof of proposition [ prophgf ] .
we mark left points with @xmath109 and top points with @xmath75 for clarity ( although , again , we do nt actually need any extra catalytic variables for the enumeration ) . in figure [ figdexample ] ,
the dashed lines illustrate how top and left points are associated with the trees .
the structural equations are as follows : @xmath113 { \ppp_{\textsf{sl}}^{\textsf{t } } } & \;=\ ; ( { \ppp_{\textsf{l}}}-\ppp)\times \seq{t\+\zzz } \label{eqpleftr } \\[5pt ] { \ttt_{\textsf{sl}}^{\textsf{t } } } & \;=\ ; { \ppp_{\textsf{sl}}^{\textsf{t}}}\:\;+\;\ : { \ppp_{\textsf{sl}}^{\textsf{t}}}\times \seqplus{{\ttt^{\textsf{t } } } } \times \seqbig{\zzz \times \seq{{\ttt^{\textsf{t } } } } } \label{eqtleftr } \\[5pt ] \ddd \setminus \hhh & \;=\ ; { \ttt_{\textsf{top}}}\:\times\ : \seq{{\ttt^{\textsf{t } } } } \:\times\ : { \ttt_{\textsf{sl}}^{\textsf{t}}}\:\times\ : \seqbig{\seq{\ell\+\zzz } \times ( { \ttt^{\textsf{t}}}+ { \ttt_{\textsf{sl}}^{\textsf{t } } } ) } \:\times\ : \seq{\ell\+\zzz } \label{eqd - h}\end{aligned}\ ] ] in , we use @xmath114 to denote a path tree together with any left points that split it . in , we use @xmath115 to denote a path tree split by at least one left point , together with any top points that occur to its right . by constraint
( b ) , a path tree may not be split by both left and top points . in , we use @xmath116 to denote a tree split by at least one left point , together with any top points that occur to the right of some of its vertices . by constraint
( b ) , these top points must occur to the right of its tip .
an element of @xmath116 is either a path tree split by at least one left point , or else is constructed ( tip downwards ) from the trunk above the uppermost branching point , split by at least one left point , which may have top points to the right of the tip , at least one subtree rooted at the the uppermost branching point on the trunk , which may have top points to the right of some of its vertices , and any points on the trunk below the uppermost branching point , each possibly rooting a subtree that may have top points to the right of some of its vertices . finally , in , the set of canonical griddings of permutations in @xmath4 whose griddings
have at least one left point is constructed , starting at the upper left , from the top tree , which , by constraint ( c ) , is never split by a left point , zero or more @xmath78 trees , which are not split by a left point , a @xmath116 tree , split by the uppermost left point , zero or more @xmath78 or @xmath116 trees , each possibly with left points above it , and zero or more left points below the lowermost tree .
translation into functional equations yields : @xmath117{\seqfrac[z]{\ell\+z } } , \\[6pt ] { p_{\textsf{sl}}^{\textsf{t}}}(z , t,\ell ) & \;=\ ; \big(\!{p_{\textsf{l}}}(z , t,\ell ) - \seqplusfrac{z}\big ) \seqfrac{t\+z } , \\[6pt ] { t_{\textsf{sl}}^{\textsf{t}}}(z , t,\ell ) & \;=\ ; { p_{\textsf{sl}}^{\textsf{t}}}(z , t,\ell ) \:\;+\;\ : \seqplusfrac[{p_{\textsf{sl}}^{\textsf{t}}}(z , t,\ell)\+]{{t^{\textsf{t}}}(z , t ) } \times \seqfrac{\seqfrac[z]{{t^{\textsf{t}}}(z , t ) } } , \\[6pt ] d(z , t,\ell ) - h(z , t ) & \;=\ ; \seqfrac[{t_{\textsf{top}}}(z , t)]{{t^{\textsf{t}}}(z , t ) } \times \seqfrac[{t_{\textsf{sl}}^{\textsf{t}}}(z ,
t,\ell)]{\seqfrac{\ell\+z}\big({t^{\textsf{t}}}(z , t)+{t_{\textsf{sl}}^{\textsf{t}}}(z , t,\ell)\big ) } \times \seqfrac{\ell\+z } .\end{aligned}\ ] ] solving , expanding , simplifying and rearranging yields a ( very complicated ) algebraic expression for @xmath118 . the univariate generating function then results from setting @xmath82 and @xmath119 and simplifying . from the generating function , we can determine the asymptotic behaviour .
[ cord ] the following properties hold in the limit : asymptotically , @xmath120 , so the probability that a random large permutation avoiding @xmath0 and @xmath1 also avoids @xmath2 tends to @xmath121 .
the number of left points in the canonical gridding of a permutation in @xmath4 has asymptotic mean @xmath122 and standard deviation @xmath123 .
the number of top points in the canonical gridding of a permutation in @xmath124 has asymptotic mean @xmath85 and standard deviation @xmath86 . for part ( i ) ,
the asymptotics of @xmath125 results from the application of proposition [ propasympt ] to the first nonconstant term of the puiseux expansion for @xmath126 around its dominant singularity @xmath89 : @xmath127 the ratio @xmath128 gives the proportion of permutations in @xmath4 avoiding @xmath2 .
the asymptotic limit of this ratio is @xmath129 . for part ( ii ) , substituting @xmath82 and simplifying yields @xmath130 where @xmath131 q(z,\ell ) & \;=\ ; ( 2 - 3\+\ell ) \:-\ : z \+
( 3 - 5\+\ell - 8\+\ell^2 ) \:+\ : z^2\+ ( 1 - 3\+\ell - 15\+\ell^2 - 7\+\ell^3 ) \\ & \qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad \:+\ : z^3\+ ( \ell + 8\+\ell^2 + 9\+\ell^3 + 2\+\ell^4 ) \:-\ : z^4\+ ( \ell^2 + 2\+\ell^3 + \ell^4 ) .\end{aligned}\ ] ] the asymptotic mean and standard deviation for the number of left points follow by the application of propositions [ propasympt ] and [ propmoments ] .
the limiting distribution is illustrated in figure [ figleftpointdistrib ] . ] for part ( iii ) , substituting @xmath119 and simplifying yields @xmath132 where @xmath133 s(z , t ) & \;=\ ; - 1 \:+\ : ( 7 + 3\+t + t^2)\+z \:-\ : ( 14 + 14\+t + 6\+t^2)\+z^2 \:+\ : ( 9 + 22\+t + 13\+t^2)\+z^3 \\ & \qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad \:-\ : ( 12\+t + 12\+t^2)\+z^4 \:+\ : 4\+t^2 \+z^5 .\end{aligned}\ ] ] the asymptotic mean and standard deviation for the number of left points follow by the application of propositions [ propasympt ] and [ propmoments ] .
the first twelve terms of the sequence @xmath125 are 1 , 2 , 6 , 22 , 88 , 366 , 1556 , 6720 , 29396 , 129996 , 580276 , 2611290 .
more values can be found at http://oeis.org/a165537[a165537 ] in oeis @xcite .
we have been unable to determine whether the number of top points in a permutation in @xmath4 has a gaussian limit distribution .
although @xmath134 can easily be expressed in the form @xmath135 , when this is done , both @xmath136 and @xmath137 are singular at @xmath138 .
thus , proposition [ propgaussian ] is not applicable and ca nt be used to deduce an asymptotic gaussian distribution .
it would be interesting to establish whether top points are asymptotically normal as would be expected , or whether the presence of left points somehow distorts the distribution .
the somewhat unexpected result that more than half of the permutations in @xmath139 are , in fact , members of @xmath13 , is a consequence of the fact that @xmath128 has a positive limit as @xmath30 tends to infinity .
this phenomenon deserves further investigation , as does the weaker property that the growth rates of @xmath66 and @xmath4 are equal : @xmath140 .
we conclude with the observation which initiated this investigation .
suppose @xmath143 is a finite collection of permutation classes .
we say that another permutation class @xmath16 is _
@xmath144-griddable _ if there is a finite matrix @xmath8 each entry of which is one of the @xmath145 , such that @xmath16 is a subset of the generalised grid class @xmath6 .
vatter gave the following characterisation of the property of being .
[ propkgriddable ] the permutation class @xmath16 is @xmath144-griddable if and only if it does not contain arbitrarily long sums or skew sums of basis elements of members of @xmath144 , that is , if there exists a constant @xmath146 so that @xmath16 does not contain @xmath147 or @xmath148 for any sequence @xmath149 of basis elements of members of @xmath144 . | we determine the structure of permutations avoiding the patterns @xmath0 and @xmath1 .
each such permutation consists of the skew sum of a sequence of plane trees , together with an increasing sequence of points above and an increasing sequence of points to its left .
we use this characterisation to establish the generating function enumerating these permutations .
we also investigate the properties of a typical large permutation in the class and prove that if a large permutation that avoids @xmath0 and @xmath1 is chosen uniformly at random , then it is more likely than not to avoid @xmath2 as well . |
we should like to thank a. hoang and a. leike for useful discussions .
10 x. calmet and h. fritzsch , phys .
b * 496 * , 161 ( 2000 ) [ hep - ph/0008243 ] , x. calmet , phys .
b * 510 * , 221 ( 2001 ) [ hep - th/0008189 ] , x. calmet and h. fritzsch , hep - ph/0103333 . s. l. glashow , nucl . phys .
* 22 * , 579 ( 1961 ) , s. weinberg , phys . rev .
* 19 * , 1264 ( 1967 ) , a. salam , elementary particle physics , in proceedings of the 8th nobel symposium , 1968 .
k. osterwalder and e. seiler , annals phys .
* 110 * ( 1978 ) 440 , e. fradkin and s. h. shenker , phys .
* d19 * ( 1979 ) 3682 , phys . lett . * b164 * ( 1985 ) 121 . g. t hooft , in erice 1998 , from the planck length to the hubble radius , 216 - 236 , hep - th/9812204 , hep - th/9812204 , g. t hooft , in `` recent developments in gauge theories '' , carges 1979 , ed .
t hooft et al .
plenum press , new york , 1980 , lecture ii , p.117 , see also t. banks and e. rabinovici , nucl . phys .
* b160 * ( 1979 ) 349 , s. dimopoulos , s. raby and l. susskind , nucl .
* b173 * ( 1980 ) 208 .
m. veltman , acta phys .
b * 12 * , 437 ( 1981 ) , g. passarino and m. veltman , nucl .
b * 160 * , 151 ( 1979 ) .
j. fleischer and f. jegerlehner , phys .
d * 23 * , 2001 ( 1981 ) .
e. ma , phys .
d * 47 * , 2143 ( 1993 ) [ hep - ph/9209221 ] .
m. sher , phys .
rept . * 179 * , 273 ( 1989 ) , m. sher , phys . lett .
b * 317 * , 159 ( 1993 ) [ b * 331 * , 448 ( 1993 ) ] [ hep - ph/9307342 ] , j. r. espinosa and m. quiros , phys .
b * 353 * , 257 ( 1995 ) [ hep - ph/9504241 ] , j. a. casas , j. r. espinosa and m. quiros , phys .
b * 342 * , 171 ( 1995 ) [ hep - ph/9409458 ] .
t. kawamoto , contributed to 36th rencontres de moriond on qcd and hadronic interactions , les arcs , france , 17 - 24 mar 2001 , hep - ex/0105032 .
x. calmet and h. fritzsch , phys .
* b496 * , 190 ( 2000 ) [ hep - ph/0008252 ] . | we compute the higgs mass in a model for the electroweak interactions based on a confining theory .
this model is related to the standard model by the complementarity principle .
a dynamical effect due to the large typical scale of the higgs boson shifts its mass above that of the w - bosons .
we obtain @xmath0 gev .
@sh[#1]#2 fmsl@sh fmsl@sh fmsl@sh fmsl@sh#1#2#3@th hep - ph/0107085 + lmu 01/10 + xavier calmet + and harald fritzsch + _ ludwig - maximilians - university munich , sektion physik _ + _ theresienstrae 37 , d-80333 munich , germany _
recently we have proposed a model for the electroweak interactions based on a confining @xmath1 theory @xcite .
it was shown that , at least at low energies , this model is complementary or dual to the electroweak standard model @xcite .
the complementarity principle states that there is no phase transition between the higgs and the confinement phase if there is a higgs boson in the fundamental representation of the gauge group @xcite .
the lagrangian of the theory under consideration is exactly that of the standard model before gauge symmetry breaking .
however the sign of the higgs boson squared mass is changed , i.e. it is positive , and the gauge symmetry is thus unbroken .
we have the following fundamental left - handed dual - quark doublets , which we denote as d - quarks : + [ cols= " < , < , < " , ] + the right - handed particles are those of the standard model . as the gauge symmetry is unbroken , physical particles must be singlets under @xmath1 transformation , and we thus get the following particle spectrum @xmath2 where @xmath3 is the coupling constant of the gauge group @xmath4 and @xmath5 is the corresponding covariant derivative .
we have used the unitary gauge @xmath6 where @xmath7 is the parameter appearing in the expansion of the bound states ( [ def2 ] ) . matching the expansion of the higgs boson to the standard model we get @xmath8 gev @xcite . using this expansion , we can associate a certain scale , which is proportional to @xmath7 , to each particle .
the scale of the @xmath9-bosons is then @xmath10 gev . as can be seen from the expansion for the higgs boson ( [ def2 ] ) ,
there is a factor four between the expansion parameter of the higgs boson and that of the w - bosons , thus one finds @xmath11 gev .
this factor four is dictated by the algebraic structure of the underlying gauge theory . in the confinement phase
the higgs boson is the @xmath12-wave of the @xmath1 theory , whereas the @xmath9-bosons are the corresponding @xmath13-waves .
thus one expects the higgs boson to be lighter than the @xmath9-bosons .
but , as we shall show , a dynamical effect shifts the higgs boson mass above that of the @xmath9-bosons mass .
the reason for this phenomenon is the large higgs boson scale compared to that of the @xmath9-bosons .
the masses of the physical higgs and w - bosons , being bound states consist of a constituent mass @xmath14 , where @xmath15 is the mass of the scalar @xmath16-quark and of dynamical contributions .
we have to consider two types of diagrams : the one - particle reducible diagrams ( 1pr ) and the one - particle irreducible diagrams ( 1pi ) . for the higgs boson mass
, we have to take the self - interaction and the contribution of the @xmath17 and @xmath18-bosons into account ( see figures [ fig1 ] , [ fig2 ] and [ fig3 ] ) . the fermions couple via yukawa coupling to the higgs boson , and as this interaction is not confining , fermions can not contribute to the dynamical mass of the higgs boson .
the first task is to extract the constituent mass from the experimentally measured @xmath9-bosons mass .
the fermions contribute to the dynamical mass of the @xmath9-bosons as they couple via @xmath1 couplings to the electroweak bosons but the divergence is only logarithmic @xcite and we shall only keep the quadratic divergences . [ fig3 ] ] [ fig3 ] ] [ fig3 ] ] [ fig6 ] ] [ fig6 ] ] [ fig6 ] ] we have considered the tadpoles and the one - particle - irreducible contributions at the one loop order ( the diagrams contributing to the @xmath9-bosons mass are similar to those contributing to the higgs - boson mass ) . using the duality described in @xcite , these duality diagrams can be related to the feynman graphs of figures [ fig4 ] , [ fig5 ] and [ fig6 ] .
the feynman graphs have been evaluated in ref .
@xcite as a function of a cut - off parameter and we will only keep the dominant contribution which is quadratically divergent .
we obtain : @xmath19 this equation can be solved for @xmath20 : @xmath21 we can now compute the dynamical contribution to the higgs boson mass . the exact one loop , gauge invariant counterterm has been calculated in refs .
@xcite , @xcite and @xcite . using the results of ref .
@xcite , where this counterterm was calculated as a function of a cut - off , we obtain : @xmath22 the unknown of this equation is the higgs boson s mass @xmath23 .
this equation can be solved by numerical means .
we obtain two positive solutions : @xmath24=14.1 gev and @xmath25=129.6 gev .
the first solution yields an imaginary constituent mass and is thus also discarded .
the second solution is the physical higgs boson mass .
we thus obtain @xmath26=129.6 gev in the one loop approximation .
the constituent mass is then @xmath27=78.8 gev . as expected the dynamical contribution to the @xmath9-bosons masses
is small and the higgs boson mass is shifted above that of that of the @xmath9-bosons mass because of the large higgs boson scale .
note that our prediction @xmath26=129.6 gev is in good agreement with the requirement of vacuum stability in the standard model which requires the mass of the higgs boson to be in the range 130 gev to 180 gev if the standard model is to be valid up to a high energy scale @xcite .
we can thus deduce that the duality we have described @xcite must also be valid up to some high energy scale .
our result is also in good agreement with the expectation @xmath28 gev based on electroweak fits @xcite .
this has also consequences for the model proposed in @xcite , where we assumed that the higgs boson does not couple to @xmath29-quarks . because of the dynamical effect we have discussed , the higgs boson is relatively heavy and should decay predominantly into electroweak bosons . |
the first exact results @xcite for the o(@xmath0 ) critical properties were obtained for a model on the honeycomb lattice , and revealed not only the critical point , but also some universal parameters of the critical state , as well as the low - temperature phase , as a function of @xmath0 .
the derivation of these results depends on a special choice of the o(@xmath0)-symmetric interaction between the @xmath0-component spins of the o(@xmath0 ) model , which enables a mapping on a loop gas @xcite .
these results were supposed to apply to a whole universality class of o(@xmath0)-symmetric models in two dimensions . since then , also o(@xmath0 ) models on the square and triangular lattices were investigated @xcite .
indeed , branches were found with the same universal properties as the honeycomb model , but in addition to these , several other branches of critical behavior were reported . among these , we focus on ` branch 0 ' as reported in refs .
the points on this branch appear to describe a higher critical point . for @xmath4 ,
it can be identified with the so - called @xmath5 point @xcite describing the collapse of a polymer in two dimensions , which has been interpreted as a tricritical o(@xmath4 ) model .
it has indeed been found that the introduction of a sufficiently strong and suitably chosen attractive potential between the loop segments changes the ordinary o(@xmath4 ) transition into a first - order one @xcite , such that this change precisely coincides with the @xmath4 point of branch 0 .
thus , the @xmath5 point plays the role of a tricritical o(@xmath4 ) transition .
furthermore , it has been verified that tricriticality in the o(@xmath0 ) model can be introduced by adding a sufficient concentration of vacancies into the system @xcite .
more precisely , the introduction of vacancies leads to a branch of higher critical points , of which the points @xmath4 and @xmath7 belong to universality classes ( of the @xmath5 point and the tricritical blume - capel model respectively ) that have been described earlier as tricritical points . however , the critical points of branch 0 on the square lattice appear to display universal properties that are different from those of the branch of higher critical points of the o(@xmath0 ) model with vacancies @xcite , except at the intersection point of the two branches at @xmath4 .
it thus appears that the continuation of the @xmath5 point at @xmath4 to @xmath8 can be done in different ways , leading to different universality classes . in order to gain further insight in this situation
, the present work considers an o(@xmath0 ) loop model on the kagome lattice with the purpose to find a @xmath5-like point , to continue this point to @xmath8 and to explore the resulting universality .
the partition function of the spin representation of @xmath1-state potts model on the honeycomb lattice @xmath9 depends on the temperature @xmath10 by the coupling @xmath11 , where @xmath12 is the nearest - neighbor spin - spin interaction .
the spins @xmath13 can assume values 1 , 2 , @xmath14 , @xmath1 and their index @xmath15 labels the sites of the honeycomb lattice .
the first summation is over all possible spin configurations @xmath16 , and the second one is over the nearest neighbor spin pairs .
this potts model can be subjected to a series of mappings which lead , via the random - cluster model and a fully - packed loop model , to a dilute o(@xmath0 ) loop model which can also be interpreted as an o(@xmath0 ) spin model .
the introduction of bond variables , and a summation on the spin variables map the potts model onto the random - cluster ( rc ) model @xcite , with partition function @xmath17 where @xmath18 is the number of bonds , @xmath19 the number of clusters , and @xmath20 the weight of a bond .
the sum is on all configurations @xmath21 of bond variables : each bond variable is either 1 ( present ) or 0 ( absent ) . in eq .
( [ zrc ] ) , @xmath1 can be considered a continuous real number , playing the role of the weight of a cluster . here ,
a cluster is either a single site or a group of sites connected together by bonds on the lattice . a typical configuration of the rc model on the honeycomb lattice is shown in fig .
[ mapping ] .
the next step is a mapping of the rc model on the honeycomb lattice onto a fully packed loop ( fpl ) model on the kagome lattice , which proceeds similarly as in the case of the square lattice @xcite .
the sites of fpl model sit in the middle of the edges of the honeycomb lattice , and thus form a kagome lattice @xcite .
fully packed here means that all edges of the kagome lattice are covered by loop segments .
the one - to - one correspondence between these two configurations is established by requiring that the loops do not intersect the occupied edges ( bonds ) of the honeycomb rc model , and always intersect the empty edges , as illustrated in fig .
[ mapping ] . according to eq .
( [ zrc ] ) , and that of the corresponding fpl configuration is @xmath22 according to eq .
( [ zfpl ] ) . ] to specify the boltzmann weights of the fpl model , we assign a weight @xmath0 to each loop , a weight @xmath23 to each vertex where the loop segments do not intersect an edge which is occupied by a bond of the rc model , and a weight @xmath24 to each vertex where the loop segments intersect an edge which is empty in the rc model , as illustrated in fig .
[ 2vs ] . .
the weight of a vertex where two loops intersect an unoccupied edge ( dashed line ) is @xmath24 . ] the partition function of the fpl model on the kagome lattice is thus defined as @xmath25 where @xmath26 is the number of type-1 vertices , @xmath27 is the number of type-2 vertices and @xmath28 the number of loops .
the sum is on all configurations @xmath29 of loops covering all the edges of the kagome lattice .
the one - to - one correspondence between rc configurations and fpl configurations makes it possible to express the configuration parameters @xmath26 , @xmath27 and @xmath28 of the fpl in those of the rc model , namely @xmath18 and @xmath19 .
each vertex of type-1 corresponds with a bond of the rc model on the honeycomb lattice , thus @xmath30 the total number of the two kinds of vertices is equal to the number of edges on the honeycomb lattice , i.e. , @xmath31 where @xmath32 is the total number of sites of the honeycomb lattice .
here we ignore surface effects of finite lattices .
furthermore , a loop on the kagome lattice is either one surrounding a random cluster on the honeycomb lattice , or one following the inside of a loop formed by the bonds of a random cluster .
thus @xmath33 where @xmath34 is the loop number of the rc model . together with the euler relation @xmath35 eqs . ( [ e1 ] ) to ( [ e3 ] ) yield the numbers of vertices and loops on the kagome lattice
as @xmath36 substitution in the partition function ( [ zfpl ] ) leads to @xmath37 the weight of a given loop configuration is thus equal to the corresponding rc weight @xmath38 if @xmath39 which completes the mapping of the rc onto the fpl model .
next we map the fpl model on the kagome lattice onto a dilute loop ( dl ) model on the same lattice , using a method due to nienhuis ( see e.g. ref .
the partition function of the fpl model on the kagome lattice is slightly rewritten as @xmath40^{m_l } \label{zfpl2}\ ] ] with @xmath41 and @xmath42 .
( [ zfpl2 ] ) invites an interpretation in terms of colored loops , say red with loops of weight @xmath43 and green loops of weight 1 .
each of the @xmath44 terms in the expansion of @xmath45^{m_l}$ ] thus specifies a way to color the loops : @xmath46^{m_l}=\sum _ { \{\rm colorings\}}(n-1)^{l_r}1^{l_g } \ , , \ ] ] where @xmath47 and @xmath48 denote the number of red loops and green loops respectively , @xmath49 .
let @xmath50 denote a graph @xmath29 in which the colors of all loops are specified .
the partition sum can thus be expressed in terms of a summation over all colored loop configurations @xmath50 .
the vertices of the kagome lattice are visited by two colored loops , and can thus be divided into 6 types , shown in fig .
[ 6vs ] with their associated weights @xmath51 and @xmath52 .
thus , eq . ( [ zfpl2 ] ) assumes the form @xmath53 the sum @xmath54 on all colored loop configurations may now be replaced by two nested sums , the first of which is a sum @xmath55 on all dilute loop configurations of red loops , and the second sum @xmath56 is on all configurations of green loops @xmath57 that are consistent with @xmath58 , i.e. , the green loop configurations that cover all the kagome edges not covered by a red loop .
thus @xmath59 for each vertex visited by green loops only , there are precisely two possible local loop configurations . since the loop weight of the green loops is 1 , the summation over such pairs of configurations is trivial : @xmath60 where @xmath61 is the number of green - only vertices .
the fpl partition sum thus reduces to that of a dilute loop model , involving only red loops of weight @xmath43 : @xmath62 where the partition function of the dilute loop model is defined as @xmath63 in which we forget the color variable , and denote the number of loops in a dilute configuration @xmath64 as @xmath34 .
the dilute vertices are shown in fig .
[ 5vsdl ] , together with their weights .
the exponents of the vertex weights in eq .
( [ zdl ] ) represent the numbers of the corresponding vertices . because of the similarity with the derivation of branch 0 on the square lattice
, we refer to the model ( [ zdl ] ) as branch 0 of the kagome o(@xmath0 ) loop model . the transformation between the fpl and the dl model is illustrated in fig .
[ pdl ] . .
] or @xmath65 ( see fig .
[ 6vs ] ) . for the simple case shown here ,
there are two possible configurations ( a ) and ( b ) , of which the relative weights are @xmath66 and @xmath67 respectively .
addition of these weights yields the weight @xmath68 of the dl configuration shown in ( c ) . ] the boltzmann weights in eq .
( [ zdl ] ) contain , besides the loop weights , only local weights associated with the vertices of the kagome lattice . just as in the case of the
o(@xmath0 ) model on the square lattice described in ref .
@xcite , there are precisely four incoming edges at each vertex .
this implies that there is an equivalent o(@xmath0 ) spin model : @xmath69 of which the local weights have the same relation with the vertex weights as for the square lattice model of ref .
thus , the partition sum of the spin model is expressed by @xmath70 \prod_{v } [ 1 + x_1 ( \vec{s}_{v1}\cdot\vec{s}_{v2}+ \vec{s}_{v3}\cdot\vec{s}_{v4})+ \mbox{\hspace{10 mm}}\ ] ] @xmath71 \ , .
\label{spin}\ ] ] the product is on all vertices @xmath72 of the kagome lattice .
the spins @xmath73 sit on the midpoints of the edges of the kagome lattice .
their subscript `` @xmath74 '' specifies the vertex @xmath72 as well as the position @xmath15 ( with @xmath75 ) with respect to the vertex .
the label @xmath76 runs clockwise around each vertex , such that the spins @xmath77 and @xmath78 sit on the same side of the honeycomb edge passing through vertex @xmath72 .
the spins have @xmath0 cartesian components and are normalized to length @xmath79 .
there are two different notations for each spin ( because each spin is adjacent to two vertices ) , but a given subscript @xmath74 refers to only one spin . here
the number @xmath0 is restricted to positive integers , of which only the case @xmath7 is expected to be critical . since the critical point of the rc model on the honeycomb lattice is known @xcite as a function of @xmath1 , namely @xmath80 the corresponding critical point of the @xmath81 fpl model on the kagome lattice is also known . according to eq .
( [ wfpl ] ) @xmath82 from which the corresponding critical point of the dl model with loop weight @xmath83 on the kagome lattice follows as @xmath84
the transformations described in sec . [ mapsec ] leave ( apart from a shift by a constant ) the free energy unchanged , and lead to relations between the thermodynamic observables of the various models .
thus , the conformal anomaly and some of the critical exponents of the fpl and the dl models can be obtained from the existing results for the random - cluster model .
thus , like in the analogous case of the o(@xmath0 ) model on the square lattice @xcite , the fpl model on the kagome lattice should be in the universality class of the low - temperature o(@xmath0 ) phase .
however , the representation of magnetic correlations in our present cylindrical geometry leads to a complication .
the kagome lattice structure , together with the fpl constraint , imposes the number of loop segments running along the cylinder to be even .
since the o(@xmath0 ) spin - spin correlation function is represented by a single loop segment in the loop representation , which can not be embedded in an fpl model on the kagome lattice , it is not clear how to represent magnetic correlations in this model .
thus we abstain from a further discussion of the scaling dimensions of the fpl model .
_ conformal anomaly _ for the fpl model with loop weight @xmath0 on the kagome lattice ,
the conformal anomaly @xmath85 is equal to that of the @xmath86 potts model @xcite : @xmath87 in the coulomb gas language @xcite , it can be expressed as a function of the coulomb gas coupling constant @xmath88 , with @xmath89 : @xmath90 the conformal anomaly @xmath85 of the branch-@xmath91 critical o(@xmath0 ) dl model on the kagome lattice with loop weight @xmath0 is given by the same formula , but with @xmath0 replaced by @xmath92 : @xmath93 the conformal anomaly is , via the number @xmath94 , related to a set of scaling dimensions @xmath95 as determined by the kac formula @xcite : @xmath96 ^ 2 - 1}{2 m ( m+1 ) } \ , , \label{cfxi}\ ] ] where @xmath97 and @xmath98 are integers for unitary models .
_ temperature exponent _ for the branch-@xmath91 critical dl model with loop weight @xmath0 on the kagome lattice
, the temperature exponent is expected to be the same as that for branch 0 on the square lattice @xcite , namely @xmath99 with @xmath100 in eq .
( [ cfxi ] ) .
_ magnetic exponent _ the magnetic exponent of the branch-@xmath91 dl model with @xmath4 on the kagome lattice is _ not _ equal to the magnetic exponent of the low temperature o(@xmath92 ) loop model .
the same situation was found earlier for the branch-@xmath91 o(@xmath91 ) model on the square lattice @xcite . according to the reason given in @xcite , the magnetic exponent is equal to the temperature one , i.e. , the @xmath101 entry of eq .
( [ cfxi ] ) .
the geometry of the underlying fpl model , where the number of dangling bonds is restricted to be even , plays here an essential role .
note that the magnetic exponent of the tricritical dilute o(@xmath0 ) model @xcite , even at the @xmath5 point , is different from that of branch 0 .
these results for @xmath102 and @xmath103 are expressed in the coulomb gas language as @xmath104
the transfer matrix is constructed for an @xmath105 loop model wrapped on a cylinder , with its axis perpendicular to one of the lattice edge directions of the kagome lattice .
the finite size @xmath106 is defined such that the circumference of the cylinder is spanned by @xmath107 elementary hexagons ( corner to corner ) .
the cylinder is divided into @xmath108 slices , of which @xmath106 sites form a cyclical row , while each of the @xmath107 remaining sites forms an equilateral triangle with two of the sites of the cyclical row .
the length of the cylinder is thus @xmath109 .
the partition function of this finite - size dl model is given by eq .
( [ zfpl ] ) , but with @xmath110 instead of @xmath111 , in order to specify the length @xmath108 of the cylinder : @xmath112 there are open boundaries at both ends of the cylinder , so that there are @xmath106 dangling edges connected to the vertices on row @xmath76 , as well as on row @xmath108 .
the way in which the end points of the dangling edges are pairwise connected by the loop configuration @xmath110 is defined as the ` connectivity ' , see ref .
@xcite for details .
here we ignore the dangling edges of row 1 ( except for a topological property that will be considered later ) and focus on the @xmath106 dangling edges of row @xmath108 .
since it is determined by the loop configuration , the connectivity @xmath113 at row @xmath108 is written as a function of @xmath110 : @xmath114 .
the partition sum is divided into a number of restricted sums @xmath115 , each of which collects all terms in @xmath116 having connectivity @xmath113 on row @xmath108 , i.e. : @xmath117 an increase of the system length @xmath108 to @xmath118 leads to a new configuration @xmath119 which can be decomposed in @xmath110 and the appended configuration @xmath120 on row @xmath118 .
the graph @xmath120 fits the dangling edges of the loop graph @xmath110 on the @xmath108-row lattice .
the addition of the new row increases the number of the four kinds of vertices and of the number of loops by @xmath121 , @xmath122 , @xmath123 , @xmath124 and @xmath125 respectively .
the restricted partition sum of the system with @xmath118 rows is @xmath126 @xmath127 the last sum is on all sub - graphs @xmath128 that fit @xmath110 .
the connectivity @xmath129 depends only on the connectivity @xmath113 on row @xmath108 , and on @xmath120 , so that we may write @xmath130 .
thus eq . ( [ zm1 ] ) assumes the form @xmath131 the third sum depends only on @xmath132 and @xmath113 , and thus defines the elements of the transfer matrix @xmath133 as @xmath134 substitution of @xmath135 and eq .
( [ psum ] ) in eq .
( [ lrec ] ) then yields the recursion of the restricted partition sum as @xmath136 in order to save memory and computer time , the transfer matrix of a system with finite size @xmath106 is decomposed in @xmath137 sparse matrices : @xmath138 where @xmath139 denotes an operation which adds a new vertex @xmath15 on a new row , as illustrated in fig .
most of these sparse matrices are square , but @xmath140 is not , because it increases the number of dangling bonds by two .
the action of the other rectangular matrix , @xmath141 , reduces the number of dangling bonds again to @xmath106 . during the actual calculations , we only store the positions and values of the non - zero elements of a sparse matrix , in a few one - dimensional arrays . moreover ,
this need not be done for all the sparse matrices , because there are only four independent matrices .
the other ones are related to these by the action of the translation operator @xcite .
steps and is denoted @xmath142 ( which are executed from right to left )
. each step adds a new site to the lattice .
two of these steps are illustrated in ( a ) to ( c ) .
the number of dangling bonds does not change during these steps .
the second part consists of @xmath106 steps and is denoted @xmath143 .
the first step of these , @xmath144 , adds a new vertex to the sub - row and increases the number of dangling bonds by 2 as shown in ( d ) .
the following steps @xmath145 append vertices sequentially , and do not change the number of dangling bonds .
after adding the last vertex by @xmath146 to the sub - row , the construction of a new row has been completed and the size of the system shrinks from @xmath147 to @xmath106 . ] while the construction of the transfer matrix is formulated in terms of connectivities on the topmost rows @xmath108 and @xmath118 , the connectivity on row 1 is not entirely negligible . in particular , the number of dangling loop segments on that row can be even or odd . as a consequence
the number of dangling loop segments on the topmost row is then also even or odd respectively .
this leads to a decomposition of the transfer matrix in an even and an odd sector .
the odd sector corresponds with a single loop segment running in the length direction of the cylinder . for a model on an infinitely long cylinder with finite size @xmath106 ,
the free energy per unit of area is determined by @xmath148 where @xmath149 is the largest eigenvalue of @xmath10 in the @xmath150 sector . from the finite - size data for @xmath151 we estimated the conformal anomaly
@xmath85 @xcite .
the magnetic correlation length @xmath152 is related to the magnetic gap in the eigenvalue spectrum of @xmath10 as @xmath153 where @xmath154 is the largest eigenvalue of @xmath10 in the @xmath155 sector .
the temperature correlation length @xmath156 is related to the temperature gap in the eigenvalue spectrum of @xmath10 as @xmath157 where @xmath158 is the second largest eigenvalue of @xmath10 in the @xmath150 sector . using cardy s conformal mapping @xcite of an infinite cylinder on the infinite plane
, one can thus estimate the temperature dimension @xmath102 and @xmath103 .
we calculated the finite - size data for the free energies of the fpl model at the critical points given by eq .
( [ cpfpl ] ) for system sizes @xmath159 .
these data include the case @xmath4 ; this is possible because , for @xmath160 one has @xmath161 , so that the ratio between @xmath162 and @xmath163 in eq .
( [ cpfpl ] ) remains well defined in this limit .
the additional loop configurations allowed by the dilute model lead to a larger transfer matrix for a given system size , so that our results at the critical points given by eqs .
( [ cpdl ] ) are restricted to sizes @xmath164 .
the latter results also include the temperature and magnetic gaps .
the finite - size data for the fpl and dl models displayed a good apparent convergence , and were fitted using methods explained earlier @xcite , see also ref .
@xcite . in the kagome lattice fpl model , it is not possible to introduce one single open loop segment running in the length direction of the cylinder .
the presence of a single chain would force unoccupied edges into the system , in violation of the fpl condition .
therefore , we have no results for @xmath103 . furthermore , in the case of the low - temperature o(@xmath0 ) phase , the eigenvalue associated with @xmath102 decreases rapidly when @xmath0 becomes smaller than 2 , and becomes dominated by other eigenvalues .
therefore , also results for @xmath102 are absent for the fpl model , and our results are here restricted to the conformal anomaly @xmath85 .
the resulting estimates for the fpl model are listed in tab .
[ tab1 ] .
.conformal anomaly @xmath85 of the fpl model as determined by the transfer - matrix calculations described in the text .
the sizes of the system @xmath106 are from @xmath165 to @xmath166 . estimated error margins in the last decimal place
are given in parentheses .
the numerical results are indicated by ` num ' .
for comparison , we include theoretical values indicated by ` th ' , as given by eq .
( [ cfpl ] ) . [ cols="<,<,<",options="header " , ]
we found a branch of critical points of the dilute loop model on the kagome lattice as a function of the loop weight @xmath0 , which is related to the @xmath2-state potts model on the honeycomb lattice .
the critical properties of these critical points are conjectured and verified by numerical transfer matrix calculations and a finite - size - scaling analysis .
as expected , the model falls into the same universality class as branch @xmath91 of the o(@xmath0 ) loop model @xcite on the square lattice .
the analysis did , however , yield a difference .
this is due to the geometry of the lattice .
for the square lattice , it was found @xcite that there exists a magnetic scaling dimension @xmath167 as revealed by the free - energy difference between even and odd systems .
such an alternation is absent in the free - energy of the present model on the kagome lattice .
while the number of dangling edges may be odd or even for the square lattice , it can only be even in the present case of the kagome lattice .
the numerical accuracy of the results for the conformal anomaly and the exponents is much better than what can be typically achieved for an arbitrary critical point , whose location in the parameter space has to be determined in advance by so - called phenomenological renormalization @xcite .
this seems not only due to the limited precision of such a critical point .
we suppose that the main reason is that irrelevant scaling fields tend to be suppressed in exactly solvable parameter subspaces .
we are much indebted to bernard nienhuis , for making his insight in the physics of o(@xmath0 ) loop models available to us .
this research is supported by the national science foundation of china under grant # 10675021 , by the beijing normal university through a grant as well as support from its hscc ( high performance scientific computing center ) , and by the lorentz fund ( the netherlands ) .
widest - label b. nienhuis , phys .
lett . * 49 * , 1062 ( 1982 ) ; j. stat . phys . * 34 * , 731 ( 1984 ) .
e. domany , d. mukamel , b. nienhuis and a. schwimmer , nucl .
b * 190 * [ fs3 ] , 279 ( 1981 ) . h. w. j. blte and b. nienhuis , j. phys .
a * 22 * , 1415 ( 1989 ) . y. m. m. knops , h. w. j. blte and b. nienhuis , j. phys .
a * 26 * , 495 ( 1993 ) .
b. duplantier and h. saleur , phys .
* 59 * , 539 ( 1987 ) .
h. w. j. blte , m. t. batchelor and b. nienhuis , physica a * 251 * , 95 ( 1988 ) .
guo , b. nienhuis and h. w. j. blte , int . j. mod
c * 10 * , 291 ( 1999 ) ; phys . rev .
lett . * 96 * , 045704 ( 2006 ) .
p. w. kasteleyn and c. m. fortuin , j. phys .
* 46 * ( suppl . ) , 11 ( 1969 ) ; c. m. fortuin and p. w. kasteleyn , physica ( amsterdam ) * 57 * , 536 ( 1972 ) .
r. j. baxter , s. b. kelland , f. y. wu , j. phys . a * 9 * , 397 ( 1976 ) .
i. syzi , progr . theor
* 6 * , 306 ( 1951 ) . f. y. wu , rev .
54 * , 235 ( 1982 ) .
h. w. j. blte , j. l. cardy and m. p. nightingale , phys .
lett . * 56 * , 742 ( 1986 ) .
i. affleck , phys .
lett . * 56 * , 746 ( 1986 ) .
b. nienhuis , in _ phase transitions and critical phenomena _ , eds . c. domb and j. l. lebowitz ( academic press , london , 1987 ) , vol .
* 11*. d. friedan , z. qiu and s. shenker , phys .
lett . * 52 * , 1575 ( 1984 ) .
h. w. j. blte and m. p. nightingale , physica a * 112 * , 405 ( 1982 )
. j. l. cardy , j. phys .
a * 17 * , l385 ( 1984 ) . for reviews , see e.g. m. p. nightingale in _ finite - size scaling and numerical simulation of statistical systems _ , ed .
v. privman ( world scientific , singapore 1990 ) , and m. n. barber in _ phase transitions and critical phenomena _ , eds . c. domb and j. l. lebowitz ( academic , new york 1983 ) , vol .
* 8*. m. p.
nightingale , proc .
b * 82 * , 235 ( 1979 ) . | a critical dilute o(@xmath0 ) model on the kagome lattice is investigated analytically and numerically .
we employ a number of exact equivalences which , in a few steps , link the critical o(@xmath0 ) spin model on the kagome lattice to the exactly solvable critical @xmath1-state potts model on the honeycomb lattice with @xmath2 .
the intermediate steps involve the random - cluster model on the honeycomb lattice , and a fully packed loop model with loop weight @xmath3 and a dilute loop model with loop weight @xmath0 , both on the kagome lattice .
this mapping enables the determination of a branch of critical points of the dilute o(@xmath0 ) model , as well as some of its critical properties . for @xmath4 ,
this model reproduces the known universal properties of the @xmath5 point describing the collapse of a polymer .
for @xmath6 it displays a line of multicritical points , with the same universal properties as a branch of critical behavior that was found earlier in a dilute o(@xmath0 ) model on the square lattice .
these findings are supported by a finite - size - scaling analysis in combination with transfer - matrix calculations . |
observations of type ia supernovae ( snia ) indicate that currently the observable universe is undergoing an accelerating expansion @xcite .
this cosmic acceleration has also been confirmed by numerous observations of large scale structure ( lss ) @xcite and measurements of the cosmic microwave background ( cmb ) anisotropy @xcite .
the cause for this cosmic acceleration is generally dubbed as `` dark energy '' , a mysterious exotic energy which generates large negative pressure , whose energy density is dominated the universe ( for a review see e.g. @xcite ) .
the astrophysical nature of dark energy is that it does not cluster at any scale unlike normal baryonic matter which form structures .
the combined analysis of cosmological observations suggests that the universe is spatially flat , and consists of about 70% dark energy , 30% dust matter ( cold dark matter plus baryons ) , and negligible radiation .
the nature of dark energy as well as its cosmological origin remain enigmatic at present .
the future of the universe crucially depends on the nature of dark energy : if it is quintessence ( having equation of state parameter eos @xmath0 ) then the energy density of quintessence dilutes with the expansion and the acceleration will be replaced by deceleration in far future ; if the cause of cosmic acceleration is cosmological constant ( @xmath1 ) then the universe will accelerate forever since its constant energy density provides a continuous source of vacuum energy to produce acceleration ; however if the dark energy is phantom energy ( @xmath2 ) then acceleration of the universe will convert into super - acceleration in far future which will eventually destroy every gravitationally stable structure in the universe @xcite . in recent years , the thermodynamics of the accelerating universe has got much attention and numerous interesting results are obtained . in particular , the generalized second law ( gsl ) of thermodynamics has been widely studied in the cosmological context ; the law states that the entropy of a closed isolated system along with the entropy of its boundary is always an increasing function of time .
it needs to be stressed that this law has some informal proofs but on several instances in cosmology , it is violated @xcite .
furthermore , the validity of gsl crucially depends on the choice of boundary of the frw universe : for instance the gsl is respected if the frw boundary is the dynamical apparent horizon @xcite but it is conditionally valid if the boundary is future event horizon @xcite . in this work ,
we choose the boundary as the future event horizon and show that the gsl is violated in all the cases studied in this paper .
the main motivation and the organization of this work are as follows : in section ii , we introduce basic equations and solutions for dbi - essence model .
two particular solutions are found and the scalar field and corresponding potentials are analyzed . in section iii , we calculate statefinder parameters for dbi dark energy and its nature are investigated during evolution of the universe . in section iv , we develop correspondence between dbi - essence and modified chaplygin gas and reconstruct the potential and warped brane tension as well as dynamics of scalar field are analyzed for two types of solutions . in section
v , we study gsl in the presence of mcg and dbi - essence and examine the validity of gsl during evolution of the universe bounded by the event horizon .
final section is devoted to the discussion .
note that a simple scalar field , a ` quintessence field ' , is a suitable candidate as an alternative to the ` cosmological constant ' @xcite .
the dynamics of the scalar fields depend on the scalar potentials . however , scalar fields with inverse power - law potentials have attracted lot of research interests since in this case the equations of motion yield attractor solutions .
it ensures that the late time behavior of the universe is independent to the choice of arbitrary initial conditions .
however , this exquisite behavior of the quintessence field comes at a price of extreme fine - tuning of the cosmological parameters .
this problem of fine - tuning can be resolved if the quintessence field is modeled via approaches beyond the standard model of particle physics , for instance , string theory @xcite . in this paper , we proceed with a scalar field model where the kinetic term is non - canonical .
such non - canonical terms in the lagrangian generally appear in the braneworld gravity @xcite . here
, the kinetic term has a dirac - born - infeld ( dbi ) form .
physically , this originates from the fact that the action of the system is proportional to the volume traced out by the brane during its motion .
this volume is given by the square - root of the induced metric which automatically leads to a dbi kinetic term @xcite .
the action of the dirac - born - infeld ( dbi ) scalar field @xmath3 can be written as ( choosing @xmath4 ) @xcite @xmath5,\ ] ] where @xmath6 is the self - interacting potential and @xmath7 is the warped brane tension . in the later analysis , we shall determine exact forms of these two functions .
now let us consider the matter content of the universe is composed of dbi type dark energy scalar field .
the background spacetime is the spatially flat friedmann - robertson - walker ( frw ) , for which the einstein field equations are @xmath8 @xmath9 where @xmath10 is the hubble parameter .
notice that @xmath11 produces super - acceleration while @xmath12 corresponds to accelerated expansion due to cosmological constant .
the energy density and pressure of the scalar field are respectively given by @xmath13 @xmath14 where the quantity @xmath15 is reminiscent from the usual relativistic lorentz factor and is given by @xmath16 from above expression ( 6 ) , we observe that @xmath17 and @xmath18 always . from ( 4 ) and ( 5 ) we have @xmath19 which is always @xmath20 . from energy conservation equation
, we have the wave equation for @xmath3 as @xmath21=0.\ ] ] now we consider two cases : ( i ) @xmath22 constant and ( ii ) @xmath23 constant . * case i : * @xmath22 constant . in this case , assume @xmath24 , ( @xmath25 ) and @xmath26 , ( @xmath27 is a positive constant ) so that @xmath28 . in these choices
we have the following solutions : @xmath29 @xmath30 @xmath31 and @xmath32 where , @xmath33,\nonumber\\ a_{0}&=&\big(\frac{3\gamma c\sqrt{(\gamma-1)n+m}}{\sqrt{3}[2(m - n)+(2n-1)\gamma^{3}]}\big)^{\beta},\nonumber\end{aligned}\ ] ] and @xmath34 with @xmath35 and @xmath36 are constants . from ( 10 ) and ( 11 ) , we see that @xmath7 and @xmath6 always decrease as @xmath3 increases . for this solution ,
the deceleration parameter @xmath37 becomes , @xmath38 for acceleration of the universe , @xmath37 must be negative i.e. @xmath39 and hence @xmath40 .
* case ii * : @xmath23 constant .
let us assume , @xmath41 , so from ( 6 ) we have @xmath42 .
since @xmath43 , so that we have @xmath44 .
let us also assume @xmath45 .
in this case , we have the solutions : @xmath46 @xmath47 where @xmath48 is the integration constant and the expression for deceleration parameter @xmath37 as @xmath49 from ( 13 ) and ( 14 ) , we see that the solution is valid for @xmath50 . for acceleration of the universe , @xmath37 must be negative i.e , @xmath51 .
since there are various candidates for the dark energy model , we often face with the problem of discriminating between them , which were solved by introducing statefinder parameters @xcite . these statefinder diagnostic pair i.e. , @xmath52 parameters are of the form : @xmath53 these parameters are dimensionless and allow us to characterize the properties of dark energy in a model independent manner .
the statefinder is dimensionless and is constructed from the scale factor of the universe and its time derivatives only .
the parameter @xmath54 forms the next step in the hierarchy of geometrical cosmological parameters after @xmath55 and @xmath37 . for cosmological constant with a fixed equation of state ( @xmath1 ) and a fixed newton s gravitational constant
, we have @xmath56 .
moreover @xmath57 represents the standard cold dark matter model containing no radiation while einstein static universe corresponds to @xmath58 @xcite . in literature ,
the diagnostic pair is analyzed for various dark energy candidates including holographic dark energy @xcite , agegraphic dark energy @xcite , quintessence @xcite , dilaton dark energy @xcite , yang - mills dark energy @xcite , viscous dark energy @xcite , interacting dark energy @xcite , tachyon @xcite , modified chaplygin gas @xcite and @xmath59 gravity @xcite to name a few . in case
i , we have the expressions of @xmath54 and @xmath60 as @xmath61 and @xmath62 , which are constants . in case
ii , we have found the relation between density and pressure as @xmath63 and the relation between @xmath54 and @xmath60 as @xmath64 the behavior of the parameters @xmath65 in ( 19 ) is shown in fig.1 . from the figure ,
we have seen that @xmath60 decreases from some negative value to @xmath66 as @xmath54 increases upto a certain stage but they obey negative sign .
after that @xmath60 also decreases from @xmath67 to some negative value as @xmath54 increases from negative label to positive label during evolution of the universe .
for the dbi model in case ii.,title="fig : " ] +
here , it is interesting to find the possible relation between the dbi - essence and the modified chaplygin gas ( mcg ) @xcite .
the mcg best fits with the @xmath68year wmap and the sdss data with the choice of parameters @xmath69 and @xmath70 @xcite which are improved constraints than the previous ones @xmath71 @xcite .
recently it is shown that the dynamical attractor for the mcg exists at @xmath72 , hence mcg crosses this value from either side @xmath73 or @xmath74 , independent to the choice of model parameters @xcite .
a generalization of mcg is suggested in @xcite by considering @xmath75 , where @xmath76 and @xmath77 are constants .
the mcg is the generalization of generalized chaplygin gas @xmath78 @xcite with the addition of a barotropic term .
this special form also appears to be consistent with the wmap @xmath79year data and henceforth the support the unified model with dark energy and matter based on generalized chaplygin gas @xcite . in the cosmological context , the chaplygin gas was first suggested as an alternative to quintessence and demonstrated an increasing @xmath80 behavior for the evolution of the universe @xcite .
recent supernovae data also favors the two - fluid cosmological model with chaplygin gas and matter @xcite .
recently , several works on chaplygin gas @xcite and other dark energy model like tachyonic field @xcite have been discussed for interacting and non - interacting scenarios of the accelerating universe . in this section , we will show that , by choosing a proper potential , the dbi - essence can be described by a modified chaplygin gas at late times . to find the possible relation between the dbi - essence and the modified chaplygin gas
, we set @xmath81 from energy conservation equation , we have the solution of @xmath82 in modified chaplygin gas as @xmath83^{\frac{1}{1+\alpha}},\ ] ] where @xmath35 is an arbitrary positive integration constant . from equations ( 4 ) - ( 6 )
, we have @xmath84 @xmath85 now consider the following two cases : against @xmath86 respectively in case i for @xmath87.,title="fig : " ] + against @xmath86 respectively in case i for @xmath87.,title="fig : " ] + against @xmath86 respectively in case i for @xmath87.,title="fig : " ] + * case i * : @xmath22 constant . from equations ( 6 ) and ( 22 ) , it is easy to seen that the expression of @xmath88 is @xmath89 from ( 20 ) - ( 24 ) , we get the solutions @xmath90 @xmath91^{-\frac{\alpha}{1+\alpha}},\end{aligned}\ ] ] @xmath92\nonumber\\&&\times \big[c+\frac{b}{1+a}~a^{3(1+a)(1+\alpha ) } \big]^{-\frac{\alpha}{1+\alpha}}.\end{aligned}\ ] ] the variations of @xmath3 , @xmath7 and @xmath6 against @xmath86 in case i have been drawn in figs . 2 , 3 and 4 respectively for @xmath87 .
from these figures , we see that dbi scalar field @xmath3 and potential @xmath93 are increasing and warped brane tension @xmath94 is decreasing with the evolution of the universe . * case ii * : @xmath23 constant : let for simplicity @xmath95 . from ( 6 ) and ( 22 ) we have @xmath96 the solutions are @xmath97^{\frac{1}{2(1+\alpha)}},\end{aligned}\ ] ] @xmath98^{\frac{2}{1+\alpha}}\nonumber\\&&\times\big(\big[c+\frac{b}{1+a}~a^{3(1+a)(1+\alpha ) } \big]^{2}\nonumber\\&&-c^{2}(1+a)^{2}\big[c+\frac{b}{1+a}~a^{3(1+a)(1+\alpha ) } \big]^{\frac{2}{1+\alpha}}\big)^{-1},\end{aligned}\ ] ] @xmath99^{\frac{1}{1+\alpha } } } { c(1+a)a^{3(1+a)}+\big[c+\frac{b}{1+a}~a^{3(1+a)(1+\alpha ) } \big]^{\frac{\alpha}{1+\alpha } } } \nonumber\\&&-\frac{a^{3(1+a)}\big[ac-\frac{b}{1+a}~a^{3(1+a)(1+\alpha ) } \big]}{c(1+a)a^{3(1+a)}+\big[c+\frac{b}{1+a}~a^{3(1+a)(1+\alpha ) } \big]^{\frac{\alpha}{1+\alpha } } } .\end{aligned}\ ] ] the variations of @xmath3 , @xmath7 and @xmath6 against @xmath86 in case ii have been drawn in figs . 5 , 6 and 7 respectively for @xmath100 .
from these figures , we see that dbi scalar field @xmath3 and potential @xmath93 are increasing and warped brane tension @xmath94 is decreasing with the evolution of the universe .
so the dbi scalar field , potential and warped brane tension can be reconstructed by modified chaplygin gas model . against @xmath86 respectively in case ii for @xmath100.,title="fig : " ] + against @xmath86 respectively in case ii for @xmath100.,title="fig : " ] + against @xmath86 respectively in case ii for @xmath100.,title="fig : " ] +
gibbons & hawking conjectured that event horizon area , including cosmological event horizons , might quite generally have associated entropy @xcite .
a prominent example is de sitter space .
we consider the frw universe as a thermodynamical system with the future even horizon surface as a boundary of the system , which is a valid assumption @xcite . this horizon has got recent attention since it yields a correct equation of state of dark energy , namely for the holographic dark energy @xcite . in general ,
the radius of the event horizon @xmath101 is not constant but changes with time ( or expansion of the universe ) .
let @xmath102 be an infinitesimal change in the radius of the future event horizon during a time of interval @xmath103 .
this small displacement @xmath102 will produce an infinitesimal change @xmath104 in the volume @xmath93 of the event horizon .
each spacetime describing a thermodynamical system and satisfying einstein s equations differs infinitesimally in the extensive variables volume , energy and entropy by @xmath104 , @xmath105 and @xmath106 , respectively , while having the same values for the intensive variables temperature @xmath94 and pressure @xmath107 .
thus , for these two spacetimes describing two thermodynamical states , there must exist some relation among these thermodynamic quantities .
to study the generalized second law of thermodynamics through the universe we deduce the expression for normal entropy using the first law of thermodynamics @xmath108 where , @xmath109 and @xmath110 are respectively temperature , entropy , pressure , volume and internal energy within the event horizon . here the expression for internal energy can be written as @xmath111 . now
the volume of the sphere is @xmath112 , where @xmath113 is the radius of the event horizon defined by @xmath114 which immediately gives @xmath115 the temperature of the event horizon is @xcite @xmath116 so using the above relations ( 33)-(35 ) , equation ( 32 ) can be written as @xmath117 also the entropy on the event horizon is @xcite @xmath118 using ( 34 ) , ( 36 ) and ( 37 ) , we have the rate of change of total entropy as @xmath119 the generalized second law states that total entropy can not be decrease i.e. , @xmath120 now we shall examine the validity of gsl of thermodynamics for dbi - essence and modified chaplygin gas separately .
+ @xmath121 * dbi - essence : * + * case i : * @xmath122 constant : for the dbi solution ( 8) , the radius of the event horizon is @xmath123 in this case , the rate of change of total entropy becomes @xmath124 so from ( 41 ) we see that , generalized second law can not be satisfied in case i for dbi - essence model .
+ against @xmath86 in dbi model in case ii.,title="fig : " ] + * case ii : * @xmath125 constant : for the dbi solutions ( 13 ) and ( 14 ) and using ( 2)-(6 ) , ( 33 ) and ( 34 ) , we get the radius of the event horizon as @xmath126 where @xmath127 represents the complementary error function .
the rate of change of total entropy becomes @xmath128\end{aligned}\ ] ] the above expression is very complicated form in @xmath86 .
so we have drawn the figure of @xmath129 against @xmath86 in fig .
8 . from the figure
, we see that @xmath130 for all values of @xmath86 .
hence we conclude that gsl can not be satisfied during evolution of the universe in case ii of dbi - essence model . against @xmath131 in modified chaplygin gas model for @xmath132.,title="fig : " ] + @xmath121 * modified chaplygin gas : * + for chaplygin gas model , let us assume @xmath133 , so that the solution for density ( 21 ) reduces to @xmath134 using equations ( 2 ) , ( 33 ) and ( 44 ) , the radius of the event horizon can be expressed as @xmath135\big],\end{aligned}\ ] ] where @xmath136\big]_{x=\infty}.\end{aligned}\ ] ] from equation ( 38 ) , we have the deviation of total entropy as @xmath137\big]\nonumber\\ & & \times \big [ x^{\frac{1 + 3a}{6(1+a)(1+\alpha)}}\big(\frac{2}{1+a}-\frac{3}{1+x } \big)\nonumber\\&&\times_{2}f_{1}[\frac{1 + 3a}{6(1+a)(1+\alpha)},\frac{1}{2(1+\alpha)},\nonumber\\ & & 1+\frac{1 + 3a}{6(1+a)(1+\alpha ) } , -x]\nonumber\\ & & + \big(\frac{1}{1+a}+\frac{3}{1+x } \big)d\nonumber\\ & & -\frac{1 + 3a}{1+a } x^{\frac{1 + 3a}{6(1+a)(1+\alpha ) } } ( 1+x)^{-\frac{1}{2(1+\alpha ) } } \big].\end{aligned}\ ] ] the above expression is very complicated form in @xmath131 .
so we have drawn the figure of @xmath138 against @xmath131 in fig .
9 . from the figure
, we see that @xmath139 for all values of @xmath131 .
hence we conclude that gsl always satisfied during evolution of the universe in modified chaplygin gas model .
in this work , we have considered the frw universe with dbi - essence dark energy model , which is a scalar field having a non - canonical kinetic term .
we have found the exact solutions of potential , warped brane tension and dbi scalar field .
we also calculate the statefinder parameters for our model that make it distinguishable among numerous dark energy models . from the fig . 1 , we have seen that @xmath60 decreases from some negative value to @xmath66 as @xmath54 increases up to a certain stage but they obey negative sign . after that @xmath60 also decreases from @xmath67 to some negative value as @xmath54 increases from negative label to positive label during evolution of the universe .
moreover , we establish correspondence between dbi - essence and modified chaplygin gas and hence we have reconstructed the potential and warped brane tension in cases i and ii . by this reconstruction and from figs . 2 - 7 ,
we have seen that dbi scalar field and potential increase and warped brane tension decreases during evolution of the universe .
we have also considered total entropy as sum of the entropies of a cosmological event horizon and the entropy of the dbi - essence .
we have investigated the validity of the generalized second law ( gsl ) of thermodynamics in the presence of dbi - essence and modified chaplygin gas separately . in all of the cases ( cases i and ii ) of dbi - essence model ,
we have observed , the time derivative of the total entropy is remaining at the negative level during the evolution .
this means that the total entropy is a decreasing function of time in the situations considered in this work .
so the gsl breaks down for dbi - essence model in both the cases .
it is also observed that the gsl always satisfied for mcg model .
riess a. et al . , 1998 astron .
j. , 116 , 1009 tegmark m. et al .
, 2004 phys . rev .
d 69 , 103501 bennett m. et al .
, 2003 astrophys .
148 , 1 copeland e.j . ,
sami m. , & tsujikawa t. , 2006 int . j. mod .
phys d. , 15 , 1753 zhang x. , & wu f - q .
, 2007 phys .
d 76 , 023502 wall a.c . , 2009 jhep 0906 , 021 jamil m. , saridakis e.n . , &
setare m.r .
, 2010 phys . rev .
d 81 , 023007 sadjadi h.m . &
jamil m. , 2010 epl 92 , 69001 ratra b. & peebles p.j.e .
, 1988 phys . rev .
d 37 , 3406 brax s. , arxiv:0711.2428 [ hep - ph ] kachru s. et al . , 2003 jcap 0310 , 013 martin j. & yamaguchi m. , 2008 phys . rev .
d 77 , 123508 sahni v. et al . , 2003 jetp lett .
77 , 201 debnath u. , 2008 class .
25 , 205019 zhang x. , 2005 int . j. mod
d 14 , 1597 wei h. & cai r.g . ,
2007 phys . lett .
b 655 , 1 huang j.z et al . , 2008 astrophys .
space sci .
315 , 175 zhao w. , 2008 int . j. mod .
d 17 , 1245 zimdahl w. & pavon d. , 2004 gen . relativ . grav .
36 , 1483 hu m. & meng x.h .
, 2006 phys .
b 635 , 186 shao y. & gui y. , 2008 mod . phys .
lett . a 23 , 65 chakraborty w. & debnath u. , 2007 mod . phys . lett . a 22 , 1805 setare m.r . & jamil m. , 2011 gen . rel .
43 , 293 benaoum h.b .
, hep - th/0205140 lu j. et al . , 2008 phys . lett .
b 662 , 87 dao - jun l. & l. xin - zhou . , 2005 chin
22 , 1600 jing h. et al . ,
25 , 347 debnath u. , 2007 astrophys .
space sci .
312 , 295 barreiro t. & sen a. a. , 2004 phys . rev .
d 70 , 124013 carturan d. & finelli f. , 2003 phys . rev .
d 68 , 103501 barreiro et al , 2008 , phys . rev .
d 78 , 043530 makler m. et al . , 2003 phys .
b 555 , 1 kamenshchik a. et al 2001 phys . lett .
b 511 , 265 panotopoulos g. 2008 phys .
d 77 , 107303 setare m.r . , 2007 phys .
b 648 , 329 setare m.r . ,
2007 phys .
b 654 , 1 setare m.r . , 2007 eur . phys .
j. c 52 , 689 setare m.r .
, 2007 phys .
b 653 , 116 setare m.r . , 2009 phys . lett .
b 673 , 241 gibbons g.w . & hawking s.w . , 1977 phys .
d 15 , 2738 davis t.m .
et al . , 2003 class .
20 , 2753 jamil m. et al .
, 2009 phys .
b 679 , 172 | in this work , we have considered the dbi - essence dark energy model in frw universe . we have found the exact solutions of potential , warped brane tension and dbi scalar field .
we also calculate the statefinder parameters for our model that make it distinguishable among numerous dark energy models .
moreover , we establish correspondence between dbi - essence and modified chaplygin gas ( mcg ) and hence reconstruct the potential and warped brane tension . by this reconstruction
, we observe that dbi scalar field and potential increase and warped brane tension decreases during evolution of the universe .
finally , we investigate the validity of the generalized second law ( gsl ) of thermodynamics in the presence of dbi - essence and modified chaplygin gas .
it is observed that the gsl breaks down for dbi - essence model but gsl always satisfied for mcg model . |
galactic archaeology is the detailed study of nearby galaxies that aims to reconstruct their evolutionary histories .
a key , but often elusive , goal is to understand the assembly of an individual galaxy in terms of its initial formation and subsequent mass growth via mergers / accretion of stars and gas .
the formation of massive galaxies ( which are predominately early - type ) is often described in two stages or phases . in the first phase
, it is thought that gas collapses via strong radial inflows , into a turbulent , clumpy rotating disc ( keres et al .
2005 ; dekel et al . 2009 ; ceverino et al . 2010 ; zolotov et al .
the vast bulk of the stars form in - situ within the central galaxy regions . in the second phase ,
mass growth occurs via stars that are formed ex - situ , i.e. within other galaxies that are later accreted / merge onto the host galaxy ( naab et al . 2007 ; oser et al . 2010 , 2012 ; lackner et al . 2012
; dubois et al . 2013 ) . in this picture , accretion may dominate the mass growth at low redshift , with more massive galaxies accreting a larger fraction of their mass ( khochfar et al .
2011 ) .
recently , naab et al . (
2014 ) presented detailed cosmological zoom simulations with hydrodynamics of 44 massive galaxies formed in two - phases as described above , while also placing them in the context of previous galaxy formation modelling .
in particular , they modelled the shapes and kinematics of their present day simulated galaxies in two dimensions ( 2d ) with a spatial resolution of 0.4 kpc ( about 1/10 of the typical effective radius ) .
this included 2d kinematics of velocity , velocity dispersion ( @xmath1 ) and higher order moments h@xmath2 ( which indicates the skewness of the velocity distribution ) and h@xmath3 ( which indicates the kurtosis , or peakiness , of the velocity distribution ) similar to that provided by today s integral field spectroscopy observations .
different mass assembly histories leave distinct imprints in the 2d kinematic maps and 1d kinematic profiles of the final galaxy . from these imprints
6 distinct evolutionary pathways were identified .
these pathways involved different merger ratio contributions ( minor or major ) , timescales ( early or recent mergers ) and gas contents ( rich or poor ) . although only a small sample , these represent possible pathways for the growth of massive galaxies in a cosmological context . using the same suite of models as naab et al .
( 2014 ) , wu et al .
( 2014 ) showed that these different assembly pathways were also reflected in the outer kinematic radial profiles of present day simulated galaxies .
the different assembly histories are also naturally associated with different mean fractions of stars formed in - situ and ex - situ ( i.e. accreted ) . in particular , the fraction of stars formed ex - situ is predicted to correlate with galaxy stellar mass and mean stellar age ( naab et al .
2014 ) . using similar initial conditions to the naab et al .
( 2014 ) models , remus et al .
( 2013 ) and hirschmann et al .
( 2015 ) showed that total mass density profiles and stellar metallicity outer gradients respectively , also correlate with the ex - situ fraction . using a semi analytic model approach , khochfar et al .
( 2011 ) focused on the assembly history of centrally fast and slow rotators in a similar cosmology to naab et al .
they concluded that slow rotators accreted over 50 - 90% of their mass in 1 - 2 major mergers and a series of minor mergers .
for fast rotators the accreted fraction is less than 50% and is dominated by minor mergers .
this picture agrees in broad strokes with that of naab et al .
( 2014 ) , although the hydrodynamic simulations suggest that a fraction of slow rotators may have experienced only minor mergers .
moody et al . ( 2014 ) also focused on reproducing fast and slow rotators . in their controlled binary and multiple merger simulations
, they found that binary mergers almost always result in a fast rotator while round , slow rotators as observed in nature require multiple mergers .
the first large 2d kinematic survey of early - type galaxies was the atlas@xmath4 survey ( cappellari et al .
they found a variety of kinematic structures , including the prevalence of embedded kinematic discs .
based on this , they classified 86% of their 260 early - type galaxies to be centrally fast or regular rotators ( i.e. disc - like kinematics with high angular momentum in the inner regions ) and 12% to be slow rotators with low angular momentum . from the sluggs ( sages legacy unifying globulars and galaxies , http://sluggs.swin.edu.au ) survey , arnold et al .
( 2014 ) and foster et al .
( 2015 ) found that centrally slow rotators maintain their slow rotation at larger radii , but centrally fast rotators show significant diversity .
their rotation may plateau , decline or steepen with increasing radius .
similar results were obtained from observations of pne ( coccato et al . 2009 ; cortesi et al . 2013 ) . as well as sluggs , a number of other 2d integral field spectroscopy surveys are underway or near completion e.g. califa , massive , sami , manga ( see brodie et al .
2014 for a summary ) .
each has its own comparative advantages .
a number of previous studies have compared detailed observations with tailored models to infer the past history of individual galaxies .
however , we are aware of only one independent study ( spiniello et al . 2015 ) that has directly compared their own 2d results with the simulations of naab et al .
( 2014 ) and thus assigned an assembly class within the two - phase model of galaxy formation .
the spiniello et al .
( 2015 ) data reached out to 0.7 effective radii ( r@xmath0 ) in ngc 4697 and confirmed the disc - like kinematics found by atlas@xmath4 and the sluggs surveys ( krajnovic et al . 2011 ; arnold et al .
they found its kinematic properties to be most consistent with class a , i.e. having undergone gas - rich minor mergers with gradual dissipation ( however a recent gas - rich major merger may also be a possibility ) . from their stellar population analysis
, they determined that the disk formed some 3 gyr after the bulge .
here we use the high s / n , large radial extent and excellent velocity resolution of the 2d kinematics from the sluggs survey of nearby early - type galaxies ( brodie et al . 2014 ) to compare their stellar properties with those of the simulated galaxies of naab et al .
( 2014 ) . combining with the central ( @xmath51 r@xmath0 ) kinematics from the atlas@xmath4 survey where available ,
we classify our galaxies into one of the six classes of assembly history .
we then compare their v/@xmath1 profiles with average values from the simulated galaxies of wu et al .
( 2014 ) , and the average ex - situ mass fractions of different assembly classes with mean stellar age , mass density profile slope and stellar metallicity gradients .
finally , we discuss the relative frequency of the different assembly classes for the sluggs galaxies and contrast them with model predictions .
based on cosmological zoom simulations of 44 model galaxies , naab et al .
( 2014 ; hereafter n14 ) defined 6 ` assembly classes ' with distinct evolutionary pathways .
the galaxies build up their mass via mergers and accretions , at different times , with variations in gas content .
this gives rise to a range in the relative fraction of ex - situ formed stars to the total stellar mass .
the present day total stellar masses of their galaxies span 2.6 @xmath5 m@xmath6 @xmath5 57 @xmath7 10@xmath8 m@xmath9 , and lie within dark matter halos of mass 2.2 @xmath5 m@xmath6 @xmath5 370 @xmath7 10@xmath10 m@xmath9 .
the different pathways are encoded in the _
present day _ isophotal shapes and 2d kinematics of the galaxies .
we note that the resolution and feedback of the models are limited and refer the reader to n14 for a see a discussion .
below we briefly describe the key properties of each n14 assembly class .
we focus on 2d kinematics since the isophotal shape of galaxies does not provide much leverage between the different classes ( nor do the models cover the full range of observed ellipticities ) .
the 2d kinematic properties are based on the appearance of the 2d velocity and velocity dispersion maps out to 2 r@xmath0 , and the distribution of individual pixel values of the higher order moments h@xmath2 and h@xmath3 versus v/@xmath1 within 1 r@xmath0 .
a 1d property is also included by n14 , i.e. the shape of the @xmath11 profile ( a proxy for projected angular momentum ) to 2 r@xmath0 as defined by emsellem et al .
+ + notes : the inner metallicity gradients ( second column ) are measured in the radial region @xmath12 , while the outer metallicity gradients ( third column ) are measured in the radial region @xmath13 . in the cases where the 2d maps do not extend to @xmath14
, the radial profile is extrapolated from the available points outside @xmath15 . | early - type ( e and s0 ) galaxies may have assembled via a variety of different evolutionary pathways . here
we investigate these pathways by comparing the stellar kinematic properties of 24 early - type galaxies from the sluggs survey with the hydrodynamical simulations of naab et al .
( 2014 ) .
in particular , we use the kinematics of starlight up to 4 effective radii ( r@xmath0 ) as diagnostics of galaxy inner and outer regions , and assign each galaxy to one of six naab et al . assembly classes .
the majority of our galaxies ( 14/24 ) have kinematic characteristics that indicate an assembly history dominated by gradual gas dissipation and accretion of many gas - rich minor mergers .
three galaxies , all s0s , indicate that they have experienced gas - rich major mergers in their more recent past .
one additional elliptical galaxy is tentatively associated with a gas - rich merger which results in a remnant galaxy with low angular momentum .
pathways dominated by gas - poor ( major or minor ) mergers dominate the mass growth of six galaxies .
most sluggs galaxies appear to have grown in mass ( and size ) via the accretion of stars and gas from minor mergers , with late major mergers playing a much smaller role .
we find that the fraction of accreted stars correlates with the stellar mean age and metallicity gradient , but not with the slope of the total mass density profile .
we briefly mention future observational and modelling approaches that will enhance our ability to accurately reconstruct the assembly histories of individual present day galaxies .
[ firstpage ] galaxies : evolution galaxies : kinematics and dynamics |
cataclysmic variables ( cvs ) are close binary systems transferring matter from a low - mass dwarf secondary to a white dwarf .
the transferred matter forms an accretion disk . in dwarf novae ( dne )
, a class of cvs , thermal - viscous instability in the accretion disk causes outbursts .
su uma - type dwarf novae , a subclass of dne , show long outbursts ( superoutburst ) in addition to ordinary short outbursts , and semi - periodic modulations called superhumps are detected during the superoutbursts .
the superhumps have periods a few percent longer than the orbital period .
it is now widely believed that the 3:1 resonance in the accretion disk causes the eccentric deformation of the disk , resulting superhumps [ tidal instability : @xcite ; @xcite ; @xcite ] .
the superoutburst can be understood as a result of the increased tidal dissipation and removal of the angular momentum when the tidal instability works [ thermal tidal instability ( tti ) model : @xcite ; @xcite ] .
this process produces a relaxation oscillation in the total angular momentum of the disk , and the superoutbursts recur .
the interval between the successive superoutbursts is called supercycle . [ for general information of cvs , dne , su uma - type dwarf novae and superhumps , see e.g. @xcite ] . in a series of papers @xcite , @xcite , @xcite @xcite and @xcite , we systematically surveyed su uma - type dwarf novae particularly on period variations of superhumps .
the change in the superhump period reflects the precession angular velocity of the eccentric ( or flexing ) disk , and is expected to be an excellent probe for studying the structure of the accretion disk during dwarf nova outbursts . in recent series of papers , we dealt with a various of topics related to su uma - type dwarf novae and superhumps : in @xcite , we also studied kepler data and made a pilot study of variations of superhumps amplitudes motivated by @xcite . in @xcite , we systematically studied er uma - type dwarf novae ( a class of su uma - type dwarf novae with very short supercycles , see e.g. @xcite ; @xcite ; @xcite ) and helium dwarf novae ( am cvn - type objects ) . in @xcite , we made a pilot study of the decline rate of the superoutburst motivated by @xcite and studied negative superhumps ( superhumps having periods shorter than the orbital period and are considered as a manifestation of a tilted disk , see e.g. @xcite ; @xcite ; @xcite ) particularly in bk lyn , which displays both states of a novalike variable ( a thermally stable cv ) and an er uma - type dwarf nova .
we continue this extended , comprehensive research of su uma - type dwarf novae and superhumps in general in this paper .
since most of the objects treated in this paper is little documented in the past , and since a compilation of historical descriptions of dwarf novae has not been issued for long since @xcite , we intend these series of papers to be also a source of compiled information of individual dwarf novae . the present advances in understanding of the superhump periods and their variations started with the new interpretation of the kepler observation @xcite , who used the negative superhumps as a probe for the variation of the disk radius over the supercycle , and confirmed the radius variation predicted by the tti model .
combined with @xcite , @xcite , the tti model is currently the only viable model to explain the su uma - type phenomenon . in @xcite , we demonstrated that most of the @xmath1 diagrams of superhumps in su uma - type dwarf novae can be expressed by three distinct stages : initial growing stage ( stage a ) wuth a long period and fully developed stage ( stage b ) with a systematically varying period and later stage c with a shorter , almost constant period .
[ see @xcite for the notation of stages a - b - c of superhumps ] .
the origin of these three stages of superhumps was a mystery when it was documented in @xcite .
after examination of the kepler data , @xcite proposed that the appearance of the pressure effect is responsible for the transition from stage a to stage b. in this interpretation , stage a reflects the state when the 3:1 resonance is confined to the resonance region .
this interpretation allowed a new method to determine the mass ratio ( @xmath2 ) only from superhump observations and the orbital period @xcite .
this method is particularly suitable for measuring mass ratios in wz sge - type dwarf novae ( su uma - type dwarf novae with very long supercycles , and are considered to be the terminal stage of the cv evolution ) and for discriminating hitherto very poorly known period bouncers ( cvs which have passed the minimum orbital period in evolution : @xcite ; @xcite ) .
superhumps are now not only a powerful tool for diagnosing the accretion disk , but also a powerful tool for illuminating the cv evolution . the material and methods of analysis are given in section [ sec : obs ] , observations and analysis of individual objects including short discussions on individual objects are given in section [ sec : individual ] , the general discussion is given in section [ sec : discuss ] and the summary is given in section [ sec : summary ] .
the data were obtained under campaigns led by the vsnet collaboration @xcite . for some objects
, we used the public data from the aavso international databasehttp://www.aavso.org / data - download@xmath3 . ] .
the majority of the data were acquired by time - resolved ccd photometry by using 30 cm - class telescopes located world - wide , whose observational details will be presented in future papers dealing with analysis and discussion on individual objects of interest .
the list of outbursts and observers is summarized in table [ tab : outobs ] .
the data analysis was performed just in the same way described in @xcite and @xcite and we mainly used r softwarehttp://cran.r - project.org/@xmath3 . ] for data analysis . in de - trending the data , we used both lower ( 15th order ) polynomial fitting and locally - weighted polynomial regression ( lowess : @xcite ) .
the times of superhumps maxima were determined by the template fitting method as described in @xcite .
the times of all observations are expressed in barycentric julian dates ( bjd ) .
the abbreviations used in this paper are the same as in @xcite : @xmath4 means the orbital period and @xmath5 for the fractional superhump excess . since @xcite ,
the alternative fractional superhump excess in the frequency unit @xmath6 has been introduced because this fractional superhump excess can be directly compared to the precession rate .
we therefore used @xmath7 in referring the precession rate .
we used phase dispersion minimization ( pdm ; @xcite ) for period analysis and 1@xmath8 errors for the pdm analysis was estimated by the methods of @xcite and @xcite .
we also used least absolute shrinkage and selection operator ( lasso ) method ( @xcite ; @xcite ) , which has been proven to yield very sharp signals . in this paper , we used the two - dimensional lasso power spectra as introduced in the analysis of the kepler data such as in @xcite ; @xcite ; @xcite .
these two - dimensional lasso power spectra have been proven to be helpful in detecting negative superhumps ( cf .
@xcite ) as well as superhumps with varying frequencies ( cf .
although the application of two - dimensional lasso power spectra to the kepler data dealt with almost uniformly sampled data , we have demonstrated in @xcite and @xcite that two - dimensional lasso power spectra are also effective in detecting multiple signals and their variations in non - uniformly sampled ground - based data .
the resultant @xmath9 , @xmath10 and other parameters are listed in table [ tab : perlist ] in same format as in @xcite .
the definitions of parameters @xmath11 and @xmath10 are the same as in @xcite .
we also presented comparisons of @xmath1 diagrams between different superoutbursts since this has been one of the motivations of these surveys ( cf .
@xcite ) and it has been demonstrated that a combination of @xmath1 diagrams between different superoutbursts can better describe the overall pattern of the period variation @xcite . in drawing combined @xmath1 diagrams , we usually used @xmath12 for the start of the superoutburst , which usually refers to the first positive detection of the outburst .
this epoch usually has an accuracy of @xmath131 d for well - observed objects , and if the outburst was not sufficiently observed , we mentioned how to estimated @xmath14 in such an outburst .
we also present representative @xmath1 diagrams and light curves especially for wz sge - type dwarf novae , which are not expected to undergo outbursts in the near future . in all figures ,
the binned magnitudes and @xmath1 values are accompnied by 1@xmath8 error bars , which are omitted when the error is smaller than the plot mark .
we used the same terminology of superhumps summarized in @xcite .
we especially call attention to the term `` late superhumps '' .
we only used `` traditional '' late superhumps when an @xmath130.5 phase shift is confirmed [ @xcite ; see also table 1 in @xcite for various types of superhumps ] , since we suspect that many of the past claims of detections of `` late superhumps '' were likely stage c superhumps [ cf .
@xcite ; note that the kepler observation of v585 lyr also demonstrated this persistent stage c superhumps without a phase shift @xcite ] .
early superhumps are double - wave humps seen during the early stages of wz sge - type dwarf novae , and have period close to the orbital periods ( @xcite ; @xcite ; @xcite ) .
we used the period of early superhumps as approximate orbital period .
the validity of this assumption is also reviewed in this paper .
as in @xcite , we have used coordinate - based optical transient ( ot ) designations for some objects , such as catalina real - time transient survey ( crts ; @xcite)http://nesssi.cacr.caltech.edu / catalina/@xmath3 . for the information of the individual catalina cvs ,
see @xmath15http://nesssi.cacr.caltech.edu / catalina / allcv.html@xmath3 . ]
transients and listed the original identifiers in table [ tab : outobs ] .
when available , we preferably used the international astronomical union ( iau)-format names provided by the crts team in the public data releasehttp://nesssi.cacr.caltech.edu / datarelease/@xmath3 . ] ccccl subsection & object & year & observers or references & i d + [ obj : foand ] & fo and & 2013 & aka & + [ obj : dhaql ] & dh aql & 2000 & oud , btw & + [ obj : bbari ] & bb ari & 2013 & gfb , okc , hac , mhh & + [ obj : uzboo ] & uz boo & 2013 & ku , gfb , dpv , rit , rui , mhh , & + & & & aavso , rpc , kai , cri , imi , & + & & & mdy , iak , ioh + [ obj : v342cam ] & v342 cam & 2013 & dpv & + [ obj : v452cas ] & v452 cas & 2013 & imi , rpc , dpv & + [ obj : v359cen ] & v359 cen & 2014 & hac & + [ obj : yzcnc ] & yz cnc & 2014 & aka , mdy , hac & + [ obj : gzcnc ] & gz cnc & 2014 & kis , mdy , imi , kai , aka & + [ obj : alcom ] & al com & 2013 & oku , nka , ku , irs , cri , & + & & & dks , kis , imi , aavso & + [ obj : v503cyg ] & v503 cyg & 2013 & rpc & + [ obj : ixdra ] & ix dra & 2012 & mev , aavso , ujh & + [ obj : mndra ] & mn dra & 2012 & ast , cri & + & & 2013 & cri , kai , ter , ku & + [ obj : cperi ] & cp eri & 2013 & okc , swi & + [ obj : v1239her ] & v1239 her & 2013 & rpc , imi & + [ obj : cthya ] & ct hya & 2014 & mdy & + [ obj : vwhyi ] & vw hyi & 2012 & hac , cta , hen & + [ obj : wxhyi ] & wx hyi & 1977 & @xcite & + & & 2014 & hac & + + + + [ cols="^,^,^,^,<",options="header " , ] ( 88mm,110mm)fig85.eps ( 88mm,110mm)fig86.eps ( 88mm,95mm)fig87.eps ( 88mm,95mm)fig88.eps
in addition to basic data of superhumps of the objects observed in this paper , the major findings we obtained can be summarized as follows .
* we report the detection of negative superhumps in quiescence of vw hyi in 2012 .
we conclude that type l supercycle in vw hyi ( reduced number of normal outbursts ) is associated with the presence of negative superhumps as in v1504 cyg and v344 lyr .
* master j005740 is the first eclipsing wz sge - type dwarf nova showing the probable eclipse of the white dwarf .
the sharp structure in the profile of the early superhumps is interpreted as the eclipse of the accretion disk , which has been difficult to distinguish from the profile of the early superhump itself in other wz sge - type dwarf novae .
the symmetric profile of the eclipse indicates that the disk itself , not the enhanced hot spot , is eclipsed .
this finding gives observational support to @xcite who interpreted that the source of the early superhumps is not the hot spot as suggested by an enhanced mass - transfer model @xcite .
we also provided a model calculation of the eclipse light curve of this object during the phase of early superhumps .
* we detected stage a superhumps with growing amplitudes in mn dra and likely stage a ( from the @xmath1 diagram ) in gz cnc , both of which have long orbital periods .
the stage a superhumps in these systems lasted longer than expected .
we interpreted that the 3:1 resonance was confined in the region of excitement because these objects have mass - ratios critically close to the condition in which the tidal instability occurs .
this may provide an interpretation of large negative period derivatives recorded in the past in systems with long orbital periods . * the 2013 superoutburst of uz boo was followed by four post - superoutburst rebrightening as was observed in the 2003 superoutburst .
this observation suggests that the pattern of rebrightening tends to be reproducible in the same object . *
the wz sge - type dwarf novae al com and asassn-13ck showed a long - lasting ( plateau - type ) rebrightening . in the early phase of the rebrightening
, both objects showed a precursor - like outburst , suggesting that the long - lasting rebrightening is triggered by a precursor outburst .
both objects showed small dip(s ) during the rebrightening .
* we have reviewed the observation of early superhumps of wz sge - type dwarf novae and found that the fractional superhump excesses for early superhumps have a typical value of @xmath16% .
* we have succeeded in detecting a positive period derivative of superhumps in the helium cv cp eri .
this object also showed oscillation - type rebrightenings .
* we have established the long - sought superoutburst of the eclipsing dwarf nova v893 sco after fifteen years since the rediscovery .
this work was supported by the grant - in - aid `` initiative for high - dimensional data - driven science through deepening of sparse modeling '' from the ministry of education , culture , sports , science and technology ( mext ) of japan .
the authors are grateful to observers of vsnet collaboration and vsolj observers who supplied vital data .
we acknowledge with thanks the variable star observations from the aavso international database contributed by observers worldwide and used in this research .
this work is deeply indebted to outburst detections and announcement by a number of variable star observers worldwide , including participants of cvnet and baa vss alert .
we thank dr .
brian skiff for making historical materials about wx hyi available to us .
the ccd operation of the bronberg observatory is partly sponsored by the center for backyard astrophysics .
the ccd operation by peter nelson is on loan from the aavso , funded by the curry foundation .
we are grateful to the catalina real - time transient survey team for making their real - time detection of transient objects available to the public . | continuing the project described by @xcite , we collected times of superhump maxima for 56 su uma - type dwarf novae mainly observed during the 20132014 season and characterized these objects .
we detected negative superhumps in vw hyi and indicated that the low number of normal outbursts in some supercycle can be interpreted as a result of the disk tilt .
this finding , combined with the kepler observation of v1504 cyg and v344 lyr , suggests that the disk tilt is responsible for modulating the outburst pattern in su uma - type dwarf novae .
we also studied the deeply eclipsing wz sge - type dwarf nova master ot [email protected] and found evidence of a sharp eclipse during the phase of early superhumps .
the profile can be reproduced by a combination of the eclipse of the axisymmetric disk and the uneclipsed light source of early superhumps .
this finding confirms the lack of evince of a greatly enhanced hot spot during the early stage of wz sge - type outburst .
we detected growing ( stage a ) superhumps in mn dra and give a suggestion that some of su uma - type dwarf novae situated near the critical condition of tidal instability may show long - lasting stage a superhumps .
the large negative period derivatives reported in such systems can be understood a result of the combination of stage a and b superhumps .
the wz sge - type dwarf novae al com and asassn-13ck showed a long - lasting ( plateau - type ) rebrightening . in the early phase of the rebrightening
, both objects showed a precursor - like outburst , suggesting that the long - lasting rebrightening is triggered by a precursor outburst . |
in the uncapacitated facility location ( ufl ) problem the goal is to open facilities in a subset of given locations and connect each client to an open facility so as to minimize the sum of opening costs and connection costs . in the penalty avoiding ( prize collecting ) variant of the problem , a fixed penalty can be paid instead of connecting a client . in the @xmath0-level uncapacitated facility location problem with penalties ( @xmath0-level uflwp ) , we are given a set @xmath1 of clients and a set @xmath2 of facilities ( locations to potentially open a facility ) in a metric space .
facilities are of @xmath0 different types ( levels ) , e.g. , for @xmath3 one may think of these facilities as shops , warehouses and factories .
each set @xmath4 contains all facilities on level @xmath5 and the sets @xmath4 are pairwise disjoint . each client
@xmath6 can either be connected to precisely one facility at each of @xmath0 levels ( via a path ) , or be rejected in which case the penalty @xmath7 must be paid ( @xmath7 can be considered as the loss of profit ) . to be more precise , for a client @xmath6 to be connected , it must be connected with a path @xmath8 , where @xmath9 is an open facility on level @xmath5 .
the cost of connecting points @xmath10 , is the distance between @xmath11 and @xmath6 , denoted by @xmath12 .
the cost of opening facility @xmath11 is @xmath13 ( @xmath14 ) .
the goal is to minimize the sum of the total cost of opening facilities ( at all levels ) , the total connection cost and the total penalty cost . in the uniform version of the problem
all penalties are the same , i.e. , for any two clients @xmath15 we have @xmath16 .
if @xmath17 are big enough , @xmath0-level uflwp is the @xmath0-level ufl problem , for which krishnaswamy and sviridenko @xcite showed @xmath18-hardness of approximation for general @xmath0 and @xmath19-hardness for @xmath20 .
actually , even for @xmath21 guha and khuller @xcite showed that the approximation ratio is at least @xmath22 , unless @xmath23 . the current best known approximation ratio for this simplest case @xmath21 is @xmath24 by li @xcite .
for @xmath25-level ufl problem shmoys , tardos , and aardal @xcite gave the first constant factor approximation algorithm by extending the algorithm for @xmath26-level and obtaining an approximation ratio @xmath27 .
subsequently , aardal , chudak , and shmoys @xcite used randomized rounding to get the first algorithm for general @xmath0 , which had approximation ratio of @xmath28 .
ageev , ye and zhang @xcite gave a combinatorial @xmath29-approximation algorithm for general @xmath0 by reducing the @xmath0-level directly into @xmath26-level problem .
by recursive reduction , i.e. , reducing @xmath0-level to @xmath30 level , they obtained an improved @xmath31-approximation for @xmath20 and @xmath32 for @xmath3 .
later , this was improved by zhang @xcite , who combined the maximization version of @xmath26-level ufl problem and dual - fitting to get a @xmath33-approximation algorithm for @xmath20 , and a @xmath34-approximation for @xmath3 .
byrka and aardal @xcite improved the ratio for @xmath3 to @xmath35 .
for @xmath36 the ratio was recently improved by byrka and rybicki @xcite to @xmath37 for @xmath3 , @xmath38 for @xmath39 , and the ratio converges to 3 when @xmath40 .
ufl with penalties was first introduced by charikar et al .
@xcite , who gave a @xmath28-approximation algorithm based on a primal - dual method .
later , jain et al .
@xcite indicated that their greedy algorithm for ufl could be adapted to uflwp with the approximation ratio @xmath25 .
xu and xu @xcite proposed a @xmath41-approximation algorithm based on lp - rounding and a combinatorial @xmath42-approximation algorithm by combining local search with primal - dual .
later , geunes et al . @xcite presented an algorithmic framework which can extend any lp - based @xmath43-approximation algorithm for ufl to get an @xmath44-approximation algorithm for ufl with penalties .
as a result , they gave a @xmath45-approximation algorithm for this problem .
recently , li et al .
@xcite extended the lp - rounding algorithm by byrka and aardal @xcite and the analysis by li @xcite to uflwp to give the currently best @xmath46-approximation algorithm . for multi - level uflwp , asadi et al .
@xcite presented an lp - rounding based @xmath47-approximation algorithm by converting the lp - based algorithm for uflwp by xu and xu @xcite to @xmath0-level . to the best of our knowledge ,
this is the only algorithm for multi - level uflwp in the literature .
we first show that algorithms whose performance can be analysed with a linear function of certain instance parameters , like the chudak and shmoys algorithm @xcite for ufl , can be easily combined and analysed with a natural factor revealing lp .
this simplifies the argument of shi li @xcite for his @xmath24-approximation algorithm for ufl _ since an explicit distribution for the parameters obtained by a linear program is not necessary in our factor revealing lp_. with this tool one can easily randomize the scaling factor in lp - rounding algorithms for various variants of the ufl problem .
_ we demonstrate this by randomizing the algorithm for @xmath0-level uflwp . for @xmath0-level ufl
we can get the same approximation ratios as for @xmath0-level uflwp by setting @xmath48 .
_ note that the previously best ratio is 4 for @xmath0-level uflwp ( @xmath49 ) @xcite and @xmath46 for @xmath21 @xcite .
the following table shows how much we improve the approximation ratios of our algorithm for @xmath50 by involving randomization of the scaling factor .
_ irrespective of the way in which we choose @xmath51 , deterministically or randomly , approximation ratio converges to three . _
.comparison of ratios . [ cols="^,^,^,^,^,^,^,^,^,^,^",options="header " , ] note that the reduction above does not work for the non - uniform case , because then the distance from client @xmath6 to the penalty - facility of client @xmath52 could be smaller than @xmath7 .
nevertheless we will show that lp - rounding algorithms in this paper can be easily extended to the non - uniform penalty variant .
_ for non - uniform case , _ our algorithm is based on rounding a solution to the extended lp - relaxation of the problem .
this extended lp may either be seen as the standard lp on a modified graph ( see appendix [ sec : modiffication ] ) as described in @xcite , or originate from the @xmath0-th level of the sherali adams hierarchy , or explicitly be written in terms of paths on the original instance .
here we use the explicit construction . note that in the optimal solution to @xmath0-level uflwp each facility
is connected to at most one facility on the higher level .
we will impose this structure on the fractional solution by creating multiple copies of the original facility , one for each path across the higher levels of facilities . to describe the linear program we have to give a few definitions .
let @xmath53 be the set of paths which start in a client and end in a facility on level @xmath0 .
let @xmath54 be the set of paths which start on level @xmath5 and end on the highest level @xmath0 , i.e. , in a root of some tree .
by @xmath55 we denote the set of all paths , i.e. , @xmath56 .
the cost of the path denoted by @xmath57 depends on the kind of path .
if @xmath58 , then @xmath59 . if @xmath60 , then @xmath61 .
@xmath62 the natural interpretation of the above lp is as follows .
inequality ( [ lp : out_one ] ) states that each client is assigned to at least one path or is rejected
. inequality ( [ lp : thomas_order ] ) encodes that opening a lower level facility implies opening its unique higher level facility . the most complicated inequality ( [ lp : f_open_enough ] ) for a client @xmath63 and a facility @xmath64 , imposes that the opening of @xmath9 must be at least the total usage of it by the client @xmath6 .
let @xmath65 be an optimal solution to the above lp .
the approximation algorithm @xmath66 presented below is parameterized by @xmath67 . formulate and solve the extended lp ( 12)-(17 ) to get an optimal solution @xmath65 ; scale up facility opening and client rejecting variables by @xmath67 , then recompute values of @xmath68 for @xmath69 to obtain a minimum cost solution @xmath70 + divide clients into two groups @xmath71 and @xmath72 cluster clients in @xmath73 ; round facility opening ( tree by tree ) ; connect each client @xmath6 with a closest open connection path unless rejecting it is a cheaper option .
our final algorithm is as follows : run algorithm @xmath74 for each @xmath75 and select a solution with the smallest cost .
clustering is based on rules described in @xcite which is generalized in @xcite for @xmath0-level instances . rounding on a tree
was also used in @xcite .
nevertheless , for completeness we give a brief description of step 4 and 5 in the following subsections . from now on we are considering only scaled up instance @xmath76 . for any client @xmath77 ,
let @xmath78 be the set of top - level facilities fractionally serving @xmath6 in @xmath76 . as discussed in section [ sec : one_level ] , wlog the fractional connectivity of @xmath6 to a set of facilities may be assumed to be the fractional opening of these facilities .
sort facilities @xmath79 from @xmath78 by non - decreasing distance from client @xmath77 , and select the smallest subset of @xmath78 with volume one - this is the set of close facilities @xmath80 , the rest of facilities from @xmath78 are distant facilities @xmath81 . by @xmath82 and @xmath83
we denote the average distances from @xmath6 to close , distant and all facilities in set @xmath78 respectively .
moreover by @xmath84 we denote the maximal distance from @xmath6 to a close facility .
formal definitions are as follows : @xmath85 using the similar arguments as in @xcite we can define @xmath86 and express @xmath87 and @xmath88 using @xmath89 .
@xmath90 two clients @xmath91 are called neighbors if @xmath92 .
select unclustered client @xmath77 that minimizes @xmath93 , form a new cluster containing @xmath6 and all its unclustered neighbors from @xmath94 , call @xmath6 the center of the new cluster ; the above clustering procedure ( just like in @xcite ) partitions all clients into groups called clusters .
such partition has two important properties .
first : there are no two neighbors from @xmath94 which are ( both ) centers of clusters .
second : distance from any client in cluster to his cluster center is not too big .
consider an arbitrary cluster center @xmath6 .
since lp solutions have a form of a forest , we only need to focus on rounding single tree serving @xmath6 .
for clarity , within this rounding procedure we will refer to facilities as vertices ( of a tree ) , and use @xmath95 to denote the fractional opening of vertex ( facility ) @xmath96 and @xmath97 to denote the extent in which the cluster center @xmath6 uses @xmath96 in @xmath76 , i.e , @xmath98 .
note that @xmath99 for each @xmath96 and @xmath100 if @xmath96 is not the root of a tree . the main idea is to open exactly one path for cluster center @xmath6 but keep the probability of opening of each vertex @xmath96 equal to @xmath95 in the randomized procedure .
in @xcite we gave a token - passing - based adaptation of the procedure by garg konjevod and ravi @xcite , that stores the output in @xmath101 and @xmath102 , and has exactly the desired properties .
@xmath103 = x_v$ ] and @xmath104 = y_v$ ] for all @xmath105 .
it is essential that the probability of opening at least one path in a set @xmath106 can be lower bounded by a certain function @xmath107 , where @xmath108 is the total flow from client @xmath6 to all paths in @xmath109 and @xmath0 is the number of levels in the considered instance .
it can be shown that @xmath110 and the following lemma ( from @xcite ) hold . for more details see appendix [ sec : functions of probability ] and @xcite .
inequality @xmath111 implies @xmath112 .
_ the high level idea is that we can consider the instance of @xmath0-level uflwp as a corresponding instance of @xmath0-level ufl by showing that the worst case approximation ratio is for clients in set @xmath94 and we can treat the penalty of client @xmath113 as a penalty - facility " in our analysis .
that is , we can overcome penalties by solving an equivalent @xmath0-level ufl without penalties . _
it is standard in uncapacitated location problems to split facilities to obtain a so called _ complete _ solution , where no facility is used less than it is open by a client ( see @xcite for details ) . for our algorithm , to keep the forest structure of the fractional solution , we must slice the whole trees instead of splitting individual facilities to obtain the following .
[ completeness ] each solution of our linear program for @xmath0-level uflwp can be transformed to an equivalent complete solution .
we should give two copies @xmath114 and @xmath115 of tree @xmath116 ( instead of it ) if there is some client @xmath63 with a positive flow @xmath117 to one of the paths @xmath118 in the tree @xmath116 which is smaller than the path opening @xmath119 .
let the opening of such problematic path be equal to flow @xmath117 in tree @xmath114 . in tree @xmath120 it has value equal to the opening in @xmath116 decreased by @xmath117 .
in general each facility in tree @xmath114
( @xmath120 ) has the same opening as in @xmath116 times @xmath121 ( @xmath122 ) .
note that the value of flow from client @xmath6 ( and other clients which are connected with both trees now ) should be the same as before adding trees @xmath114 and @xmath120 instead of @xmath116 .
all clients recompute " their connection values .
we sort all paths in increasing connection cost for client @xmath6 and connect with them ( in that order ) as strong as it is possible until client @xmath6 has flow equal to one or it is cheaper to pay penalty instead of connecting with any open path .
the important fact is that the expected connection and penalty cost of each client remain the same after above operations . in the process of coping and replacing trees we add at most @xmath123 new trees . because each client has at most one `` problematic '' ( not saturating ) path . for the clarity of the following analysis we will use a one - level " description of the instance and fractional solution despite its @xmath0-level structure . because the number of levels will have influence only on the probabilities of opening particular paths in our algorithm .
consider set @xmath124 of paths which start in client @xmath6 and end in the root of a single tree @xmath116 . instead of thinking about all paths from set @xmath124
separately we can now treat them as one path @xmath125 whose fractional opening is @xmath126 and ( expected ) cost is @xmath127 . observe that our distance function @xmath128 satisfy the triangle inequality .
from now on we will think only about clients and facilities ( on level @xmath0 ) and ( unique ) paths between them .
accordingly , we will now encode the fractional solution as @xmath129 , to denote the fractional connectivity , opening and penalty components .
[ ineq_proof ] @xmath130 .
@xmath131 @xmath132 @xmath133 + the second inequality holds because @xmath134 .
moreover to justify the last equality we should observe that @xmath135 .
[ worst_case ] the worst case approximation ratio is for clients from set @xmath94 .
we have two types of clients divided for two sets @xmath94 and @xmath136 .
lets sort facilities in nondecreasing distances from client @xmath6 . in that proof
@xmath137 is number of facilities which has positive flow from @xmath6 in considering ( scaled up ) fractional solution .
suppose the first case @xmath138 , then we can upper bound his connection and penalty cost like that @xmath139 \leq \sum_{i = 1}^{l}(f_k(\sum_{j = 1}^{i } y_j ) - f_k(\sum_{j = 1}^{i-1 } y_j ) ) d(q , j)\ ] ] @xmath140 @xmath141 @xmath142 inequality holds because @xmath143 in other case @xmath144 would be connected with facility in that distance instead of using penalty . in the second case we have
that @xmath145 .
connection and penalty cost of client @xmath144 can be upper bounded in below way @xmath139 \leq ( f_k(\sum_{j = 1}^{l } y_j ) - f_k(\sum_{j = 1}^{l-1 } y_j ) ) d(q , j ) + ( 1 - f_k(\gamma ( 1 - g ) ) ) p_q\ ] ] note that for each client @xmath146 the truth is @xmath147 , so from lemma [ ineq_proof ] we have that the worst case approximation ratio is for clients from set @xmath94 .
[ g_0 ] for clients @xmath77 we can treat its penalty as a facility . if @xmath6 is a cluster center , @xmath6 will have at least one ( real ) facility open in its set of close facilities .
thus , its connection and penalty cost are independent of the value of @xmath148 .
if @xmath6 is not a cluster center and we pretend its penalty as a facility , no other client @xmath52 will consider to use this fake facility . because @xmath52 only looks at facilities fractionally serving him , and the facilities which serve the center of the cluster containing @xmath52 .
a single algorithm @xmath149 has expected facility opening cost @xmath150\leq \gamma \cdot f^*$ ] and expected connection and penalty cost @xmath151 \leq max \{3 - 2 \cdot f_k(\gamma ) , \frac{2 - f_k(\gamma ) - f_k(1)}{1 - \frac{1}{\gamma}}\ } \cdot ( c^ * + p^*)$ ] ( see appendix [ sec : single_algorithm ] for a detailed proof ) . to obtain an improved approximation ratio we run algorithm @xmath66 for several values of @xmath51 and select the cheapest solution .
the following lp gives an upper bound on the approximation ratio .
@xmath152 since the number of levels has influence on connection probabilities , the values of @xmath153 need to be defined more carefully than for ufl .
in particular , for @xmath154 we now have @xmath155 and @xmath156 for @xmath157 .
the table [ improved_ratios ] summarizes the obtained ratios for a single algorithm ( run with the best choice of @xmath51 for particular @xmath0 ) and for a group of algorithms .
( i.e. , distances to facilities ) for @xmath158 obtained from solution of the lp in section [ group_analize ] . _
x - axis is volume of a considered set and y - axis represents distance to the farthest facility in that set .
values of function @xmath159 are in one - to - one correspondence with values of @xmath160 in lp from section [ group_analize]._,height=215 ] in a randomized alg .
( from the dual lp ) for @xmath158 . left figure : general view ; right figure : close - up on small probabilities.,width=226 ] in a randomized alg .
( from the dual lp ) for @xmath158 .
left figure : general view ; right figure : close - up on small probabilities.,width=226 ] 99 k. aardal , f. chudak , d. shmoys : a 3-approximation algorithm for the k - level uncapacitated facility location problem .
inf . process .
72(5 - 6 ) : 161 - 167 ( 1999 ) a. ageev , y. ye , j. zhang : improved combinatorial approximation algorithms for the k - level facility location problem .
icalp 2003 : 145 - 156 m. asadi , a. niknafs , m. ghodsi : an approximation algorithm for the k - level uncapacitated facility location problem with penalties .
csicc 2008 , ccis 6 : 41 - 49 . j. byrka , k. aardal : an optimal bifactor approximation algorithm for the metric uncapacitated facility location problem .
siam j. comput .
39(6 ) : 2212 - 2231 ( 2010 ) j. byrka , m. ghodsi , a. srinivasan : lp - rounding algorithms for facility - location problems .
corr abs/1007.3611 ( 2010 ) j. byrka , b. rybicki : improved lp - rounding approximation algorithm for k - level uncapacitated facility location .
icalp ( 1 ) 2012 : 157 - 169 m. charikar , s. khuller , d. mount , g. narasimhan : algorithms for facility location problems with outliers .
soda 2001 : 642 - 651 f. chudak , d. shmoys : improved approximation algorithms for the uncapacitated facility location problem .
siam j. comput .
33(1 ) : 1 - 25 ( 2003 ) n. garg , g. konjevod , r. ravi : a polylogarithmic approximation algorithm for the group steiner tree problem . soda 1998 : 253259 j. geunes , r. levi , h. romeijn , d. shmoys : approximation algorithms for supply chain planning and logistics problems with market choice . math . program .
130(1 ) : 85 - 106 ( 2011 ) s. guha , s. khuller : greedy strikes back : improved facility location algorithms .
j. algorithms 31(1 ) : 228 - 248 ( 1999 ) k. jain , m. mahdian , e. markakis , a. saberi , v. vazirani : greedy facility location algorithms analyzed using dual fitting with factor - revealing lp .
j. acm 50(6 ) : 795 - 824 ( 2003 ) r. krishnaswamy , m. sviridenko : inapproximability of the multi - level uncapacitated facility location problem .
soda 2012 : 718 - 734 s. li : a 1.488 approximation algorithm for the uncapacitated facility location problem .
comput . 222 : 45 - 58 ( 2013 ) y. li , d. du , n. xiu , d. xu : improved approximation algorithm for the facility location problems with linear / submodular penalties .
cocoon 2013 : 292 - 303 .
d. shmoys , e. tardos , k. aardal .
approximation algorithms for facility location problems ( extended abstract ) .
stoc 1997 : 265 - 274 .
m. sviridenko : an improved approximation algorithm for the metric uncapacitated facility location problem .
ipco 2002 : 240 - 257 g. xu , j. xu : an lp rounding algorithm for approximating uncapacitated facility location problem with penalties .
94(3 ) : 119 - 123 ( 2005 ) g. xu , j. xu : an improved approximation algorithm for uncapacitated facility location problem with penalties
. j. comb .
17(4 ) : 424 - 436 ( 2009 ) j. zhang : approximating the two - level facility location problem via a quasi - greedy approach .
math . program .
108(1 ) : 159 - 176 ( 2006 )
the idea is to construct a graph in which each facility on level @xmath5 is connected with exactly one facility on level @xmath161 .
we will describe in a few words how to do it , but the best idea is to read section 2 in @xcite .
let @xmath162 and @xmath163 be the set of facilities before and after modification respectively . for the highest level nothing change which means @xmath164 .
for each facility @xmath165 we have @xmath166 copies each connected with a different facility in @xmath4 .
the cardinality of set @xmath167 is equal to @xmath168 . in general : for each @xmath169 set @xmath4 has @xmath170 copies of each element in set @xmath171 and each copy is connected with a different element in the set @xmath172 .
note that there is an optimal integral solution with the form of a forest .
so we do not lose anything important for this optimal solution by modifying the graph in a way described above .
lets consider set @xmath175 of paths which start in client @xmath63 and end in the root of a tree @xmath116 .
we say that client @xmath6 has flow of value @xmath176 to tree @xmath116 if the total value of paths in set @xmath124 is equal to @xmath176 .
byrka et al . in @xcite
gives the following definition of function which is a lower bound for the probability of at least one path of a tree will be open as a result of rounding procedure on that tree .
we use @xmath177 to denote @xmath178 , similar for @xmath179 .
@xmath180 it is a product of the probability of opening the root node , and the ( recursively bounded ) probability that at least one of the subtrees has an open path , conditioned on the root being open .
now we are ready to give a function @xmath181 to bound the probability of opening at least one path when we have flow @xmath108 from one client to more than one tree .
let @xmath182 , which is one minus the biggest chance that no tree gives a route from the root to a leaf , using the previously defined @xmath183 function to express the success probability on a single tree .
now we can upper bound the expected connection and penalty cost of single algorithm . as it was proved in lemma [ worst_case ]
the worst case scenario is for client @xmath77 which is not a cluster center , so to upper bound the expected connection and penalty cost we can concentrate on clients from @xmath94 .
moreover from lemma [ g_0 ] we can suppose that @xmath184 .
the value of @xmath185 is a chance that at least one facility will be open in the set of close facilities .
@xmath186 expresses the chance that at least one distant facility of the considered client is open , but all close facilities are closed .
the remaining @xmath187 is the probability of connecting the considered client to the open facility by its cluster center .
the cost of this connection is bounded in lemma [ cluster_close_distance ] .
suppose @xmath188 is the cluster center of @xmath189 .
p_c \cdot d_{av}^{c}(j ) + p_d \cdot d_{av}^{d}(j ) + p_s \cdot ( \gamma d_{av}(j ) + ( 3 - \gamma)d_{max}(j)))\ ] ] @xmath191 @xmath192 @xmath193 @xmath194 you can find the justification for above inequalities in @xcite . summing over all clients we get the lemma . | the state of the art in approximation algorithms for facility location problems are complicated combinations of various techniques . in particular , the currently best 1.488-approximation algorithm for the uncapacitated facility location ( ufl ) problem by shi li is presented as a result of a non - trivial randomization of a certain scaling parameter in the lp - rounding algorithm by chudak and shmoys combined with a primal - dual algorithm of jain et al . in this paper
we first give a simple interpretation of this randomization process in terms of solving an auxiliary ( factor revealing ) lp . _ then , armed with this simple view point , we exercise the randomization on a more complicated algorithm for the @xmath0-level version of the problem with penalties in which the planner has the option to pay a penalty instead of connecting chosen clients , which results in an improved approximation algorithm . _ |
direct imaging and characterization ( through low resolution spectroscopy ) of extrasolar terrestrial planets ( etps ) is an exciting but challenging scientific application of coronagraphy . while the angular separation is well within the reach of a mid - sized telescope in the visible , the high star - planet contrast ( about @xmath1 in the visible ) requires a wavefront stability that can only be obtained with a space telescope .
the low apparent luminosity of an earth - like planet ( @xmath2 for an earth at 10pc ) requires the precious planetary photons to be well isolated from the stellar light ; otherwise , photon noise from the stellar light drives the exposure time unreasonably high and speckles from the stellar image , if not calibrated , prevent detection . a system capable of delivering @xmath1 contrast at less than 0.1 is therefore required for efficient detection and characterization of etps around a reasonably large ( @xmath330 ) sample of nearby stars .
the purpose of this work is to identify and quantify how fundamental physics imposes hard limits on what coronagraphs can achieve for the detection of terrestrial planets , and to compare these limits with what existing coronagraph designs can achieve .
coronagraphs designs simulated in this work are introduced and [ sec : existingc ] , where they are tentatively grouped in a four families . the configuration adopted for each design in this study
is also described . in [ sec : idealcase ] , a metric to quantify coronagraph performance is proposed , allowing existing coronagraph designs to be evaluated in the `` ideal '' case ( monochromatic light , unresolved central star ) .
a theoretical limit of coronagraph performance in this ideal case is also derived . in the next 2 sections ,
we closely look at the most fundamental ( unavoidable ) deviations from this ideal case , and their effects on coronagraphic performance : stellar angular size ( [ sec : stellarsize ] ) and presence of a zodiacal / exozodiacal background ( [ sec : zodi ] ) .
finally , a detailed simulation of coronagraphic observations of a sample of nearby stars is performed in [ sec : montecarlo ] to derive the efficiency of coronagraphs for detection of etps .
the coronagraphs studied in this work ( see table [ tab : coronagraphs ] ) are all the coronagraphs known to us to theoretically achieve a @xmath1 psf contrast within 5 @xmath4 of the central source , as illustrated in figure [ fig : psfs ] .
it is impossible to compare all the different kinds of coronagraph designs that have been suggested in all their details . but without these concrete examples , we could not communicate our new perspective and illustrate what are the fundamental physical effects limiting coronagraphic performance .
we offer this brief summary of previous work as historical background prior to using these designs .
more detailed descriptions of these coronagraphs can be found in the references given in this section .
we only consider in this paper unobstructed circular pupils .
we note that several techniques have been proposed to adapt `` unfriendly '' pupil shapes ( central obstruction , spiders ) to specific coronagraphs @xcite .
l c c c + achromatic interferometric coronagraph & aic & @xcite & + common - path achromatic interferometer - coronagraph & cpaic & @xcite & ( = aic ) + visible nulling coronagraph , x - y shear ( @xmath5 order null ) & vnc & @xcite & shear distance = @xmath6 pupil radius + pupil swapping coronagraph & psc & @xcite & shear distance = 0.4 pupil diameter + + + conventional pupil apodization and shaped - pupil & cpa & @xcite & prolate ( @xmath7 , 8% throughput ) + achromatic pupil phase apodization & ppa & @xcite & @xmath8 ; @xmath9 ; @xmath10 + phase induced amplitude apodization coronagraph & piaac & @xcite & prolate apodization + phase induced zonal zernike apodization & pizza & @xcite & not simulated + + + apodized pupil lyot coronagraph & aplc & @xcite & @xmath11 + apodized pupil lyot coronagraph , @xmath12 steps & aplc@xmath13 & @xcite & @xmath14 + band limited , @xmath5 order & bl4 & @xcite & @xmath15 intensity mask , @xmath16 + band limited , @xmath17 order & bl8 & @xcite & @xmath18 + + + phase mask & pm & @xcite & with mild prolate pupil apod .
+ 4 quadrant & 4qpm & @xcite & + achromatic phase knife coronagraph & apkc & @xcite & ( = 4qpm ) + optical vortex coronagraph , topological charge @xmath19 & ovc@xmath20 & @xcite & @xmath21 + angular groove phase mask coronagraph & agpmc & @xcite & ( = ovc ) + optical differentiation & odc & @xcite & mask : @xmath22 + + these coronagraphs look much like nulling interferometers : they rely on interferometric combination of discrete beams derived from the entrance pupil .
* aic : the achromatic interferometric coronagraph @xcite uses a beam splitter to destructively combine 2 copies of the entrance pupil , one of them achromatically @xmath23-phase shifted and flipped .
the final image exhibits a 180 degree ambiguity which may be removed at the expense of loosing achromaticity @xcite .
the common path achromatic interferometric coronagraph ( cpaic ) developed by @xcite achieves the same achromatic @xmath23-phase shifted and flipped nulling with a common path interferometer , and is therefore optically more robust .
* vnc : the visible nulling coronagraph , @xmath5 order @xcite is the coronagraph equivalent of a double - bracewell nulling interferometer .
two successive shears in perpendicular directions produce 4 beams , which , when combined , yield a @xmath5 order null in the pupil plane . by producing an image from the nulled pupil ,
this coronagraph combines a deep null with good imaging capabilities . in this paper ,
the shear distance is chosen to be 30% of the pupil size , which places the first transmission maximum ( 36% throughput ) at @xmath24 from the optical axis . increasing the shear distance allows smaller iwa , but reduces the throughput .
we note that another coronagraph , the @xmath5 order band limited coronagraph ( see [ ssec : lyot ] ) could be designed to perform the same pupil shear and nulling operations .
both designs are simply two different optical implementations of the same principle , and are referred to as vnc / bl4(1 ) in this paper .
* psc : in the pupil swapping coronagraph @xcite , parts of the pupils are geometrically swapped prior to destructive interferometric combination , thus avoiding the throughput loss due to the shear in the vnc . the psc design considered in this work is the @xmath5 order psc described in @xcite .
the pupil complex amplitude can be modified to yield a psf suitable for high contrast imaging , a property used by many coronagraph concepts .
apodization can be performed by a pupil plane amplitude mask ( conventional pupil apodization - cpa ) , which can be continuous @xcite or binary @xcite .
apodization by mach - zehnder type pupil plane interferometry was also suggested by @xcite to produce a continuous apodization , but is not explored in this work . within the cpa `` family '' ,
we adopt for our comparative study a prolate spheroidal apodized pupil design @xcite , optimally tuned , according to the performance criteria defined in [ sec : usefult ] , for a @xmath25 angular separation . in this design
, the apodizer transmits 8% of the light , and at @xmath25 separation , exactly half of this light is detectable according the the criteria defined in [ sec : usefult ] .
this apodization is nearly identical to the general prolate apodization function shown in figure 8 of @xcite . in cpas ,
tradeoffs exist between iwa , throughput and `` discovery space '' ( fraction of the field of view usable for planet detection ) .
a description of these tradeoffs can be found in @xcite , and they have been extensively explored for the design of binary masks / shaped pupils @xcite .
for example , in the checkerboard - mask design proposed in figure 1 of @xcite , the psf surface brightness reaches @xmath26 at @xmath27 , but at this separation , the companion needs to be on the diagonal to be observed , and only a quarter of its light then falls in a `` black '' quadrant . in this particular small - iwa design , at 2.8 @xmath4 , the peak throughput is then 3.8% for a square pupil , and the radially averaged throughput is @xmath32% . at large ( @xmath28 ) separation , the average and peak throughputs are 15.1% .
our prolate spheroidal apodization is not the optimal design if peak throughput is to be maximized ( in this case , the `` single prolate '' shaped pupil described in @xcite seems much superior ) .
our choice is therefore not optimal if the companion s position is previously known , in which case it could be intentionally placed at a given position angle .
binary pupil masks have the advantage of being achromatic and significantly easier to manufacture than continuous transmission masks .
this technological advantage however comes at a non - negligible cost in planet detection performance : * in the psf delivered by a non - obstructed , non - apodized circular aperture ( airy pattern ) , 84% of the energy transmitted by the pupil is contained in the central diffraction spot ; if the pupil is apodized with a continuous mask , this fraction is close to 100% ( but the spot is larger and less light is transmitted by the mask ) . in the psf delivered by binary apodization masks ,
this fraction is typically about 50% : there is a factor @xmath29 difference between the mask throughput and the `` airy throughput '' @xcite useful for detection . within the central diffraction spot of a planet
s image , the planet flux is proportional to the airy throughput while the background light contribution is proportional to the mask throughput .
binary masks therefore `` amplify '' the amount of background light mixed with the planet image .
this effect , discussed in [ sec : def ] is also found in sparse aperture interferometers , where additional background light is collected from the secondary lobes . *
the difference between mask throughput and airy throughput is diffracted in regions of the psf where planet detection is impossible . in a properly designed mask
, this unwanted light avoids a large central region around the psf s center . as the number of openings or slits in a binary mask increases ,
these two effects become less of a concern , and the mask asymptotically becomes identical to a continuous mask .
if only one half of the focal plane is considered , an amplitude pupil apodization can be replaced by a phase - only apodization ( pupil phase apodization , ppa ) , just as phase corrections in the pupil can only cancel focal plane speckles in one half of the field of view . high contrast imaging with phase apodization
was proposed by @xcite , who found solutions for broadband imaging and obtained contrast / throughput performances similar to amplitude apodization designs ( although only over a quarter of the field of view ) .
@xcite independently computed a ppa solution for the hubble space telescope pupil to suppress diffraction in half of the field of view .
the ppa design adopted in this work uses a square pupil mask and the sum of two identical phase apodizations ( equation 6 in @xcite ) , one in the x - direction , the other in the y - direction .
although the ppa mask has a 100% throughput within the square aperture , it exhibits large phase slopes at its edges , which are necessary to suppress diffraction in a quarter of the field but significantly increase the intensity of diffracted light in other parts of the field , reduce the amount of light in the central core of the psf and broaden the psf core . just as for binary amplitude apodization masks ,
the ppa s large ratio between the pupil mask throughput ( here equal to 0.6 - area of the largest square that fits within the pupil ) and the light concentrated in the psf core ( @xmath30 ) is problematic in the presence of zodiacal and exozodiacal backgrounds .
this coronagraph uses a lossless amplitude apodization of the pupil performed by geometric redistribution of the light rather than selective absorption @xcite .
the same beam remapping technique is known as beam shaping in laser science @xcite and radio astronomy .
the geometrical remapping of the telescope pupil amplifies phase slopes , therefore enabling observations at angular separations @xmath3 3 times smaller than with a conventional amplitude apodization .
two designs are considered in this work : * * piaa : * the detection occurs in the first focal plane after the beam remapping .
starlight therefore falls on the detector but is concentrated in a single diffraction spot .
this design is adopted for practical reasons in [ sec : idealcase ] and [ sec : stellarsize ] to avoid having to tune a focal plane mask diameter . * * piaac : * a focal plane mask occults starlight in the first focal plane .
an inverse beam remapping unit then restores the original pupil to recover field of view .
while a focal plane mask size needs to be chosen for this design , it is adopted in [ sec : montecarlo ] to maintain a sharp planet image over the zodi and exozodi backgrounds .
an interesting alternative to cpa techniques is phase induced zonal zernike apodization , pizza , @xcite , which achieves the pupil amplitude apodization without compromising the telescope s throughput or angular resolution .
pizza , which is not simulated in this work , may be too chromatic to be practically suitable for imaging etps .
the iwa delivered by pizza , cpa and ppa are similar . in a lyot coronagraph , the telescope pupil and the focal plane mask
are not perfectly fitted one for another , resulting in strong stellar leaks ( much higher than the @xmath26 goal ) within the coronagraph s lyot stop . for a lyot coronagraph to deliver high contrast performance ,
the complex amplitude in the `` intermediate '' focal plane ( equal to the fourier transform of the entrance pupil multiplied by the focal plane occulter ) needs to be `` band - limited '' : its fourier transform ( @xmath31 complex amplitude in the lyot pupil plane ) should have nearly zero power in a frequency range which is then selectively transmitted by the lyot pupil stop .
improved performance over the original lyot design can therefore be obtained by either : * adapting the pupil to the hard edge focal plane mask .
this is achieved in the apodized pupil lyot coronagraph , aplc , described in @xcite , where the entrance pupil of a lyot coronagraph with a hard edge focal plane occulter is optimally apodized . a slightly different approach , explored by @xcite ( see figure 3 of the paper ) ,
is to apodize the pupil after the hard edge focal plane occulter . as suggested by @xcite , the output of an aplc can be used as the input of a second stage aplc : these are the multistep aplcs ( noted aplc@xmath13 , where @xmath12 is the number of steps ) . * adapting the focal plane mask to the telescope pupil .
this is achieved by using `` band limited masks '' in the focal plane : the @xmath5 order band limited coronagraph , bl4
@xcite and 8th order band limited coronagraph , bl8 @xcite . a visible nuller interferometer ( vnc ) equivalent to the bl4 coronagraph design adopted in this work could be designed , and we therefore refer to this bl4 design as vnc / bl4(2 ) .
phase mask coronagraphs ( pm , 4qpm , ovc , odc ) introduce phase shifts in the focal plane .
this allows smaller iwas than offered by aplcs and bls .
* pm : the phase mask coronagraph @xcite uses a circular @xmath23-shifting focal plane mask , and , to allow @xmath1 contrast , a mild pupil amplitude apodization @xcite .
* 4qpm : the 4-quadrant phase mask @xcite uses a focal plane mask which shifts 2 out of 4 quadrants of the image by @xmath23 .
the achromatic phase knife coronagraph @xcite is another form of 4qpm .
* ovc : in the optical vortex coronagraph @xcite and the angular groove phase mask coronagraph @xcite , a focal plane vortex phase mask replaces the 4 quadrant phase mask of the 4qpm , thus avoiding the `` dead zones '' of the 4qpm . in @xmath32 polar
coordinates , the mask phase is equal to @xmath33 , where @xmath19 is the _
topological charge_. ovcs with high topological charges ( @xmath34 ) exhibit low sensitivity to low - order aberrations .
* odc : the optical differentiation coronagraph @xcite is using a phase and amplitude focal plane mask .
some of the designs listed in table [ tab : coronagraphs ] may be combined in series .
the multistep aplc concept @xcite illustrates how such a `` cascade '' can simultaneously improve iwa and throughput by designing each step for a lower contrast level .
@xcite showed how a nulling interferometer can benefit from a cpa , and a similar combination may also be advantageous on a circular pupil ( vnc - cpa for example ) .
the entrance pupil apodization required for the aplcs may be performed by a piaa unit to improve iwa and throughput .
although combinations of coronagraphic techniques are not studied in this paper , the results obtained on individual designs can provide a good insight into which combinations may be most beneficial . pupil replication
@xcite has been proposed to transform a phase slope ( equivalent to angular separation on the sky ) into phase discontinuities in the replicated pupil . for the off - axis planet , this creates secondary diffraction peaks far from the central source image .
the authors pointed out that it could enhance the performance of a coronagraph with which it would be combined .
@xcite , in a sensitivity study of pupil replication , suggested to use it after a 4qpm .
the tip - tilt sensitivity of both techniques ( pupil replication introduces phase steps in the wavefront of an off - axis source ; see [ sec : ttsens ] for the 4qpm ) may be a concern for direct imaging of terrestrial planets .
a coronagraphic effect can be obtained by placing an occulting screen , at least as big as the telescope diameter , between the star and the telescope @xcite .
this external occulter needs to be designed to not diffract starlight within the telescope aperture , needs to be placed far from the telescope ( on a separate spacecraft ) , and the telescope - occulter - star alignment needs to be maintained during observations .
while maintaining the alignment is especially challenging , a `` conventional '' imaging telescope can be used without the need for exquisite control of wavefront aberrations .
since we aim at quantitatively comparing the performance of coronagraph designs , we need a
fair performance metric : one that can be applied uniformly to all coronagraph designs .
coronagraph performance is usually quantified with inner working angle ( usually defined as the minimal angular distance at which the throughput for the planet is half of the maximal throughput ) , throughput ( measured at large angular separation ) , and search area ( fraction of the focal plane image within which a planet can be detected ) .
the iwa is somewhat arbitrary , as there is no well - defined stellar - to - companion light ratio threshold beyond which companion light suddenly becomes unmeasurable .
iwa is coupled with throughput : for example , a coronagraph with a 10% maximum throughput achieving 5% throughput at 1 @xmath4 has a smaller iwa than a coronagraph with a 100% maximum throughput achieving 30 % throughput at 1 @xmath4 . in this example
, the larger iwa coronagraph outperforms the smaller iwa coronagraph .
search area is also somewhat arbitrary and often can not be quantified as a single number .
for example , diffraction from 4 spider vanes in 2 perpendicular directions ( we assume here that the width of the diffraction spikes created by the spiders is equal to the iwa of the coronagraph ) covers @xmath35 of the field of view at the iwa and @xmath36 at large angular separations . * metrics commonly used to quantify coronagraph performances ( iwa , throughput and search area ) are therefore difficult to use to directly compare coronagraph designs . * a simpler , more direct measure of how well a coronagraph can separate planet light from star light , the `` useful throughput '' ,
is proposed here .
it is defined as the amount of planet light which can be used toward detection , and is expressed as a function of the planet position relative to the star and the planet / star contrast .
when averaged over all position angles , it conveniently quantifies the coronagraph performance with a single number ( making it possible to directly compare coronagraph performances ) , function of the angular separation .
this function accurately includes the effects of iwa , throughput and search area , and is a good estimation of the coronagraph performance . to compute this quantity ,
we must first define when planet light in the coronagraph detector plane becomes `` unusable '' because of excessive starlight mixed with it .
theoretically , direct detection of companions embedded in much brighter starlight is possible if the starlight component on the image is well characterized .
practically , due to calibration problems and photon noise ( exoplanets are faint ) , companion light mixed with much brighter starlight is unusable .
there is however no well - defined stellar - to - companion light ratio threshold beyond which companion light becomes unmeasurable : the criteria we will use to define the
useful throughput ( hereafter denoted throughput ) is therefore somewhat arbitrary .
graphical representation of the useful throughput . in this 1-d example , the stellar and planet psf are shown with some overlapping .
the useful throughput is obtained by integrating the companion ( planet ) light from @xmath37 to @xmath38 ; in this interval , the integrated flux contributions from the central source and the companion are equal . ]
as illustrated in figure [ fig : usablet ] , the `` useful throughput '' is the maximum fraction of the planet s light gathered by the telescope which can be separated from starlight .
the `` separation '' criteria used is that , integrated over the area of the focal plane considered , there is as much planet light as there is residual stellar light .
physically , this area may be a single pixel or a group of pixels on a focal plane array detector . to compute the useful throughput of a coronagraph at a particular location in the image plane , we first model the images @xmath39 and @xmath40 produced respectively by a star and a much fainter planet in the coronagraph .
pixels are then sorted in decreasing order of @xmath41 ; for the first pixels in this sorted list , a large fraction of the light collected originates from the planet while the last pixels of this list are dominated by starlight .
the number @xmath42 is chosen such that , over the first @xmath42 pixels of this list , the total starlight collected is equal to the total planet light collected .
this total amount of planet light , divided by the the flux in planet light that is collected by the entrance aperture , is the * useful throughput*. we note that if , on every pixel of the image , starlight exceeds planet light , then the useful throughput will be zero .
figure [ fig : usefult ] shows the useful throughput for the coronagraphs listed in table [ tab : coronagraphs ] . for small angular separations ( @xmath43 ) the aic is the most efficient coronagraph , with a 50% useful throughput at @xmath44 . based on our assumptions , the phase mask coronagraphs ( especially ovc@xmath45 , pm and 4qpm ) and piaa are the best choices from @xmath44 outward .
many coronagraphs in this list never achieve a very high throughput because of strong entrance pupil apodization ( cpa , aplc , and to a lesser degree aplc@xmath13 ) , beam splitting ( psc , vnc ) , or pupil diaphragming ( vnc / bl4 , bl8 ) .
the range of performance is very large : the aic offers an iwa about 8 times smaller than the cpa , and the piaa s throughput is about 10 times the cpa s throughput beyond @xmath46 .
many of the coronagraph designs studied in this paper were developed in the last few years , and there is therefore hope that a large number of other coronagraph designs will be discovered in the near future .
it is particularly interesting to wonder how much of the upper left area of figure [ fig : usefult ] ( high throughput at small angular distance ) can / will be accessed . in the following section ,
we derive a fundamental limit of coronagraph throughput ( shown in figure [ fig : usefult ] ) using a universal
model of coronagraphs .
we consider an optical system , shown in figure [ fig : matrixu ] , in which light enters through a pupil ( which may or may not be circular ) which we represent by a finite ( but large ) number @xmath12 of regularly spaced elements .
similarly , the output of this optical system , defined by the points at which light is detected , blocked or exits the system , is represented by @xmath12 points . to a coherent light source ( such as a point source at infinity ) corresponds a distribution of complex amplitude at the pupil of this system , written here as a vector @xmath47 of @xmath12 complex numbers .
although @xmath47 is written as a vector , it represents a 2d distribution of complex amplitude across the pupil : each of its coefficient is the complex amplitude of the incident wave across one of the @xmath12 elements in the entrance pupil .
we note that this description is only valid to the extent that complex amplitude is constant across each pupil element - this limitation is equivalent to a field of view limit in the focal plane , and our model is therefore not suitable for point sources at more than @xmath48 from the optical axis . for convenience ,
we choose to normalize this vector such that the total light intensity in the pupil is unity ( @xmath49 ) .
optical systems are ( usually ) linear in complex amplitude : the complex amplitude anywhere in the system can be written as a linear combination of complex amplitudes in the entrance pupil .
for example , fourier transforms and fresnel propagation simulations are linear operations commonly performed to compute coronagraphic images .
the output complex amplitude vector @xmath50 can therefore be written as : @xmath51 where @xmath52 is a nxn complex matrix which entirely describes the optical system and @xmath53 is the point source position in the sky .
the output vector @xmath50 includes light that is blocked by masks ( turned into heat ) , exits the system ( for example , reflected back to the sky ) or falls on the detector ( these are the only coronagraph outputs that can be measured and therefore used for planet detection ) .
conservation of energy in the system imposes @xmath54 : @xmath52 is therefore a * complex unitary matrix , function of the coronagraph optics , but independent of the entrance wavefront * ( and therefore independent of the source position @xmath53 in the sky ) .
wavefront control elements , if present , are also part of @xmath52 .
as illustrated in figure [ fig : matrixu ] , in a coronagraph , the goal is to create system output(s ) for which an on - axis source will produce very little light , but a closeby off - axis source s flux will be transmitted .
mathematically speaking , a matrix @xmath52 needs to be chosen such that some coefficients of @xmath55 ( we denote @xmath56 the number of such coefficients ) are very small for @xmath57 , but relatively high for other values of @xmath53 . to
this set of coefficients corresponds a sub - space @xmath58 of dimension @xmath42 ( shown as the bottom part of vector @xmath50 on figure [ fig : matrixu ] ) .
the coronagraph optimization problem can be further restricted by fixing a 2d vector @xmath59 , where @xmath60 is the planet off - axis position at which the coronagraph s throughput is to be maximized . in this case , a change of coordinate system ( rotation ) within the sub - space @xmath58 can concentrate all the planet flux within @xmath58 on a single coefficient .
this change of coordinate system is a function of @xmath60 , and aligns , within @xmath58 , the 2d vector @xmath59 with one of the directions of the new coordinate system .
physically this is equivalent to placing multiple beam splitters and phase shifters to concentrate all of the planet light within @xmath58 in a single output beam : this would be possible thanks to the fact that planet light is fully coherent .
this norm - preserving coordinate change is itself a unitary matrix which can be integrated within @xmath52 ( by direct multiplication with the `` old '' @xmath52 ) , in which case only a single column of this new @xmath52 matrix is now relevant to the problem : we denote @xmath61 this column ( @xmath61 is a vector ) . since @xmath52 is unitary , @xmath62 . in this new basis
, @xmath63 is the square absolute value of one of the coefficients of the new vector @xmath50 and is equal to the coronagraph throughput at the sky position @xmath53 : it is therefore the quantity to be minimized for @xmath64 and maximized for @xmath65 .
we use in this section the new base described at the end of the previous section , where the sub - space @xmath58 corresponds to a single coefficient of @xmath50 . in this `` optimized coronagraphic projection '' , the energy in this single coefficient is the useful throughput of the coronagraph , as long as , within this single `` pixel '' , the stellar flux is smaller than the planet flux . excluding all other coefficients of @xmath50
, we can now equivalently refer to the coronagraph throughput or useful throughput .
we denote @xmath66 the on - axis coronagraph throughput ( no more than @xmath26 for a system designed to image etps in the visible ) and @xmath67 the input complex amplitude for an on - axis source .
we note that @xmath66 is a throughput , not a contrast , and should therefore ideally be no more than @xmath26 times the coronagraph throughput for the planet .
it is now possible to design an ideal theoretical coronagraph for detecting a source at the position @xmath60 by choosing a complex vector @xmath61 of norm 1 such that : 1 .
@xmath68 : this is the coronagraphic `` contrast '' at position @xmath60 .
@xmath69 is as large as possible : this is the the square root of the coronagraph intensity throughput at position @xmath60 .
graphical representation of the coronagraph optimization problem . ]
this way of looking at the problem shows that optimizing a coronagraph is equivalent to choosing a projection axis @xmath61 which is `` mostly '' perpendicular to @xmath67 ( constraint ( 1 ) ) and as co - linear as possible with @xmath59 ( constraint ( 2 ) ) .
a glance at figure [ fig : projection ] reveals that if @xmath59 and @xmath67 are close ( ie .
the planet position @xmath60 is close to the optical axis ) , the coronagraphic throughput can not be high .
as shown in figure [ fig : projection ] , the input vector @xmath59 can be decomposed into an `` along optical axis '' component and a `` perpendicular to optical axis '' component : @xmath70 where @xmath71\:a(0)\ ] ] is the projection of @xmath59 onto @xmath67 ( `` along optical axis '' ) . using this decomposition ,
the linearity of the coronagraph in complex amplitude , and the @xmath72 constraint , an upper bound can be established for the square root of the coronagraphic throughput : @xmath73 \ : \sqrt{\epsilon } + a_{\perp}(\alpha_p ) \bullet c.\ ] ] since @xmath74 and @xmath75 , @xmath76 < ||a_{\perp}(\alpha_p)||.\ ] ] upper limit on the off - axis throughput of a coronagraph for different entrance pupils . ]
the upper limit given by equation [ equ : upplim ] is shown in figure [ fig : upplim ] for a `` normal '' circular pupil ( unobstructed , unapodized ) as well as for gaussian - apodized pupils and a pupil with a large 80% central obstruction .
not surprisingly , amplitude apodization decreases the throughput at all distances and also pushes further the point at which the maximum throughput is reached ( decrease in angular resolution ) . with a large central obstruction
, the throughput reaches its maximum value quicker ( slight increase in angular resolution ) .
this result shows that a coronagraph , regardless of its on - axis throughput ( as long as it is much smaller than 1 of course ) , can not have a throughput exceeding 50% at 0.5 @xmath4 for a circular pupil .
curiously , the upper limit derived by equation [ equ : upplim ] oscillates with distance , especially in the large central obstruction case .
this is due to a slight similarity in the wavefront at 0 @xmath4 and @xmath77 for the circular pupil case . in the extreme case of a 2-pupil interferometer of baseline @xmath78
, the same effect would lead to a @xmath79 periodicity of the theoretical maximum throughput .
since @xmath80 ( right angle in figure [ fig : projection ] ) , @xmath81 by definition , and @xmath82 is the intensity of the non - aberrated non - coronagraphic psf at position @xmath60 ( equation [ equ : a0a0 ] ) : * the maximum throughput of a coronagraph is equal to one minus the non - aberrated non - coronagraphic psf of the telescope .
* in the case of a non - obstructed circular pupil , this throughput is one minus the airy pattern , as can be seen in figure [ fig : upplim ] .
this last property is a direct consequence of the linearity in complex amplitude of the coronagraph .
since the coronagraph needs to remove the central star , it also removes the `` flat non - tilted component '' of any incident wavefront , which , expressed as a function of the source position on the sky , is the airy function ( for an unobstructed pupil ) . with the above analysis
, the vector @xmath61 that would be required to reach the upper limit at a given position @xmath60 can be computed .
each of the @xmath12 elements of the pupil can be treated as a small independent beam , and we could construct the optical system with a series of beam splitters ( not necessarily 50%/50% ) and phase shifts .
each beamsplitter combines 2 beams and produces 2 output beams . since @xmath83 = 1 $ ] , the problem is equivalent to concentrating all the flux into a single beam if the entrance vector ( @xmath12 beams ) is equal to @xmath61 .
it is relatively easy to do incrementally : at each step 2 beams are combined with a phase shift + beam splitter such that all the light is sent into one output beam ( this is possible since the beams are fully coherent between them and we know exactly their amplitudes and phases ) .
@xmath84 beam splitters are required in this scheme .
if the coronagraph is only intended to work close to the optical axis , @xmath12 could be quite small ( @xmath12 scales as the square of the outer working angle ) , resulting in an optical system that could possibly be built .
unfortunately , this coronagraph is only guaranteed to reach the theoretical limit at a single planet position @xmath60 . a better way to build the `` ideal '' coronagraph
is to have the light of an on - axis point source entirely concentrated in a single output .
this would ensure that , for large values of @xmath12 , the coronagraphic throughput ( now measured on @xmath84 beams ) is equal to the theoretical limit simultaneously at all source positions .
special attention would be given to the arrangement of the beam splitters that only affect the @xmath84 `` useful '' outputs in order to allow the source position to be easily retrieved from comparisons between their intensities .
the sensitivity of some coronagraphs to tip - tilt errors has been investigated by several authors ( see for example @xcite ) .
a sun - like star at 10pc is 1mas across : with an 8 m visible telescope its apparent radius is about @xmath85 .
many coronagraphs can not maintain a @xmath1 contrast on such an extended central source . even if the telescope diameter were reduced ( or the wavelength increased ) such that the planet , at maximum elongation , would be at @xmath86 , the stellar radius would still be @xmath87 .
figure [ fig : usefulttt ] shows how coronagraphic performance ( useful throughput ) decreases as the stellar radius increases , and can be used to somewhat arbitrarily divide coronagraphs in 3 groups : * high sensitivity to stellar radius ( aic , 4qpm , pm , ovc@xmath45 , odc ) .
these coronagraphs can not maintain high contrast even with a 0.01 @xmath4 stellar radius .
they would be very difficult to use for etp imaging . *
moderate sensitivity to stellar radius ( vnc and bl4 with small iwa / large shear : vnc / bl4(1 ) , psc , aplcs , ovc@xmath13 with large value of @xmath12 ) .
these coronagraphs may be used efficiently on `` small '' diameter telescopes ( for which the stellar radius is about @xmath87 or less ) . *
low sensitivity to stellar radius ( piaac , vnc and bl4 with large iwa / small shear , bl8 , cpa ) .
these coronagraphs are suitable for direct imaging and characterization of etps , as they tolerate stellar radii of @xmath85 .
the robustness of vnc / bl4(1 ) compared to the vnc / bl4(2 ) is entirely due to the larger iwa chosen in this this study for the vnc / bl4(2 ) . for both the bl4 and vnc
, there is a iwa vs. sensitivity to stellar radius tradeoff . for most coronagraphs , increasing the stellar diameter first compromises contrast at small angular separations .
even the vnc / bl4(2 ) and piaac , which are relatively insensitive to angular size , lose throughput at small angular separations in the @xmath85 radius case .
the cpa and bl8 stand out as being extremely robust to stellar angular size : their throughput curves are almost identical from 0 to 0.1 @xmath4 stellar radius .
the cpa owes its relative immunity to tip - tilt errors to the field - invariance of its psf .
the sensitivity of the bl8 to tip - tilt errors was studied in detail by @xcite , who also find that the bl8 can tolerate relatively high levels of focus , coma , and astigmatism . direct comparison between figures [ fig : usefult ] and [ fig : usefulttt ]
shows a very disappointing result : the the best coronagraphs in the point source case ( aic , pm , 4qpm , ovc2 ) all perform very poorly when the stellar angular size is considered . increasing
the stellar size from 0 to @xmath87 has moved the smallest angular separation at which 50% throughput is reached from 0.5 @xmath4 ( for the aic ) to slightly more than @xmath86 ( for the piaa ) .
is this behavior imposed by fundamental physics and therefore unavoidable ? or is there hope to find a low - iwa coronagraph which is not so dramatically sensitive to stellar angular size ? in this section , we quantify how the angular extent of the source affects the coronagraph throughput .
we assume that the coronagraph is observing a disk of uniform surface brightness and angular radius @xmath88 . using the same framework as in [ sec : theoreticallimit_not ] , the star is now modeled by a series of vectors @xmath89 , where the @xmath90 are uniformly distributed on the stellar disk ( @xmath91 ) and @xmath92 .
an orthonormal base can be built from the vectors @xmath93 with the following process : 1 .
the vector @xmath94 ( @xmath95 initially ) of this base is chosen to be co - linear with the vector @xmath93 with the highest norm .
all vectors @xmath93 are replaced by their projection on a hyperplane perpendicular to @xmath94 .
@xmath96 is incremented and we return to step 1 with the new vectors @xmath93 .
this algorithm insures that , in the orthonormal base obtained , the coefficients of the vectors @xmath93 decrease very rapidly .
we denote @xmath97 , @xmath98 the maximum absolute value of the coefficient @xmath96 in this new base across all @xmath93 .
the 2d representation of the first 5 vectors @xmath94 are shown in figure [ fig : modes ] .
the same modes can be obtained by linear expansion of the pupil complex amplitude @xmath99 for small values of @xmath100 and @xmath101 ( here , pupil plane coordinates are ( @xmath102,@xmath103 ) and point source angular coordinates on the sky are ( @xmath100,@xmath101 ) ) .
graphical representation of the first 5 modes of the orthonormal base @xmath94 for a circular unobstructed pupil .
the vectors @xmath94 are independent of stellar size , and their coefficients are all real numbers ( no imaginary part ) . ] as previously shown , a coronagraph which needs to cancel the light across the stellar disk can be represented by a vector @xmath61 of norm 1 .
we denote @xmath104 , @xmath105 the coefficients of @xmath61 in the new base of modes @xmath94 .
to insure that the coronagraph cancels the light of the star by a factor @xmath66 , the following constraint on each @xmath104 can be imposed : latexmath:[\[\label{equ : ciconst } equivalent to stating that the integrated starlight transmitted by the coronagraph is smaller than @xmath66 , but we have verified numerically that it yields a stellar throughput which is in most cases within 20% of @xmath66 .
this can be explained by the very rapid decrease of @xmath97 with @xmath96 ( especially for smaller stars ) and the rapid quadratic increase of stellar leaks if @xmath107 goes above @xmath108
. for modest stellar sizes ( below @xmath4 ) , equation [ equ : ciconst ] constrains only the first few @xmath104 ( the others are constrained by @xmath109 ) .
we note that adopting a zero stellar size will result in @xmath110 and @xmath111 for @xmath112 . in this case ,
equation [ equ : ciconst ] only imposes a constraint for @xmath95 . to obtain the maximum throughput as a function of angular distance ,
the vector @xmath59 is decomposed in the new base , and its first coefficients are multiplied by the values of @xmath107 derived in equation [ equ : ciconst ] .
upper limit on the off - axis throughput of a coronagraph for different stellar radii . ]
the results obtained , shown in figure [ fig : upplimtt ] and also overplotted in figure [ fig : usefulttt ] , confirm the large effect of a small increase in stellar size on the coronagraphic iwa .
in the point source case , the 50% throughput can be reached at 0.5@xmath4 ; for a @xmath85-radius source , the 50% throughput limit moves to almost @xmath86 .
this behavior is contrary to a `` geometrical optics '' intuition , which would suggest that increasing the stellar size by 0.1 @xmath4 moves iwa by 0.1 @xmath4 . in the linear algebra model of coronagraphy
presented in this work , this behavior is however understandable : as the stellar diameter increases , a higher number of modes needs to be removed from the wavefront ; each mode removed increases the coronagraph s `` blind spot '' by a @xmath113 area in the focal plane . a coronagraph which follows exactly the limit shown in figure [ fig : upplimtt ]
can be assembled from discrete beam splitters and phase shifts as described in
[ ssec : canitbebuilt ] .
an example of such a design is shown in figure [ fig : dicc ] .
a @xmath114 coronagraph is often referred to as a @xmath115 order null coronagraph , while @xmath116 insures a @xmath5 order null ( and so on ... ) . in this section ,
our linear algebra - based model is used to predict and explain how different coronagraph behave at small angular separation .
these results allow us to understand how coronagraphic performance degrades as stellar angular diameter increases , and allow in [ ssec : specklenulling ] to determine if wavefront control techniques can mitigate this problem .
psfs of selected small - iwa coronagraphs as a function of angular separation . at small angular separation ,
the psf of aic , pm and 4qpm are centro - symmetric , making it impossible to determine on which side of the optical axis is the source .
the ovc@xmath118 psfs at @xmath119 and @xmath120 are dominated by numerical roundoff errors .
note that the brightness scale is not constant , as images on the left are much fainter than images on the right . ]
the smallest - iwa coronagraphs considered in this study ( iac , pm , 4qpm ) all remove mode m0 ( see figure [ fig : modes ] ) from the incoming wavefront but `` transmit '' ( at least partially ) modes m1 and m2 . at very small angular separations ,
the `` transmitted '' wavefront is therefore dominated by a combination of modes m1 and m2 .
the coefficients of this combination are proportional to the off axis distance , resulting in a coronagraphic residual light intensity proportional to the square of the source s angular separation ( @xmath115 order coronagraphs ) .
it is therefore expected that for these coronagraphs , the complex amplitude in the final image is also proportional to the source coordinates : the complex amplitude for a source at position @xmath121 is the opposite of the complex amplitude for a source at position @xmath122 . * at small angular separations , our theoretical analysis therefore predicts that the images ( square modulus of the complex amplitudes ) of two point sources on opposite sides of the optical axis are identical , and that , as the source moves away from the optical axis , its image is simply multiplied by a constant , but does not change structure . * both these predictions
are confirmed by numerical simulations , as shown in figure [ fig : psfs_smallangle ] ( top 3 rows ) . just as in the case of second - order coronagraphs ,
the linear algebra - based theory outlined in the previous sections can predict / explain the behavior of other coronagraphs at small angular separation .
the interferometric combination coronagraph ( icc6 ) which will be described in
[ sec : idealdef ] perfectly removes the first 6 modes shown in figure [ fig : modes ] , but leaves other modes untouched .
the `` transmitted '' wavefront of this theoretical coronagraph at small angular separation is therefore dominated by the @xmath123 term in the polynomial expansion of @xmath99 , corresponding to modes @xmath124 , @xmath125 , @xmath126 and @xmath127 ( not shown in figure [ fig : modes ] ) .
we can therefore predict that along any direction , its psf is a fixed image multiplied by the 6@xmath128 power of source angular separation , and that psfs for 2 point sources on opposite sides of the optical axis will be identical .
a similar behavior is seen in figure [ fig : psfs_smallangle ] for the ovc@xmath118 , which also removes the first 6 modes shown in figure [ fig : modes ] .
since this coronagraph also attenuates further modes , the psf structures and throughput are however different from the icc@xmath118 , which is designed to perfectly cancel the first 6 modes and fully transmit the others .
active elements ( deformable mirrors ) can be used to create a coherent `` anti speckle halo '' which destructively interferes with the unwanted residual focal plane speckles .
these speckle nulling techniques , first proposed by @xcite , were more recently studied by several authors @xcite and successfully tested in laboratory @xcite .
the star can be modeled as a set of point sources uniformly distributed on the stellar disk ( these sources are not mutually coherent ) .
we first assume that a second order coronagraph is used , and we consider only one of these point sources ( not at the center of the disk ) which contributes to the coronagraphic leakage by adding an image on the focal plane .
this particular image is fully coherent and can therefore be canceled by a speckle nulling technique .
the shape / state of the active optical element(s ) ( dm for example ) in the speckle control device will be such that the phase and amplitude of the `` anti speckle halo '' is the opposite ( same amplitude , @xmath23 phase offset ) of the leakage : destructive interference between the 2 will leave no light in the focal plane .
we now decide to `` freeze '' the state of the speckle nulling device and move the point source on the other side of the optical axis .
this multiplies the complex amplitude of the coronagraphic leakage by @xmath135 ( [ sssec:2ndorderc ] ) but does not change the `` anti speckle halo '' ( which is independent of source position within the small angular displacement considered ) .
the leakage and `` anti speckle halo '' will therefore add rather than subtract because of a phase mismatch , and the leakage intensity will be 4 times higher than without speckle nulling . to every point on
the stellar disk corresponds an equally bright point on the other side of the optical axis : speckle nulling can not reduce coronagraphic leaks in a second - order coronagraph .
the same `` phase mismatch '' argument applies to @xmath136 order coronagraph such as @xmath137 .
independently of this `` phase mismatch '' effect ( which would not apply to a @xmath5 order coronagraph dominated by modes m3 to m5 ) , coronagraphic leaks increase as the point on the stellar disk moves away from the optical axis , and there is therefore a similar `` amplitude mismatch '' effect when stellar points at different radii are considered .
the stellar angular size problem described in [ sec : ttsens ] truly is a fundamental limit . pointing errors ( or any higher order optics aberrations ) , on the other hand ,
produce a coherent leak which can be suppressed by wavefront control techniques .
low order aberrations can be measured from the coronagraphic leaks they produce .
a more optimal strategy is to `` catch '' aberrations before they start to increase coronagraphic leaks by utilizing `` for free '' the large amount light that is blocked by the coronagraph .
this can usually be done by re - imaging of the light reflected by the focal plane occulter ( or , for phase mask coronagraphs , the light outside the geometric pupil ) .
we note that speckle sensing and nulling usually require a small spectral bandwidth , as the speckle coherence is low in white light : the observing spectral band may have to be split with dichroics in small `` bins '' prior to applying speckle nulling .
the `` useful throughput '' introduced in [ sec : usefult ] accurately quantifies the coronagraphic performance in a system composed solely of a star and a single planet .
a planetary system is however observed against incoherent zodiacal and exozodiacal light backgrounds . for simplicity , we assume here that both background components are constant across the field ( this is an approximation for the exozodiacal light ) and that they can be numerically subtracted to the photon noise level .
we consider a planet+background system ( no central star ) in which @xmath138 is the ratio between total planet light and total background light within the planet image ( which we define here , for convenience , as the smallest focal plane area containing half of the planet s psf light )
. we denote @xmath139 this value for a non - coronagraphic imaging telescope ( airy pattern ) .
coronagraphs unfortunately tend to decrease @xmath138 ( @xmath140 ) by a factor which we denote the * diffractive efficiency factor ( def ) * : * pupil amplitude apodization and/or clipping in the lyot plane increases the angular size of the planet image , therefore mixing more background light with the planet light . *
the aic delivers double images of the planet , therefore dividing @xmath138 by two .
* coronagraphs delivering psfs in which a significant fraction of the light is outside the main diffraction core decrease @xmath138 .
three factors determine the def : the width @xmath141 of the psf core ( more exactly , the square root of the `` useful '' psf area ) , the fraction @xmath142 of the psf energy in the diffraction core , and the total throughput @xmath143 . the collectible planet energy is proportional to @xmath144 while the background flux collected in the same focal plane area is proportional to @xmath145 : the def is therefore proportional to @xmath146 .
the diffractive efficiency factors of most of the coronagraphs in table [ tab : coronagraphs ] are listed in table [ tab : coronagraphs_eval ] , along with other key characteristics .
the def is especially low on the cpa and bl8 due to their poor angular resolution . as discussed in
[ sec : cpabinarymask ] , a cpa with a binary mask ( shaped pupil ) could `` amplify '' background levels by another factor 2 : the combined effect is then equivalent to dividing the planet / background ratio by @xmath147 .
the zodiacal light background level is about @xmath148 ( can be less in some directions ) , and an earth - like planet ( 6400 km radius , 0.33 visible albedo , orbiting a sun - like star at 1 au ) , at maximum elongation , is @xmath149 where @xmath150 is the system distance in @xmath151 . if we assume that the zodiacal , exozodiacal and planet spectra are identical , the distance at which , within the planet s image ( @xmath4 wide ) , zodiacal+exozodiacal background matches the planet s flux , is , for a favorable `` face - on '' geometry : @xmath152 where @xmath153 is the exozodiacal cloud dust content , normalized to our own system , and @xmath154 is the diffractive efficiency factor described in [ sec : def ] . for example , with @xmath155 , @xmath156 , @xmath157 and @xmath158 , @xmath159 : for all but a few targets , photon noise from the background level will be the main source of noise , and will drive the exposure time required for successful detection of etps . at longer wavelength , the psf size increases , therefore gathering more zodi+exozodi background and reducing @xmath160 .
l c c c c c c c c c c c c c aic & 0.50 & 0.57 & 1.0 & 0.5 & 0.38 & 0.15 & 6.14 & @xmath161 & @xmath161 + vnc / bl4(1 ) & 0.09 & 0.36 & 1.65 & 0.37 & 1.49 & 1.27 & 1.27 & 1.43 & @xmath161 + psc & 0.18 & 0.73 & 1.0 & 0.5 & 1.13 & 0.79 & 0.79 & 3.24 & @xmath161 + cpa & 0.08 & 0.08 & 1.81 & 0.31 & 4.20 & 4.42 & 4.42 & 4.42 & 4.47 + ppa & 0.08 & 0.33 & 1.35 & 0.18 & 3.91 & 3.29 & 3.29 & 3.30 & 4.02 + aplc & 0.20 & 0.20 & 1.22 & 0.67 & 2.41 & 2.01 & 2.17 & 4.02 & @xmath161 + aplc@xmath45 & 0.27 & 0.27 & 1.11 & 0.81 & 1.64 & 1.26 & 1.78 & 4.71 & @xmath161 + vnc / bl4(2 ) & 0.35 & 0.70 & 1.19 & 0.71 & 3.02 & 2.17 & 2.20 & 2.21 & 3.77 + bl8 & 0.26 & 0.28 & 1.86 & 0.29 & 2.96 & 3.04 & 3.04 & 3.04 & 3.05 + pm & 0.71 & 0.71 & 1.0 & 1.0 & 0.69 & 0.23 & 2.94 & 3.59 & @xmath161 + 4qpm & 1.0 & 1.0 & 1.0 & 1.0 & 0.84 & 0.28 & 2.39 & @xmath161 & @xmath161 + ovc@xmath45 & 1.0 & 1.0 & 1.0 & 1.0 & 0.84 & 0.21 & 2.21 & @xmath161 & @xmath161 + ovc@xmath162 & 1.0 & 1.0 & 1.0 & 1.0 & 1.62 & 0.62 & 0.66 & 1.74 & @xmath161 + ovc@xmath118 & 1.0 & 1.0 & 1.0 & 1.0 & 2.45 & 1.02 & 1.02 & 1.19 & 2.87 + ovc@xmath163 & 1.0 & 1.0 & 1.0 & 1.0 & 3.26 & 1.41 & 1.42 & 1.42 & 3.01 + odc & - & 0.70 & 1.18 & 0.72 & 0.97 & 0.50 & 4.59 & @xmath161 & @xmath161 + piaa & 1.0 & 1.0 & 1.0 & 1.0 & 1.88 & 1.28 & 1.38 & 1.66 & 2.0 + @xmath118 & * 1.0 * & * 1.0 * & * 1.0 * & * 1.0 * & * 1.32 * & * 0.76 * & * 0.76 * & * 0.76 * & * 1.76 * + in [ sec : ttideal ] , we have demonstrated that a coronagraph which delivers theoretically ideal throughput on a partially resolved star can be designed using beam splitters and phase shifts .
the example shown in figure [ fig : dicc ] achieves this goal by confining starlight in a few interferometric outputs ( as the required contrast and/or stellar diameter increases , the number of such outputs grows ) .
however , no care was given to the distribution of planet light across the remaining outputs , and it is likely that background and planet light are highly mixed , resulting in high diffractive efficiency factor and poor imaging capabilities .
we show here that a coronagraph can be designed to simultaneously deliver optimal throughput for a partially resolved star ( see [ sec : ttideal ] ) , full angular resolution imaging and @xmath164 . in a classical non - coronagraphic imaging telescope ,
the system matrix u shown in figure [ fig : matrixu ] is a fourier transform , offering full angular resolution imaging and @xmath164 but poor coronagraphic performance : we denote @xmath165 this `` imaging '' matrix . on the other hand , a `` coronagraphic '' matrix @xmath166 optimized for coronagraphic performance , such as the one shown in figure [ fig : dicc ] , efficiently isolates starlight but suffers from poor imaging capability and high diffractive efficiency factor .
the same iterative method used to produce the `` vectors '' c , x and y in figure [ fig : dicc ] can be used to produce any predefined unitary matrix .
unitary matrices can also be cascaded : the input of a matrix , instead of being the telescope pupil , can be the output of another matrix .
for example , it is possible to cascade a `` pupil to focal plane '' matrix @xmath165 with a `` focal plane to pupil '' @xmath167 matrix , with no net result ( the product of the two matrices is the unity matrix ) .
design of an `` ideal '' coronagraph by cascading unitary matrices . in the first step ,
multiplication of the complex amplitude in the pupil plane by the coronagraphic matrix @xmath166 concentrates starlight in a few outputs .
this operation can be inverted by multiplication by @xmath168 , therefore restoring the telescope pupil .
multiplication by @xmath165 produces a clean `` image '' of the planet . at each step ,
distribution of stellar ( vector on the left ) , planet ( center vector ) and background ( right vector ) light is shown in greyscale : from white for no light to black for maximum light .
the `` ideal '' coronagraph is obtained by simply masking the few outputs containing starlight in the `` coronagraphic outputs '' intermediate step . ]
one such cascade is shown in figure [ fig : dicc2 ] : it is designed to produce a `` coronagraphic intermediate step '' where all starlight is concentrated in a few outputs and a final `` focal plane '' step with good imaging capability and @xmath164 .
the overall system matrix @xmath169 is equivalent to a classical imaging telescope , as detection ( conversion from amplitude to intensity ) is performed after the last step of this chain .
stellar light can be easily masked in the `` coronagraphic intermediate step '' , with minimal impact on the planet light ( unless the planet is very close to the star ) and background light ( which is always spread across all outputs ) .
the sequence of optical steps to perform is therefore :
1 . select the transformation @xmath166 which best isolates stellar light from planetary light 2 .
block starlight in the few outputs where it has been concentrated .
3 . undo transformation 1 to recover the original telescope pupil ( @xmath170 ) 4 .
select the transformation which concentrates the planet light in a single bright diffraction peak ( imaging @xmath165 ) in an optimized implementation of this scheme , where the number of optical elements is to be minimized , there is no need to separate steps 2 and 3 , as the intermediate product ( the original telescope pupil ) is not used .
an identical scheme is used in the piaac design , although this coronagraph falls short of the theoretically ideal performance derived in this work . the piaac first concentrates stellar light in a very tight diffraction core ( step 1 ) , which is then blocked by a mask ( step 2 ) .
the pupil remapping used to perform step 1 unfortunately `` scrambles '' the planet image beyond @xmath171 .
a second remapping unit is therefore then used to undo the first remapping ( step 3 ) .
finally , an image is formed ( step 4 ) .
since the only currently known solution to build the ideal coronagraph requires many beam splitters , this type of coronagraph is referred to as the interferometric combination coronagraph ( icc@xmath13 , where @xmath12 is the number of modes removed in the coronagraphic intermediate step ) in this work .
we have used the above method to simulate an icc@xmath118 , designed to remove the 6 modes shown in figure [ fig : modes ] .
the resulting coronagraphic throughput is of the @xmath17 order ( just as for the bl8 coronagraph ) , with an iwa of 1.32 @xmath4 .
this particular design is well suited for direct imaging of etps and accurately represents the absolute limit of what coronagraphy can achieve for this purpose .
the useful throughput vs. angular separation of the icc@xmath118 is virtually identical to the `` theoretical limit '' shown in the upper panel ( stellar radius @xmath172 ) of figure [ fig : usefulttt ] .
the performance of this `` ideal '' coronagraph ( as well as the performance of a few other coronagraphs ) is modeled in more detail in the following section .
table [ tab : coronagraphs_eval ] concludes the previous sections by listing the main characteristics of the coronagraphs studied in this paper .
in the previous sections , coronagraph performance has been quantified by the * useful throughput * and the * diffractive efficiency factor*. together , these two quantities could be used to evaluate , for a given target ( star+planet+background ) , the signal - to - noise ( snr ) as a function of exposure time .
this simple model can unfortunately be quite inaccurate : * exozodiacal light is not a smooth background ( it is significantly brighter closer to the star and in a high inclination system , the exozodi disk can be a narrow linear feature ) : its effect can not be accurately modeled by the diffractive efficiency factor .
planets are brighter when in nearly full phase , but are then closer to the star and therefore over brighter portions of the exozodi cloud .
* in the presence of background light ( zodiacal + exozodiacal light ) , faint parts of the psfs that contribute to the useful throughput can have a negligible contribution to the detection snr : the spatial distribution of the planet light affects the snr .
while the useful throughput and the def provided us with valuable insight into coronagraph performances , accurate estimation of the detection sensitivity requires explicit computation of images delivered by the coronagraph . in this section , the performance of six promising coronagraphs identified in the previous sections ( cpa , piaac , bl8 , ovc6 , icc6 and vnc / bl4(2 ) )
is quantified for the direct detection of earth - type planets .
details of this simulation are given in
[ sec : simumodel ] .
as the goal of this study is to show what can be potentially achieved with various coronagraphs , very optimistic assumptions have consistently been made .
they are listed in table [ tab : montecarlo ] and further detailed in the next section .
lccl optics throughput & 0.25 & & excludes losses due to coronagraph design + wavefront quality & perfect & & no time - variable wavefront errors + detector & perfect & & no readout noise or dark current + imaging wavelength & 0.5 0.6 & @xmath173 m & + + zodiacal background & 23.28 22.24 & @xmath174 & function of target ecliptic latitude + + 1 zodi exozodiacal cloud brightness & 22.53 & @xmath174 & at the habitable zone , for a face - on system + exozodiacal cloud inner edge & 0.02@xmath175 & au & @xmath176 star bolometric luminosity + exozodiacal cloud outer edge & 5.0@xmath175 & au & @xmath176 star bolometric luminosity + exozodiacal cloud thickness & 0.0 & au & thickness unresolved by telescope + exozodi optical depth & @xmath177 & & + + planet radius & 6400 & km & + planet albedo & 0.33 & & + semi - major axis & @xmath175 & au & @xmath176 star bolometric luminosity + eccentricity & 0 & & + orbit inclination & @xmath178 & rad & statistical median for random orientation + + a list of potential target was first created by selecting stars within 100pc in the hipparcos catalog .
the effective temperature of each star was computed from the b - v color index , using the color index vs. temperature law of main sequence stars .
stars with effective temperature between 4060 k and 7200 k ( types f , g and k ) were then selected , and those listed as subarcsecond separation doubles removed .
giants were excluded from our sample using the stellar bolometric luminosity / effective temperature criteria outlined in figure [ fig : lumtcr ] .
bolometric luminosity ( 1 = sun ) and effective temperature of stars within 100pc .
the thick black line shows the criteria used in this work to exclude giants : all stars above this line are rejected from our sample of potential targets . ]
this sample now contains 16921 stars .
each star is assumed to have an etp . in this section ,
etp ( extrasolar terrestrial planet ) is used to mean a standard earth analog : lambert phase function , radius = 6400 km , albedo = 0.33 , circular orbit at 1 au times the square root of the star s bolometric luminosity ( computed assuming the star is a perfect blackbody radiator ) .
the inclination of the orbit , assumed to be coplanar with the exozodiacal disk , is set to @xmath179 for all systems : this is the median value expected for a random system orientation . a final selection criteria is to reject systems for which the planet , at maximum elongation , is within 1 @xmath4 of the star : none of the coronagraphs studied in this work can efficiently detect planets within this angular separation .
for an 8 m telescope observing in the 0.50.6 @xmath173 m band , this final sample contains 3429 stars at an average distance of 51 pc .
zodiacal light is modeled as a uniform background across the telescope s field of view . for each target , it is assumed that the observation occurs at the time of the year when the zodiacal background is minimal , which is about 40 days before or after opposition for most targets .
the resulting zodiacal background , derived from @xcite , is a function of ecliptic latitude , and ranges from @xmath180 23.28 ( ecliptic poles ) to @xmath180 22.24 ( ecliptic equator ) .
the uniform zodiacal background is sampled by a square grid of points ( 0.2 @xmath4 sampling in 2 orthogonal directions - about 30000 psfs in the 20 @xmath4 radius field of view ) .
the psf is computed for each point , and the final image delivered by the telescope / coronagraph is obtained by adding the 30000 coronagraphic psfs together . the exozodiacal cloud is assumed to be a thin disk with dust optical depth , as seen from above the 2d disk , varying as @xmath181 ( surface brightness @xmath182 ) .
this power law index is derived from modeling of our own zodiacal cloud in the earth neighborhood @xcite . for a 1 zodi exozodiacal disk , the dust optical depth at the location of the planet
is chosen to be identical to the vertical optical depth of our zodiacal cloud at 1 au . in this model ,
the @xmath183 m surface brightness of a face - on exozodi cloud at the location of the planet is thus a function of the ratio between the star bolometric luminosity ( which defines where the planet is ) and the star luminosity at @xmath183 m ( which defines how much the star illuminates the dust ) .
the inner edge of the exozodi cloud is set at 0.02 au times the square root of the star s bolometric luminosity ( dust sublimation ) , and its outer edge is 5 times the planet - star distance .
the image of the exozodiacal cloud delivered by the telescope / coronagraph is computed with the same 0.2 @xmath4 sampling as used for the zodiacal background image .
it is assumed that the stellar diameter , exozodi cloud , zodiacal background and telescope / coronagraph characteristics are perfectly known .
the planet detection snr is therefore driven by photon noise and is proportional to the square root of the exposure time . the photon noise ( from the planet , star , zodiacal and exozodiacal images ) and the signal ( the planet image )
are evaluated for each pixel of the image .
the planet * detection snr * is obtained by the quadratic sum of the pixel snrs : this optimal weighting scheme favors pixels with large planet flux and low zodi+exozodi+star flux .
we assume here that at the time of the observation , the position of the planet along its orbit is unknown . for a given exposure time ,
the planet detection snr is computed at each point along the planet orbit . a limit on the snr
is adopted ( snr @xmath184 7 ) for successful detection , and the number of points along the planet orbit for which this limit is reached yields the * planet detection probability*. this probability , also referred to as completeness @xcite , is therefore a function of exposure time .
a related metric used in this study , derived from the planet detection probability , is the single - exposure integration time required to reach a given planet detection probability .
figure [ fig : exampleimages ] shows some example frames obtained by the simulation code for the 6 coronagraphs selected .
these images nicely illustrate coronagraphic characteristics quantified in the previous section : * * useful throughput at large separation . *
the cpa , bl8 , and to a lesser extent vnc / bl4(2 ) suffer from low coronagraphic throughput . as a result ,
the planet s image , even if well outside the coronagraph mask s influence , appears noisy ( few photons detected ) .
the piaac , ovc6 and icc6 , on the other hand , enjoy nearly 100% throughput : the planet image is brighter and less noisy . * * angular resolution . *
the cpa , bl8 , and to a lesser extent vnc / bl4(2 ) have poorer angular resolution : the planet image is larger and more zodi / exozodi light is mixed with it .
* * ability to work at small angular separation .
* none of the coronagraph tested can detect the planet on a 2 m telescope with the exposure time , wavelength , and throughput used in figure [ fig : exampleimages ] .
detection appears feasible on a 4 m telescope with the vnc / bl4(2 ) , piaac , ovc6 and icc6 , but requires a 6 m telescope with the bl8 .
finally , an 8 m telescope is needed for detection with a cpa .
* * sensitivity to stellar angular size .
* the cpa and bl8 are extremely robust to stellar angular size : starlight leaks are virtually nonexistent even on the 12 m telescope . with the vnc / bl4(2 ) , piaac , ovc6 and icc6 , starlight is visible on the 12 m telescope ( equivalent to a 6 m telescope observing the same system at 5pc ) , although it is still fainter that the exozodiacal contribution .
images similar to the ones shown in figure [ fig : exampleimages ] can be generated for each possible position of the planet along its orbit .
these images have been used to compute for hip 56997 the exposure time required to reach 20% and 50% detection ( snr=7 ) probability .
results , in table [ tab : examplecase ] , show very large discrepancies between coronagraph designs .
for example , on a 6 m telescope , the piaac requires 44 times less exposure time to reach a 50% detection probability than the cpa . for this target , the performance of the piaac , ovc6 and icc6 coronagraph are similar , while the other 3 coronagraph are significantly inferior ( especially the cpa ) . for the vnc / bl4(2 ) , it was assumed that the preferential direction of the coronagraph was optimally aligned with the system orientation .
lcccccc & + 2 m & & & & & & 12.4 hour + 4 m & & 2.1 hour & 11.4 hour & 34 min & 48 min & 14.3 min + 6 m & 6.9 hour & 14 min & 44.8 min & 5.7 min & 8.6 min & 3.78 min + 8 m & 63 min & 4.8 min & 17 min & 2.1 min & 3.1 min & 110 sec + 12 m & 12 min & 2.7 min & 4.3 min & 45 sec & 56 sec & 42 sec + + & + 2 m & & & & & & + 4 m & & 12.7 hour & & 2.3 hour & 2.2 hour & 44.1 min + 6 m & & 59 min & 5.0 hour & 14 min & 19.8 min & 12.23 min + 8 m & 4.3 hour & 18 min & 1.8 hour & 5.8 min & 7.6 min & 5.4 min + 12 m & 31 min & 55 min & 19 min & 2.0 min & 2.4 min & 113 sec + in the previous section , a single target was adopted while the telescope size was variable . here , we adopt fixed telescope sizes and evaluate the telescope / coronagraph combination performance on a sample of nearby stars , defined in [ sec : targetsample ] .
results are shown in figure [ fig : idealcaseetime ] for 4 m and 8 m telescopes . in a purely background - limited detection , exposure time
is expected to scale linearly with the background level : approximately 7 times longer for the 10 zodi exozodi case than for the 1 zodi exozodi case ( assuming the local zodi background corresponds to half a zodi of exozodiacal light ) .
the fact that this is generally true in figure [ fig : idealcaseetime ] confirms the importance of the def in coronagraph designs . for coronagraphs with low defs , such as piaac , ovc6 and icc6 , increasing the exozodi however results in moderate increase of exposure time for nearby targets ( non background - limited detections ) .
this effect is most clearly visible in the 8 m telescope case for the easiest targets .
figure [ fig : idealcaseetime ] compares how many targets fall below the 1hr , 10hr and 1 day required exposure time levels with different exozodi contents . with the 4 m telescope ,
the cpa and bl8 can not access the `` habitable zone '' of targets beyond @xmath185 , resulting in a very small number of potential targets ( 8 and 16 targets respectively for cpa and bl8 if `` potential target '' is defined as one that requires less than 10hr exposure time to reach the 50% detection probability ) .
the vnc / bl4(2 ) , thanks to its higher throughput , can access 34 targets . with this telescope diameter ( 4 m ) , the stellar angular size is sufficiently small to allow improved performance over cpa and bl8 , which are more robust to stellar angular size . in increasing order of performance , the ovc6 , piaac and icc6 offer from 87 to 177 targets in the 1-zodi case .
it is interesting to note that the performance of the ovc6 and piaac is very similar , while the `` ideal '' icc6 offers almost twice as many targets as the piaac .
this is mostly due to the benefit of smaller iwa in the icc6 over the piaac : @xmath186 vs. @xmath187 , which increases the iwa - accessible volume by 2.9 .
the planet intrinsic brightness and the background level prevent access to distant systems , and there is therefore little gain in reducing the iwa below @xmath187 . with any of the 3 `` top '' coronagraph designs simulated in this work ( piaac , ovc6 and icc6 ) ,
a 4 m telescope could , with a 10hr per observation limit and a 1-zodi assumption on the exozodi , perform a @xmath188 target survey for etps . the total exposure time to complete this survey , assuming 6 observations of 10hr for each target , would be well under a year , and therefore sounds realistically achievable . under the same assumptions , strong exozodi ( 10 zodi )
would cut in half the sample size .
figure [ fig : idealcaseetime ] shows that doubling the telescope diameter ( from 4 m to 8 m ) multiplies the number of accessible targets by about 8 for all exozodi content , coronagraph type and exposure time limit combinations tested .
the relative efficiencies of the 6 coronagraphs tested stay the same , with , in order of decreasing performance , the icc6 , piaac and ovc6 able to access several hundred targets in the 1-zodi case ( with a 1hr exposure time limit criteria ) , the vnc / bl4(2 ) about 50 targets and the bl8 and cpa a few tens of targets . in this study
, we have used an earth - analog as a prototype for earth - like planets .
the broad concept of candidate earth - like ( potentially habitable ) planets likely extends to include planet diameters to several times smaller and larger than earth .
orienting a survey to optimize detection for a different planet radius would of course change the observation times for each system and the number of systems which could be surveyed .
similarly , an observing program can be optimized for any target list and planet description to maximize the productivity with respect to total detections or other criterion .
we have quantified the performance limit imposed by fundamental physics on the direct imaging of etps . a solution to build an `` optimal '' coronagraph reaching this limit
was proposed , but may be technologically difficult to implement .
fortunately , we have also demonstrated that two existing coronagraph designs ( piaac and ovc6 ) yield performances reasonably close ( within a factor 2 in number of accessible targets ) to this fundamental limit .
these recently proposed designs could image etps around up to about 100 stars with a 4 m telescope , or several hundred stars with an 8 m telescope . * * wavefront control at the sub - @xmath189level*. thanks to the large number of stellar photons available , recent detector developments , and the expectation that a space telescope can be designed to offer a relatively stable wavefront ( with little rapidly varying aberrations , as opposed to atmospheric turbulence for ground - based telescopes ) , it may be reasonable to hope for this level of correction .
several authors have recently explored promising wavefront control schemes free of aliasing and non - common path aberrations @xcite , and contrast levels obtained in laboratory @xcite are approaching the desired performance . * * chromatic effects * can seriously limit the coronagraph s usable spectral band .
a large number of solutions have been proposed to `` achromatize '' coronagraphs , many of them specific to a single coronagraph design @xcite .
in addition to coronagraph - induced chromatic effects , the beam delivered by the telescope is expected to be chromatic in amplitude and optical pathlength differences ( non - ideal coatings , chromatic diffraction propagation between optics ) . in general ,
chromatic effects can be mitigated by reducing chromaticity in individual components ( beam splitters , phase - shifters , apodization masks ) or / and by splitting the spectral band into several narrower bands with dichroics .
more fundamentally , the assumption that detection is photon noise limited seems highly optimistic : etps are embedded in an exozodiacal light unlikely to be perfectly smooth and symmetrical . since good angular resolution may help to distinguish between a planet and exozodiacal light features ( such as arcs or rings ) , it is important to avoid using coronagraphs that clip or strongly attenuate the edge of the telescope pupil if at all possible .
the results obtained in this work should therefore be cautiously considered as upper limits on what may be achieved with a particular coronagraph / space telescope design for the specific goal of imaging etps around nearby stars .
we also note that several low throughput / large iwa coronagraphs studied in this paper can be more favorably designed for ground - based applications , where contrast requirements are relaxed by about 4 orders of magnitude .
the linear algebra approach to coronagraphy presented in this work successfully explains key characteristics and fundamental limits of coronagraphy . while we have focused our study on the fundamental limitations of coronagraphy ( due to stellar angular size and background contamination ) , this model can be extended to evaluate the sensitivity of coronagraphs to wavefront aberrations other than tip - tilt .
the same model can also be used to derive fundamental limits and guide the design of coronagraphs for more `` exotic '' pupil shapes , from segmented pupils telescopes to short or long baseline interferometric arrays .
this work was carried out with the support of the national astronomical observatory of japan and jpl contract numbers 1254445 and 1257767 for development of technologies for the terrestrial planet finder mission .
we thank lyu abe and remi soummer for providing comments and helping the authors to understand tradeoffs for the apodized pupil lyot coronagraph .
we thank stuart shaklan for his useful comments and review of this work .
abe , l. , vakili , f. , boccaletti , a. 2001 , , 374 , 1161 aime , c. , soummer , r. , ferrari , a. 2001 , , 379 , 697 aime , c. , soummer , r. , ferrari , a. 2002 , , 389 , 334 aime , c. , soummer , r. 2004 , 5490 , 456 aime , c. 2005 , , 117 , 1012 aime , c. 2005 , , 434 , 785 baudoz , p. , rabbia , y. , & gay , j. 2000 , , 141 , 319 baudoz , p. , boccaletti , a. , rabbia , y. , & gay , j. 2005 , , 117 , 1004 bord , p.j . ,
traub , w.a .
2006 , , 638 , 488 brown , r.a .
2004 , , 607 , 1003 brown , r.a .
2005 , , 624 , 1010 cash , w. c. , kasdin , n. j. , seager , s. , arenberg , j. 2005 , , 5899 , 274 codona , j.l . , angel , r. 2004 , , 604 , l117 foo , g. , palacios , d.m . , & swartzlander , g.a .
2005 , opt .
, in press gay , j. , rabbia , y. 1996 ,
paris _ , 322 , ser .
ii b , 265 gonsalves , r. , nisenson , p. 2003
, , 115 , 706 greenaway , a.h . ,
spaan , f.h.p . ,
mourai , v. 2005 , , 618 , l165 guyon , o. , roddier , f.j .
2000 , , 4006 , 377 guyon , o. 2003 , , 404 , 379 guyon , o. , pluzhnik , e.a .
, galicher , r. , martinache , f. , ridgway , s.t . , woodruff , r.a .
2005 , , 622 , 744 guyon , o. 2005 , , 629 , 592 guyon , o. , shao , m. 2006 , accepted for publication in jacquinot , p. , & roisin - dossier , b. 1964 , _ prog . opt .
_ , 3 , 29 kasdin , n.j . , vanderbei , r.j . , spergel , d.n . , & littman , m.g .
2003 , , 582 , 1147 kelsall , t. , weiland , j.l . , franz , b.a . ,
reach , w.t . ,
arendt , r.g . ,
dwek , e. , freudenreich , h.t . ,
hauser , m.g . ,
moseley , s.h .
, odegard , n.p . ,
silverberg , r.f . ,
wright , e.l .
1998 , , 508 , 44 kuchner , m.j . , traub , w.a .
2002 , , 570 , 900 kuchner , m.j .
, crepp , j. , ge , j. 2005 , , 628 , 466 labeyrie , a. 2004 , proceedings of `` astronomy with high contrast imaging '' , held in nice , france , 6 - 10 october , 2003 _ eas pub .
_ , 12 levasseur - regourd , a.c . , dumont ,
r. 1980 , , 84 , 277 lloyd , j.p . , sivaramakrishnan , a. 2005 , , 621 , 1153 malbet , f. , yu , j.w . , shao , m. 1995 , , 107 , 386 martinache , f. 2004 , _ j. of opt .
, 6 , 809 martinache , f. , guyon , o. , pluzhnik , e.a . , galicher , r. , ridgway , s.t .
2006 , , 639 , 1129 mawet , d. , riaud , p. , absil , o. , surdej , j. 2005 , , 633 , 1191 mennesson , b.p .
, shao , m. , levine , b.m . ,
wallace , j.k . ,
liu , d.t .
, serabyn , e. , unwin , s.c . , beichman , c.a .
2003 , , 4860 , 32 murakami , n. , baba , n. 2005 , , 117 , 295 nisenson , p. , & papaliolios , c. 2001 , , 548 , l201 nishikawa , j. , kotani , t. , murakami , n. , baba , n. , itoh , y. , & tamura , m. 2005 , , 435 , 379 oti , j.e . , canales , v.f . , & cagigal , m.p .
2005 , , 630 , 631 palacios , d.m .
2005 , , 5905 , 196 pluzhnik , e.a . ,
guyon , o. , ridgway , s.t .
, martinache , f. , woodruff , r.a . ,
blain , c. , galicher , r. 2006 , , in press , astro - ph/0512421 riaud , p. , mawet , d. , & absil , o. 2005 , , 628 , l81 roddier , f. , & roddier , c. 1997 , , 109 , 815 rouan , d. , riaud , p. , boccaletti , a. , clnet , y. , & labeyrie , a. 2000 , , 112 , 1479 shaklan , s.b . , & green , j.j .
2005 , , 628 , 474 shealy , d.l .
2002 , , 4770 , 28s sivaramakrishnan , a. , soummer , r. , sivaramakrishnan , a.v . , lloyd , j.p . ,
oppenheimer , b.r . ,
makidon , r.b .
2006 , , in press slepian , d. 1965 , _ j. opt
_ , 55 , 1110 soummer , r. , aime , c. , & falloon , p.e .
2003 , , 397 , 1161 soummer , r. , dohlen , k. , & aime , c. 2003 , , 403 , 369 soummer , r. 2005 , , 618 , l161 swartzlander jr .
2006 , accepted to optics letters tavrov , a.v . ,
kobayashi , y. , tanaka , y. , shioda , t. , otani , y. , & kurokawa , t. 2005 , _ optics letters _
, 30 , 2224 traub , w.a . , & vanderbei , r.j .
2003 , , 599 , 695 trauger , j.t . , moody , d. , gordon , b. , gursel , y. , ealey , m.a . , & bagwell , r.b .
2003 , , 4854 , 1 vanderbei , r.j .
, spergel , d.n . , & kasdin , n.j .
2003 , , 590 , 593 vanderbei , r.j . ,
spergel , d.n . ,
& kasdin , n.j .
2003 , , 599 , 686 vanderbei , r.j . ,
kasdin , n.j . ,
& spergel , d.n .
2004 , , 615 , 555 vanderbei , r.j . , traub , w.a .
2005 , , 626 , 1079 vanderbei , r.j .
2006 , , 636 , 528 yang , w. , & kostinski , a.b .
2004 , , 605 , 892 | many high contrast coronagraph designs have recently been proposed . in this paper , their suitability for direct imaging of extrasolar terrestrial planets is reviewed .
we also develop a linear - algebra based model of coronagraphy that can both explain the behavior of existing coronagraphs and quantify the coronagraphic performance limit imposed by fundamental physics .
we find that the maximum theoretical throughput of a coronagraph is equal to one minus the non - aberrated non - coronagraphic psf of the telescope .
we describe how a coronagraph reaching this fundamental limit may be designed , and how much improvement over the best existing coronagraph design is still possible .
both the analytical model and numerical simulations of existing designs also show that this theoretical limit rapidly degrades as the source size is increased : the `` highest performance '' coronagraphs , those with the highest throughput and smallest inner working angle ( iwa ) , are the most sensitive to stellar angular diameter .
this unfortunately rules out the possibility of using a small iwa ( @xmath0 ) coronagraph for a terrestrial planet imaging mission .
finally , a detailed numerical simulation which accurately accounts for stellar angular size , zodiacal and exozodiacal light is used to quantify the efficiency of coronagraph designs for direct imaging of extrasolar terrestrial planets in a possible real observing program .
we find that in the photon noise limited regime , a 4 m telescope with a theoretically optimal coronagraph is able to detect earth - like planets around 50 stars with 1hr exposure time per target ( assuming 25% throughput and exozodi levels similar to our solar system ) .
we also show that at least 2 existing coronagraph design can approach this level of performance in the ideal monochromatic case considered in this study . |
all the simplest paths of a given network were calculated in the dual space by converting the networks from the primal to the dual representation , where straight lines are mapped into nodes and the intersection between straight lines were mapped into edges .
straight lines are found by using a version of the _ icn _ ( intersection continuity negotiation ) algorithm @xcite .
more specifically , given an edge @xmath0 , we search among the adjacent edges attached to @xmath3 , @xmath43 , that one that is most aligned to @xmath0 . if the angle @xmath44 between @xmath0 and @xmath43 is smaller or equal to @xmath45 , we assume that these two edges belong to the same straight line .
this procedure continues until no more edges are assigned to the same straight line .
then , the procedure is repeated in opposite direction starting from the adjacent edges attached to node @xmath2 . once assigned to a straight line
, an edge is removed from the network .
as it is , this algorithm produces different networks depending on the choice for the initial edge . to overcome this ambiguity ,
our algorithm always starts with the edge that give us the longest straight line for a given network .
after this straight line is fully detected and its edges deleted , we choose the next edge that will give us the second longest straight line and so on .
the algorithm ends when there are no more edges left in the network .
once all straight lines have been identified , the dual representation is built by looking at the intersection between straight lines .
each straight line is mapped onto a node in the dual space and two nodes are connected together if their respective straight lines intersect each other at least once . to illustrate this process
, we show in fig [ fig : illus](b,1 ) an example of planar network in the primal representation where the edges are colored according the @xmath46 of the straight line they belong to . in fig .
[ fig : illus](b,4 ) we show the dual representation of the same network .
it is important to note that the longest straight lines , in this example represented by orange , red , green and magenta give rise to hubs in the dual space . in order to calculate the simplest path between nodes 1 and 2 from fig .
[ fig : illus](b,1 ) , we search for the shortest path between their respective straight lines in the dual space , cyan and blue in this case .
as it can be seen , there are two paths with length 4 , @xmath47 and @xmath48 .
each of them define a subgraph in the primal representation , here represented by the set of magenta lines in fig .
[ fig : illus](b,2 ) and red lines in fig .
[ fig : illus](b,3 ) for paths a and b , respectively . then we evaluate the shortest path distance between nodes 1 and 2 over these subgraphs and we adopted the shortest one as the simplest path - green dashed in fig .
[ fig : illus](b,3 ) . the black path in figs .
[ fig : illus](c , d ) represent the shortest path between nodes 1 and 2 .
the gini coefficient quantifies the inequalities of the lengths of straight lines , and is defined as in @xcite @xmath49 where @xmath50 is the average length of straight lines and @xmath51 is the number of straight lines .
the gini coefficient lies in the range @xmath52 $ ] and @xmath53 when all lengths are equal . on the other hand ,
if all lengths but one are very small , the gini coefficient will be close to @xmath54 .
we thank prof . a. perna for the dataset of the leaf _ hymenanthera chathamica _ and prof .
a. adamatzky for providing us the _ physarum policephalum_. mb thanks h. berestycki and m. gribaudi for interesting discussion .
mb acknowledges funding from the eu commission through project eunoia ( fp7-dg.connect-318367 ) .
mpv , es , pb , mb designed , performed research and wrote the paper .
the authors declare no competing financial interests .
strano , e. , et al .
urban street networks : a comparative analysis of ten european cities .
arxiv.1211.0259 barthelemy , m. , bordin , p. , berestycki , h. & gribaudi , m. self - organization self - organization versus top - down planning in the evolution of a city .
rep . _ * 3 * , 2153 ( 2013 ) . | shortest paths are not always simple . in planar networks , they can be very different from those with the smallest number of turns - the simplest paths .
the statistical comparison of the lengths of the shortest and simplest paths provides a non trivial and non local information about the spatial organization of these graphs .
we define the simplicity index as the average ratio of these lengths and the simplicity profile characterizes the simplicity at different scales .
we measure these metrics on artificial ( roads , highways , railways ) and natural networks ( leaves , slime mould , insect wings ) and show that there are fundamental differences in the organization of urban and biological systems , related to their function , navigation or distribution : straight lines are organized hierarchically in biological cases , and have random lengths and locations in urban systems . in the case of time
evolving networks , the simplicity is able to reveal important structural changes during their evolution .
a planar network is a graph that can be drawn on the two - dimensional plane such that no edges cross each other @xcite .
planar graphs pervade many aspects of science : they are the subject of numerous studies in graph theory , in combinatorics @xcite and in quantum gravity @xcite .
planar graphs are also central in biology where they can be used to describe veination patterns of leaves or insect wings .
in particular , the vasculature of leaves @xcite displays an interesting architecture with many loops at different scales , while in insects , the vascular network brings strength and flexibility to their wings . in city science , planar networks are extensively used to represent , to a good approximation , various infrastructure networks @xcite . in particular , transportation networks @xcite and more recently streets patterns @xcite are the subject of many studies @xcite that are trying to characterize both topological ( degree distribution , clustering , etc . ) and geometrical ( angles , segment length , face area distribution , etc . ) aspects of these networks . despite a large number of studies on planar networks ,
there is still a lack of global , high - level metrics allowing to characterize their structure and geometrical patterns .
such a characterization is however difficult to achieve and in this article , we will discuss an important aspect of planar graphs which is intimately connected to their geometrical organization . in this respect , we will define new metrics and test them on various datasets , both artificial ( roads , highways , railways , and supply networks ) and natural ( veination patterns of leaves and wings , slime mould ) enabling us to obtain new information about the structure of these networks . we will now introduce the main metrics used in this article . generally speaking , we can define different types of paths for a given pair of nodes @xmath0 .
a usual quantity is the shortest euclidean path of length @xmath1 which minimizes the distance travelled to go from @xmath2 to @xmath3 .
we can however ask for another path which minimizes the number of turns - the simplest path , of length @xmath4 ( if there are more than one such path we choose the shortest one ) . fig . [
fig : illus]a displays an example of the shortest and simplest path for a given pair of nodes on the oxford ( uk ) street network . to identify the simplest path , we first convert the graph from the primal to the dual representation , where each node corresponds to a straight line in the primal graph .
these straight lines are determined by a continuity negotiation - like algorithm , as described in material and methods .
edges in dual space , in turn , represent the intersection of straight lines in the primal graph ( see fig .
[ fig : illus]b ) .
we define the number of turns @xmath5 of a given path as the number of switches from one straight line to another when walking along this path .
this quantity is intimately related to the amount of information required to move along the path @xcite .
we have computed the probability distribution @xmath6 for all shortest and simplest paths of several networks ( see si ) and the results show that this distribution is usually centered around a smaller values for the simplest paths than for the shortest paths , as expected . more generally , we show in the supplementary information that the average number of turns @xmath7 versus the number of nodes @xmath8 indeed displays a small - world type behavior characterized by a slow logarithmic increase with @xmath8 , consistently with previous analysis of the dual network @xcite .
this feature is thus not very useful to distinguish different networks and shows that the distribution of the number of turns is a very partial information and tells very little about the spatial structure of the simplest paths . for navigation purposes ( neglecting all congestion effects ) and in order to understand the structure of the network , it is useful to compare the lengths of the shortest and the simplest paths with the ratio @xmath9
. it is then natural to introduce the _ simplicity index _ @xmath10 as the average @xmath11 the simplicity index is larger than one and exactly equal to one for a regular square lattice and any tree - like network for example .
large values of @xmath10 indicate that the simplest paths are on average much longer than the shortest ones , and that the network is not easily navigable .
we note here that we do not take into account congestion effects which can influence the path choice ( see for example @xcite ) .
this new metric is a first indication about the spatial structure of simplest paths but mixes various scales , and in order to obtain a more detailed information , we define the _ simplicity profile _
@xmath12 where @xmath13 is the euclidean distance between @xmath2 and @xmath3 and where @xmath14 is the number of pairs of nodes at euclidean distance @xmath15 .
this quantity @xmath16 is larger than one and its variation with @xmath15 informs us about the large scale structure of these graphs .
we can draw a generic shape of this profile : for small @xmath15 , we are the scale of nearest neighbors and there is a large probability that the simplest and shortest paths have the same length , yielding @xmath17 , and increasing for small @xmath15 . for very large @xmath15 , it is almost always beneficial to take long straight lines when they exist , thus reducing the difference between the simplest and the shortest paths . as a result
we expect @xmath16 to decrease when @xmath18 ( note that a similar behavior is observed for another quantity , the route - length efficiency , introduced in @xcite )
. the simplicity profile will then display in general at least one maximum at an intermediate scale @xmath19 for which the length differences between the shortest and the simplest path is maximum .
the length @xmath19 thus represents the typical size of domains not crossed by long straight lines . at this intermediate scale
, the detour needed to find long straight lines for the simplest paths is very large .
we finally note here that these indices are actually not limited to planar networks but to all networks for which the notion of straight lines has a sense and can be computed .
this would be the case for example for spatial networks which are not perfectly planar @xcite .
we introduce a null model in order to provide a simple benchmark to further analyze the results obtained by these new metrics ( the expression null model should be understood here in the sense of the benchmark and not in the usual statistical definition ) . the goal in this study
is to compare empirical results with a very simple model based on a minimal number of assumptions , but we note that it would be also interesting to compare various models generating planar networks .
we start with @xmath8 points randomly distributed in the plane and construct the voronoi graph ( see the supplementary information for further details ) .
we then add a tunable number of straight lines of length @xmath20 distributed according to @xmath21 .
examples of networks generated by this model as well as many results are shown in the si .
we first study static networks ( see fig .
[ fig : all ] ) such as the streets of cities ( bologna , italy ; oxford , uk ; nantes , france ) , the national highway network of australia , the national uk railway system , and the water supply network of central nantes ( france ) . in the case of biological networks , we study the veination patterns of leaves ( _ ilex aquifolium _ and _ hymenanthera chatamica _ ) , and of a dragongly wing .
details on these datasets can be found in the si .
we also consider three datasets describing the time evolution of networks at different scales ( see fig .
[ fig : evolution ] ) : at a small scale and in the biological realm we study the evolution of a slime mould network . at the city scale we present results on the road network of paris ( france ) from 1789 until now .
paris was largely transformed by a central authority ( the prefect haussmann under napoleon iii ) in the middle of the 19th century and the dataset studied here displays the network before and after these important transformations , offering the possibility to study quantitatively the effect of top - down planning @xcite . at the multi - town level , we study the road network of the groane area ( italy ) ( see the si for details on these datasets ) .
these networks allow us to explore different systems at very different scales from @xmath22 ( slime mould ) to @xmath23 ( australian highways ) meters .
we compute the simplicity index @xmath10 for the various datasets and for the null model as well .
the results are shown in fig .
[ fig : svalues ] as a function of the density of straight lines @xmath24 and the gini coefficient @xmath25 for the length of straight lines ( see material and methods for details ) .
the density @xmath24 of straight lines is defined as the ratio of total length of straight lines ( see fig .
si2 ) , over the total length of the network , and @xmath25 is an indicator of the diversity of the length of straight lines . the first observation from fig .
[ fig : svalues ] is that the simplicity index encodes information which is neither contained in the density @xmath24 nor in the gini coefficient @xmath25 , and reveals how the straight lines are distributed in space and participate in the flows on the network . in fig .
[ fig : svalues]a , we observe that the density of straight lines is always larger for urban systems .
more precisely , in the biological systems the density lies in the range @xmath26 $ ] , while we observe @xmath27 for artificial systems .
except for the physarum , which appears to be close to a regular lattice with a small simplicity and small gini coefficient , the simplicity index for the wing and the leaves is larger than the values obtained for the null model .
these results indicate that the organization of straight lines in biological systems is very different from artificial systems , that have very similar values of @xmath24 , @xmath25 , and @xmath10 .
in particular , we observe a hierarchy of straight lines in biological systems ( see fig [ fig : all ] ) : a main artery ( the midrib for leaves ) connects to veins which in turn are connect to smaller veins and so on . in the case of dragonfly wing ,
the main straight line is given by the external border of the network .
the existence of these main straight lines in biological systems will impact the structure of simplest paths and impose some large detour , resulting in a larger value of the simplicity index . for urban systems ,
the simplicity is very close to the null model ( of order @xmath28 in this density range ) , suggesting that in dense urban systems , long straight lines are added at random ( an exception concerns , the pre - haussmannian paris ( 1789 - 1836 ) for which we observe a simplicity smaller than for the null model , the reason being probably that the networks at these times were very sparse ) . as a result ,
navigation on urban systems requires relatively less information with no additional cost : the simplest path is not too different from the shortest path .
finally , we note an interesting effect in the null model in fig .
[ fig : svalues]a which is the existence of a maximum of the simplicity at densities of order @xmath29 . in this density regime , using straight lines implies having to make large detours .
however , when the density exceeds @xmath30 , there are enough straight lines to enable a simplest path which differs not too much from the shortest one .
we now discuss the simplicity profile shown in fig .
[ fig : all ] .
we observe that basically for most of these systems , the simplicity profile displays the generic shape with a maximum at an intermediate scale .
in urban cases , such as bologna and central nantes , we have a typical monocentric system with a dense center and a few important radial straight lines , leading to a simple profile @xmath16 . in the case of oxford and the australian highway network ,
the polycentric organization leads to multiple peaks in the simplicity profile ( fig .
[ fig : all ] ) .
interestingly , we observe that the profiles for australian highways and railways in the uk are very different , despite their similar scale , density @xmath24 , and gini coefficient @xmath25 .
in particular , the uk railway displays small values of the simplicity ( less than @xmath31 ) while for the australian highway network there are many pairs of nodes for which the simplest path is much longer than the shortest one .
we also observe that the profile for both street and water systems of nantes have a very similar shape , pointing to the fact that these networks are strongly correlated .
in addition , the position and the height of the peak ( @xmath32 ) observed for the nantes water system suggests that this distribution system has similar features compared to biological systems such as vein networks in leaves ( see below ) whose function is also distribution .
compared to urban systems , the simplicity profile of biological networks have a single well - defined , and much more pronounced peak .
we observe values of order @xmath33 and @xmath34 for @xmath35 , meaning that for this range of distance , the detour made by the simplest path is very large .
this peak is related to the existence of domains of typical size @xmath19 not crossed by large veins .
we see here a clear effect of the existence of the spatial organization of long straight lines in these systems , probably optimized for the distribution ( of water for leaves ) .
the decay for large @xmath15 is also much faster in the biological case compared to urban systems : this shows that in biological systems there are long straight lines allowing to connect far away nodes .
this is particularly evident on the leaves shown in fig .
[ fig : all ] where we can see the first levels ( primary and secondary ) veins , the rest forming a network . for streets , the organization is much less rigid and the hierarchy less strict : we have a more uniform spatial distribution of straight lines , leading to a smoother decrease of @xmath16 .
going beyond static networks , we apply our new metrics to study the structural changes of time evolving networks .
the first example of a time evolving network is the road network of the groane region which is a @xmath36 area located north of milan @xcite .
we have 7 snapshots of this network for different times from 1833 to 2007 ( see the supplementary information for details ) .
this region evolved without central planning and is thus a good example of an ` organic ' evolution of urban systems . the simplicity profile shown in fig .
[ fig : evolution](a ) allows us to distinguish two different periods . the first period from 1833 to 1955 displays a relatively small simplicity at all scales , while a distinct second regime appears from 1980 until now . in this latter regime ,
the simplicity profile is substantially larger for all scales .
this is an effect of the massive urban densification , leading to a polycentric structure where the readability and the ease to navigate are drastically lowered . at a smaller scale ,
we study the evolution of central paris between 1789 and 1999 .
this dataset provides an interesting case study , as paris experienced larges changes due to haussmann in the middle of the 19th century ( see @xcite for details and more references about this network ) .
this is an opportunity to observe quantitatively the effect of top - down planning : until 1836 , we are in the pre - haussmann paris , while from 1888 until now we are in the post - haussmann period .
the effect of haussmann s central planning is clearly visible on the network shown in fig .
[ fig : evolution](b ) . from 1789 to 1836
, we have a relatively large simplicity at all scales and we observe a decrease in that period at small scales ( @xmath37 ) which corresponds well to the fact that many religious and aristocratic domains and properties were sold and divided in order to create new houses and new roads , improving congestion inside paris . the 1826 - 1836 transition displays a decrease of the simplicity for distance larger than roughly @xmath38 kms ( corresponding to @xmath39 ) indicating that long distance routes were simplified .
it is interesting to note that during this period the eastern part of paris experienced large transformations with the construction of the channel st .
martin .
finally in the period 1836 to 1888 , when paris experienced haussmann s transformation , the simplicity profile is strongly affected : compared to 1836 , the simplicity is improved in the range @xmath40 $ ] , which can be attributed to the construction of large avenues connecting important nodes of the city .
in addition , we observe the surprising effect that at large scales @xmath41 , the simplicity is degraded by haussmann s work : this however could be an artifact of the method and the fact that we considered a portion of paris only and neglected the effect of surroundings .
finally , we note that differences between groane and paris might be explained in terms of a sparse , polycentric urban settlement ( groane ) versus a dense one ( paris ) .
in particular , in the ` urban ' phase for groane ( after 1955 ) , the simplicity profile becomes similar to the one of a dense urban area such as paris .
finally , we show the results in fig . [
fig : evolution](c ) for the _ physarum policephalum _ , a biological system evolving at the centimeter scale .
_ physarum _ is a unicellular multinucleated amoeboid that during its vegetative state takes a complex shape .
its plasmodium viscous body whose goal is to find and connect to food sources , crystallizes in a planar network - like structure of micro - tubes @xcite . in simple terms
, physarum s foraging strategy can be summarized in two phases : i ) the exploration phase in which it grows and reacts to the environment and ii ) the crystallization phase in which it connects to food sources with micro - tubes .
we inoculated active plasmodium over a single food source and observe the micro - tube network at six phases of its growth ( see si for details ) . under these conditions , we observe that the network is statistically isotropic around the food source as shown in fig . [
fig : evolution](c ) and develops essentially radially .
we first observe that the simplicity profile for the physarum is relatively low ( less than @xmath31 ) , suggesting that simplicity could be an important factor in the evolution of this organism .
a closer observation shows that during its evolution , the physarum adds new links to the previous network and also modifies the network on a larger scale , as revealed by the changes of the simplicity profile .
the evolution of the profile is similar to the one obtained for the null model when the density is increased ( see si ) , suggesting that the statistics of straight lines in this case could be described as essentially resulting from the random addition of straight lines of random lengths ( with @xmath42 ) .
we have shown that the new metrics introduced here encode in a useful way both topological and geometrical information about the global structure of planar graphs .
in particular , our results highlight the structural differences between biological and artificial networks . in the former
, we have a clear spatial organization of straight lines , with a clear hierarchy of lines ( midrib , veins , etc ) , leading to simplest paths that require a very small number of turns but at the cost of large detours .
in contrast , there is no such strong spatial organization in urban systems , where the simplicity is usually smaller and comparable to a null model with straight lines of random length and location .
these differences between biological and urban systems might be related to the different functions of these networks : biological networks are mainly distribution networks serving the purpose of providing important fluids and materials . in contrast , the role of road networks is not only to distribute goods but to enable individuals to move from one point of the city to another .
in addition , while biological networks are usually the result of a single process , urban systems are the product of a more complex evolution corresponding to different needs and technologies .
these new metrics also allow us to track important structural changes of these networks .
the simplicity profile thus appears as a useful tool which could provide a quantitative classification of planar graphs and could help in constructing a typology of leaves or street patterns for example . |
understanding the meson mass spectrum in the region below 2 gev is one of the fundamental problems of qcd . while the quark model seems to work very well for many resonances
[ see , for instance , the summary in ref .
@xcite ] , some fundamental questions , such as the constituent quark content of scalar and axial - vector resonances , are still unanswered . the quark content of the scalar mesons has been a matter of debate for many decades @xcite .
the particle data group ( pdg ) @xcite suggests the existence of five @xmath2 states in the region below 1.8 gev : @xmath3 , @xmath4 , @xmath5 , @xmath6 , and @xmath7 . also the existence of a sixth state ,
@xmath8 , close to , but distinct from , @xmath7 has been claimed @xcite .
additionally , there are scalar resonances in the isotriplet and in the isodoublet sector as well : the @xmath9 states @xmath10 and @xmath11 are well established ; the existence of the @xmath12 state @xmath13 ( or @xmath14 ) is confirmed by some @xcite and disputed by other authors @xcite , whereas the @xmath12 scalar kaon , @xmath15 , is an established resonance .
a description of all mentioned scalar states as @xmath1 states is not possible for a simple reason : the number of physical resonances is much larger than the number of resonances that can be constructed within a @xmath1 picture of mesons .
explicitly , there is only one @xmath2 state that can be constructed from the ( non - strange ) @xmath16 and @xmath17 quarks ( provided they are degenerate ) . denoting this state as @xmath18
we obtain @xmath19
. an additional @xmath2 state can be constructed , if the strange quark is considered as well : the pure strange state @xmath20 .
since @xmath18 and @xmath21 have the same @xmath22 quantum numbers , we expect the physical spectrum to consist of mixed , rather than pure , states of @xmath23 and @xmath21 ; however , such mixing will , of course , produce exactly two new states .
these states will correspond to at most two of the mentioned five ( six ) @xmath24 states , and the natural question is then : _ which two _ ?
similarly , in both the @xmath9 and @xmath12 sectors one can construct only one quark - antiquark resonance .
restricting for example to electric charge @xmath25 , one has the states @xmath26 and @xmath27 the state @xmath26 can be assigned to @xmath10 or @xmath11 , and @xmath28 to @xmath13 or @xmath29 the question is : _ which is the correct assignment ? _ an answer to these questions is inevitably complicated for several reasons .
firstly , as already indicated , states with the same @xmath22 quantum numbers are expected to mix this needs to be considered in particular in the scalar sector due to the large number of physical resonances .
secondly , basic features ( pole mass and decay width ) of some scalar mesons [ e.g. @xmath3 ] are notoriously difficult to determine experimentally , which makes it non - trivial to determine the structure of these states [ see , e.g. ref .
@xcite for an example regarding the @xmath3 resonance and ref .
@xcite for an example regarding @xmath5 ] .
the question of how to correctly determine the quark structure of the mesons is not only restricted to the scalar sector .
more recently , also the nature of the axial - vector mesons , most notably @xmath32 , but also that of the isoscalar states @xmath33 , @xmath34 , and the kaonic state @xmath35 [ or @xmath36 , see the discussion in ref . @xcite and refs . therein ] have been investigated .
should one interpret the isotriplet resonance @xmath32 as a quark - antiquark state , as the quark model suggests ( e.g. @xmath37 ) , or is this state a broad @xmath38 molecular state ?
understanding these issues is not only crucial for hadron vacuum spectroscopy but is also important at nonzero temperatures and densities , because the correct identification of the chiral partner of the pion and of the @xmath39 meson is necessary for a proper description of the in - medium properties of hadrons @xcite .
in fact , if the @xmath32 is ( predominantly ) a quark - antiquark state , it is the chiral partner of the @xmath39 meson , with which it becomes degenerate at large @xmath40 and @xmath41 @xcite .
the chiral partner of the pion is the scalar - isoscalar state @xmath42 which has been identified with the lightest scalar state @xmath3 in many works . however , as various theoretical results for the vacuum have shown , and as we shall present in detail in this work , this assignment is not correct : it turns out that @xmath5 emerges as the chiral partner of the pion .
last but not least , the question of the origin of hadron masses is important .
one aims to understand to which extent the quark condensate , @xmath43 , and the gluon condensate , @xmath44 , generate the hadron masses . to give an example , which condensate is predominantly responsible for the mass of the @xmath39 meson ? and related to this , how will the mass of the @xmath39 ( and that of other resonances ) change in the medium ?
the answers to all these fundamental questions are in principle contained in the qcd lagrangian .
unfortunately , qcd can not be solved by analytic means from first principles in the low - energy domain .
for this reason , effective theories have been developed which share some of the underlying symmetries of qcd .
the qcd lagrangian exhibits , in addition to the local @xmath45 color symmetry and the discrete @xmath46 , @xmath47 , and @xmath40 symmetries , a global chiral @xmath48 symmetry which is broken in several ways : spontaneously [ due to the chiral condensate @xmath49 @xcite ] , explicitly ( due to non - vanishing quark masses ) , as well as at the quantum level [ the @xmath50 anomaly @xcite ] . in the framework of effective theories the chiral symmetry of qcd can be realized along two lines : linearly @xcite and non - linearly @xcite . in the former case one
obtains the so - called linear sigma model : both scalar and pseudoscalar degrees of freedom are present and vectors as well as axial - vectors can be included into the model in a straightforward manner @xcite .
note that an important advantage of linear sigma models is the possibility to investigate the state of matter at large values of temperature and chemical potential [ e.g. in the region of the chiral transition , i.e. , where chiral symmetry is restored @xcite ] .
every linear sigma model contains so - called chiral partners states that mix with each other under axial transformations that become degenerate when chiral symmetry is restored . in the non - linear realization ( i.e. , the non - linear sigma models ) , the scalar and the ( axial-)vector states are integrated out and the pseudoscalar states are the only degrees of the freedom : one obtains the lagrangian of chiral perturbation theory @xcite .
if the vector mesons are not integrated out , one is left with chiral perturbation theory with vector mesons , see e.g. ref .
@xcite and refs . therein . in this paper
, we present a linear sigma model containing scalar , pseudoscalar , vector , and axial - vector mesons with both non - strange and strange quantum numbers .
although this project represents a straightforward implementation of the principles of the linear realization of chiral symmetry as already outlined in ref .
@xcite , this is to our knowledge the first time that all these degrees of freedom are considered within a single linear chiral framework . in view of the large number of the fields involved , our model shall be referred to as extended linear sigma model , or elsm .
moreover , we also exploit a
classical symmetry of the qcd lagrangian in the chiral limit : the dilatation symmetry .
this symmetry is broken by quantum effects ( trace anomaly ) and generates , through dimensional transmutation , the qcd low - energy scale @xmath51 . we describe this phenomenon by including a dilaton field in our model .
the associated potential for the dilaton field encodes the trace anomaly by an explicit breaking of dilatation symmetry .
we assume that , except for terms associated with the @xmath52 anomaly and nonzero quark masses , all other interactions are dilatation invariant .
assuming in addition analyticity in the fields ( i.e. , absence of divergences in the lagrangian in the limit of vanishing fields ) , the number of terms appearing in the lagrangian is finite .
the fluctuations of the dilaton field correspond to the glueball degree of freedom . in this work ,
we neglect the coupling of the glueball with the other mesonic degrees of freedom .
formally , this can be justified by taking the large-@xmath53 limit . in a future work
, we also plan to study the coupling of the glueball to mesons in our framework , similar to the @xmath54 study of ref .
@xcite . in refs .
@xcite , we have already presented a linear sigma model with vector and axial - vector mesons for two flavors . comparing our model with experimental data for meson vacuum phenomenology led us to conclude that the scalar @xmath1 states are located in the energy region above 1 gev in the hadron mass spectrum .
explicitly , we concluded that the resonances @xmath5 and @xmath11 are strongly favoured to be scalar quarkonia .
the present work is more general : we now consider three flavors both in the ( pseudo)scalar and ( axial-)vector channels in order to ascertain whether the conclusion of ref .
@xcite is still valid once strange mesons are considered .
we emphasize that , as discussed in ref .
@xcite , all fields entering our model describe pure quark - antiquark states .
the reason is that masses and decay widths of our theoretical states scale as @xmath55 and @xmath56 , respectively , where @xmath57 is the number of colors ; thus all decay widths vanish in the large-@xmath57 limit .
note that the inclusion of strange mesons provides us with a large number of very precise data [ in general , decisively more precise than in the case of non - strange mesons @xcite ] . as a consequence ,
our model parameters are much more constrained than in ref .
preliminary results of this work have already been presented in ref .
@xcite .
in order to ( _ i _ ) test the performance of a linear sigma model in describing the overall phenomenology of ( pseudo)scalar and ( axial-)vector mesons and ( _ ii _ ) ascertain which scalar states are predominantly quarkonia , we perform a global fit in which 21 experimentally known quantities ( decay constants , masses , and decay widths as well as amplitudes ) are included .
the situation in the scalar - isoscalar sector is extremely uncertain because five ( six ) resonances , @xmath3 , @xmath4 , @xmath5 , @xmath6 , and @xmath7 [ and , possibly in the future , @xmath8 ] are listed in the pdg .
in addition , the @xmath3 and @xmath5 decay widths and branching ratios are poorly known and @xmath4 suffers from a distortion by the @xmath59 threshold .
moreover , a scalar glueball state with a bare mass of about @xmath60 gev as predicted by lattice qcd @xcite can sizably affect the masses and the branching ratios of the scalar states above 1 gev , that is @xmath5 , @xmath6 , and @xmath7 @xcite . in view of all these reasons , we do _ not _ include the scalar - isoscalar states into the fit .
we note that , although the coupling of the glueball state to the other mesons is not considered in this paper , it does not affect our global fit .
we do , however , include the isotriplet and isodoublet quark - antiquark scalar states and we test all four combinations @xmath61 , @xmath62 , @xmath63 , and @xmath64 .
quite remarkably , the outcome of the fit is univocal : only the pair @xmath61 yields a good fit , while the other combinations do not .
we thus conclude that the @xmath9 and @xmath12 quark - antiquark scalar resonances lie above 1 gev .
in fact , the quality of our fit is surprisingly good .
we describe _ all _ experimental quantities with an average error of 5% , and most of them even to much better precision .
we perceive this to be a remarkable achievement within an ( in principle , quite simple ) effective model for the strong interaction .
we then study the two scalar - isoscalar quark - antiquark states in our model : a consequence of our fit is their mass in the large-@xmath57 limit .
these masses turn out to be about @xmath65 gev and @xmath66 gev , respectively . varying large-@xmath57 suppressed parameters which can not be determined by our fit
, one can also study their decays : the non - strange quark - antiquark state is well described by @xmath5 and the heavier @xmath67 state might be @xmath68 turning to the axial - vector channel , both masses and decays are well described by assuming the axial - vector resonances @xmath32 , @xmath33 , @xmath34 , and @xmath35 as ( predominantly ) quark - antiquark states .
detailed calculations of all formulas used in this paper as well as further discussion are presented in ref .
@xcite .
the outline of the paper is as follows . in sec .
[ model ] we discuss the @xmath69 linear sigma model with vector and axial - vector mesons . in sec .
[ fitprocedure ] we discuss the global fit and its consequences and in sec . [ sec : conclusion ] we present our conclusions .
we defer detailed calculations to appendices [ app : mass ] and [ app : decay ] .
our units are @xmath70 ; the metric tensor is @xmath71 .
in this section we present our model : a linear sigma model with ( axial-)vector mesons and global chiral @xmath69 symmetry . to this end , we discuss some important criteria for the construction of its lagrangian.the aim of our work is to emulate the properties of the qcd lagrangian in our effective approach .
this implies the necessity to consider as many symmetries of qcd as possible .
the qcd lagrangian possesses various symmetries : local ( gauge ) invariance with respect to the color group @xmath72 , discrete @xmath46 , @xmath47 , and @xmath40 symmetries , global chiral symmetry @xmath73 ( which is exact in the chiral limit ) , and also the classical dilatation ( scale ) symmetry .
the local color symmetry is automatically fulfilled when working with colorless hadronic degrees of freedom .
the discrete and chiral symmetries impose severe constraints on the terms which are kept in the lagrangian ; still , infinitely many are allowed .
indeed , in some older versions of the linear sigma model , terms with dimension larger than four were considered , see for example refs .
@xcite . in these approaches
chiral symmetry was promoted to a _ local symmetry _
, up to the vector meson mass term which breaks this symmetry explicitly . sigma models with local chiral symmetry require the inclusion of terms of order larger than four in the fields in order to correctly describe experimental data .
this procedure obviously breaks renormalizability , but this is not an issue because a hadronic theory is obviously not fundamental and is supposed to be valid only up to a mass scale of @xmath74@xmath75 gev .
still , the problem of constraining the number of terms affects these effective approaches of qcd .
we then turn our attention to the last of the mentioned symmetries of qcd : the dilatation symmetry .
it plays a key role to justify why we retain only a _ finite _ number of terms @xcite .
let us recall some of the basic features of the dilatation symmetry in the pure gauge sector of qcd : the yang - mills ( ym ) lagrangian ( qcd without quarks ) is classically invariant under dilatations , but this symmetry is broken at the quantum level .
the divergence of the corresponding current is the trace of the energy - momentum tensor @xmath76 of the ym lagrangian @xmath77 , where @xmath78 is the gluon field - strength tensor and @xmath79 the beta function of ym theory . at the composite level
, one can parametrize this situation by introducing the dilaton field @xmath80 , which is described by the lagrangian @xcite @xmath81 the dilatation symmetry is explicitly broken by the scale @xmath82 under the logarithm .
the minimum of the dilaton potential is realized for @xmath83 ; upon shifting @xmath84 a particle with mass @xmath85 emerges , which is interpreted as the scalar glueball . the numerical value has been evaluated in lattice qcd and reads @xmath86 gev @xcite . in the large-@xmath53 limit , @xmath87 ,
while the @xmath88-glueball vertex @xmath89 @xcite .
applying this to eq .
( [ ldil ] ) , we observe that @xmath90 . in the large-@xmath53 limit ,
the glueball self - interaction terms vanishes and the glueball becomes a free field .
we are now ready to present our lagrangian .
it follows from requiring symmetry under @xmath91 , ( global ) chiral @xcite , as well as dilatation transformations . in accordance with qcd
, we take the latter two symmetries to be explicitly broken only by nonzero quark masses , and dilatation symmetry to be explicitly broken only by the dilaton potential in eq . ( [ ldil ] ) as well as the @xmath92 anomaly .
therefore , all other terms in the lagrangian must be dilatation invariant and thus carry mass dimension equal to four .
this would in principle still allow for an infinite number of terms .
however , assuming in addition that there are no terms with non - analytic powers of the field variables makes the number of possible terms finite .
explicitly , the lagrangian of the model has the form @xmath93-m_{0}^{2}\left ( \frac{g}{g_{0}}\right ) ^{2}\mathop{\mathrm{tr}}(\phi^{\dagger}\phi ) -\lambda_{1}[\mathop{\mathrm{tr}}(\phi^{\dagger}\phi)]^{2}-\lambda _ { 2}\mathop{\mathrm{tr}}(\phi^{\dagger}\phi)^{2}{\nonumber}\\ & -\frac{1}{4}\mathop{\mathrm{tr}}(l_{\mu\nu}^{2}+r_{\mu\nu}^{2})+\mathop{\mathrm{tr}}\left [ \left ( \left ( \frac{g}{g_{0}}\right ) ^{2}\frac{m_{1}^{2}}{2}+\delta\right ) ( l_{\mu}^{2}+r_{\mu}^{2})\right ] + \mathop{\mathrm{tr}}[h(\phi+\phi^{\dagger})]{\nonumber}\\ & + c_{1}(\det\phi-\det\phi^{\dagger})^{2}+i\frac{g_{2}}{2}(\mathop{\mathrm{tr}}\{l_{\mu\nu}[l^{\mu},l^{\nu}]\}+\mathop{\mathrm{tr}}\{r_{\mu\nu}[r^{\mu},r^{\nu}]\}){\nonumber}\\ & + \frac{h_{1}}{2}\mathop{\mathrm{tr}}(\phi^{\dagger}\phi ) \mathop{\mathrm{tr}}(l_{\mu}^{2}+r_{\mu}^{2})+h_{2}\mathop{\mathrm{tr}}[\vert l_{\mu}\phi \vert ^{2}+\vert \phi r_{\mu } \vert ^{2}]+2h_{3}\mathop{\mathrm{tr}}(l_{\mu}\phi r^{\mu}\phi^{\dagger}){\nonumber}\\ & + g_{3}[\mathop{\mathrm{tr}}(l_{\mu}l_{\nu}l^{\mu}l^{\nu})+\mathop{\mathrm{tr}}(r_{\mu}r_{\nu}r^{\mu}r^{\nu})]+g_{4}[\mathop{\mathrm{tr}}\left ( l_{\mu}l^{\mu}l_{\nu}l^{\nu}\right ) + \mathop{\mathrm{tr}}\left ( r_{\mu}r^{\mu}r_{\nu}r^{\nu}\right ) ] { \nonumber}\\ & + g_{5}\mathop{\mathrm{tr}}\left ( l_{\mu}l^{\mu}\right ) \,\mathop{\mathrm{tr}}\left ( r_{\nu}r^{\nu}\right ) + g_{6}[\mathop{\mathrm{tr}}(l_{\mu}l^{\mu})\,\mathop{\mathrm{tr}}(l_{\nu}l^{\nu } ) + \mathop{\mathrm{tr}}(r_{\mu}r^{\mu})\,\mathop{\mathrm{tr}}(r_{\nu}r^{\nu } ) ] \text { , } \label{eq : lagrangian}\ ] ] where @xmath94 is the dilaton term ( [ ldil ] ) and @xmath95\;,\nonumber\\ l^{\mu\nu } & \equiv\partial^{\mu}l^{\nu}-iea^{\mu}[t_{3},l^{\nu}]-\left\ { \partial^{\nu}l^{\mu}-iea^{\nu}[t_{3},l^{\mu}]\right\ } \;\text{,}\nonumber\\ r^{\mu\nu } & \equiv\partial^{\mu}r^{\nu}-iea^{\mu}[t_{3},r^{\nu}]-\left\ { \partial^{\nu}r^{\mu}-iea^{\nu}[t_{3},r^{\mu}]\right\ } \;\text{.}\nonumber\end{aligned}\ ] ] the quantities @xmath96 , @xmath97 , and @xmath98 represent the scalar and vector nonets : @xmath99{ccc}\frac{(\sigma_{n}+a_{0}^{0})+i(\eta_{n}+\pi^{0})}{\sqrt{2 } } & a_{0}^{+}+i\pi^{+ } & k_{0}^{\star+}+ik^{+}\\ a_{0}^{-}+i\pi^{- } & \frac{(\sigma_{n}-a_{0}^{0})+i(\eta_{n}-\pi^{0})}{\sqrt{2 } } & k_{0}^{\star0}+ik^{0}\\ k_{0}^{\star-}+ik^{- } & { \bar{k}_{0}^{\star0}}+i{\bar{k}^{0 } } & \sigma _ { s}+i\eta_{s}\end{array }
\right ) \text { , } \label{eq : matrix_field_phi}\\ l^{\mu } & = \sum_{i=0}^{8}(v_{i}^{\mu}+a_{i}^{\mu})t_{i}=\frac{1}{\sqrt{2}}\left ( \begin{array } [ c]{ccc}\frac{\omega_{n}+\rho^{0}}{\sqrt{2}}+\frac{f_{1n}+a_{1}^{0}}{\sqrt{2 } } & \rho^{+}+a_{1}^{+ } & k^{\star+}+k_{1}^{+}\\ \rho^{-}+a_{1}^{- } & \frac{\omega_{n}-\rho^{0}}{\sqrt{2}}+\frac{f_{1n}-a_{1}^{0}}{\sqrt{2 } } & k^{\star0}+k_{1}^{0}\\ k^{\star-}+k_{1}^{- } & { \bar{k}}^{\star0}+{\bar{k}}_{1}^{0 } & \omega _
{ s}+f_{1s}\end{array } \right ) ^{\mu}\text{,}\label{eq : matrix_field_l}\\ r^{\mu } & = \sum_{i=0}^{8}(v_{i}^{\mu}-a_{i}^{\mu})t_{i}=\frac{1}{\sqrt{2}}\left ( \begin{array } [ c]{ccc}\frac{\omega_{n}+\rho^{0}}{\sqrt{2}}-\frac{f_{1n}+a_{1}^{0}}{\sqrt{2 } } & \rho^{+}-a_{1}^{+ } & k^{\star+}-k_{1}^{+}\\ \rho^{-}-a_{1}^{- } & \frac{\omega_{n}-\rho^{0}}{\sqrt{2}}-\frac{f_{1n}-a_{1}^{0}}{\sqrt{2 } } & k^{\star0}-k_{1}^{0}\\ k^{\star-}-k_{1}^{- } & { \bar{k}}^{\star0}-{\bar{k}}_{1}^{0 } & \omega _ { s}-f_{1s}\end{array } \right ) ^{\mu}\text { , } \label{eq : matrix_field_r}\ ] ] where the assignment to physical particles is shown as well sector where particle mixing takes place ( see below ) . ] . here ,
@xmath100 denote the generators of @xmath101 , while @xmath102 represents the scalar , @xmath103 the pseudoscalar , @xmath104 the vector , @xmath105 the axial - vector meson fields , and @xmath106 is the electromagnetic field .
it should be noted that here and below we use the so - called non strange strange basis in the @xmath107 sector , defined as @xmath108 which is more suitable for our calculations .
moreover , @xmath109 and @xmath110 are constant external fields defined as @xmath111{ccc}\frac{h_{0n}}{2 } & 0 & 0\\ 0 & \frac{h_{0n}}{2 } & 0\\ 0 & 0 & \frac{h_{0s}}{\sqrt{2}}\end{array } \right ) \;,\label{eq : expl_sym_br_epsilon}\\ \delta & = \delta_{0}t_{0}+\delta_{8}t_{8}=\left ( \begin{array } [ c]{ccc}\frac{\tilde{\delta}_{n}}{2 } & 0 & 0\\ 0 & \frac{\tilde{\delta}_{n}}{2 } & 0\\ 0 & 0 & \frac{\tilde{\delta}_{s}}{\sqrt{2}}\end{array } \right ) \equiv\left ( \begin{array } [ c]{ccc}\delta_{n } & 0 & 0\\ 0 & \delta_{n } & 0\\ 0 & 0 & \delta_{s}\end{array } \right ) \text { . } \label{eq : expl_sym_br_delta}\ ] ] these terms describe the effect of nonzero quark masses in the ( pseudo)scalar and ( axial-)vector sectors , respectively : @xmath112 , @xmath113 , @xmath114 , @xmath115 . throughout this work
we assume exact isospin symmetry for @xmath16 and @xmath17 quarks , such that the first two diagonal elements in eq . and
eq . are identical .
thus , only the scalar - isoscalar fields @xmath116 and @xmath80 , carrying the same quantum numbers as the vacuum , can have nonzero vacuum expectation values ( vev s ) could have a nonzero vev . ] . in the lagrangian ( [ eq : lagrangian ] ) the following terms break the original @xmath69 [ @xmath117 symmetry : _ ( i ) _ the terms proportional to the matrix @xmath109 and @xmath110 of eqs . and ( [ eq : expl_sym_br_delta ] ) , which describes the explicit symmetry breaking due to the nonzero values of the quark masses in the ( pseudo)scalar and ( axial-)vector sectors and break @xmath118 if @xmath119 and @xmath120 if @xmath121 [ for more details see , e.g. ref
. @xcite ] , and _
( ii ) _ the term proportional to the determinant and parametrized by @xmath122 , which breaks the @xmath50 symmetry and describes the axial anomaly , responsible for large mass of the @xmath123 meson .
these terms also explicitly break dilatation symmetry because they involve dimensionful coupling constants .
this is expected : the first two terms describe the bare quark masses which generate an explicit breaking of dilatation symmetry at the level of the qcd lagrangian , and the determinant term describes an anomalous breaking of the dilatation symmetry arising from the ym sector of the theory .
the interaction of the meson fields with the dilaton field @xmath80 enters only in two terms . upon condensation of the dilaton field
, these terms correspond to meson mass terms .
in addition , there are interaction terms proportional to one or two powers of the glueball field . since @xmath124 , an @xmath125-point vertex involving @xmath126 meson lines and @xmath127 glueball lines scales as @xmath128 , while an @xmath126-point vertex involving @xmath126 meson lines scales as @xmath129 .
thus , vertices with @xmath88 external lines involving glueballs vanish faster than the corresponding @xmath88-point vertices involving only mesons . as a first approximation , we assume that the glueball field completely decouples , so that we neglect it in the following .
effects from coupling the glueball to the other meson fields will be studied in a subsequent work .
let us now discuss in detail the assignment of fields in eqs .
( [ eq : matrix_field_phi ] ) ( [ eq : matrix_field_r ] ) .
if we consider isospin multiplets as single degrees of freedom , then there are 16 resonances that can be described by the model : @xmath18 , @xmath21 , @xmath130 , @xmath131 ( scalar ) ; @xmath132 , @xmath133 , @xmath134 , @xmath135 ( pseudoscalar ) ; @xmath136 , @xmath137 , @xmath138 , @xmath139 ( vector ) , and @xmath140 , @xmath141 , @xmath142 , @xmath143 ( axial - vector ) .
all fields in our model represent @xmath1 states , as discussed in ref . @xcite .
if we assign a state from our model to a physical resonance we , therefore , implicitly assume that this resonance is a @xmath144 state .
this assumption can be tested for a multitude of physical resonances in the scalar and axial - vector sectors , as discussed below ( in the pseudoscalar and the vector channels , there are no ambiguities ) . in the non - strange sector
, we assign the fields @xmath134 and @xmath132 to the pion and the non - strange part of the @xmath145 and @xmath123 mesons , @xmath146 .
the fields @xmath136 and @xmath138 represent the @xmath147 and @xmath148 vector mesons , respectively , and the fields @xmath140 and @xmath142 represent the @xmath33 and @xmath32 mesons , respectively . in the strange sector , we assign the @xmath135 fields to the kaons ; the @xmath133 field is the strange contribution to the physical @xmath145 and @xmath123 fields [ @xmath149 ; the @xmath150 , @xmath151 , @xmath152 , and @xmath143 fields correspond to the @xmath153 , @xmath154 , @xmath155 , and @xmath35 [ or @xmath36 ] mesons , respectively .
unfortunately , the assignment of the scalar fields is substantially less clear .
experimental data suggest existence of five ( six ) scalar - isoscalar states below @xmath157 gev : @xmath3 , @xmath4 , @xmath5 , @xmath6 , and @xmath7 [ as well as @xmath8 ] .
note that the existence of these five states is acknowledged by the particle data group
pdg @xcite . [ the sixth state , @xmath8 , will not be of importance for the rest of our work because its predominant @xmath158 decay renders it a putative radial excitation of @xmath5 , and our model describes ground - state quarkonia only . ]
our model contains a pure non - strange isoscalar @xmath18 and a pure strange isoscalar @xmath21
. we will demonstrate below that our model yields mixing of @xmath18 and @xmath21 , producing a predominantly non - strange state labeled as @xmath159 , and a predominantly strange state labeled as @xmath160 .
assignment of the mixed states to physical resonances is ambiguous because , as already discussed , there are five physical states all of which could , in principle , be candidates for @xmath159 and @xmath160 .
similarly , the isospin triplet @xmath161 can be assigned to different physical resonances
although , in this case , there are only two candidate states : @xmath10 and @xmath11 .
a preliminary examination of the assignment of the @xmath161 field has been performed in ref .
@xcite where it was concluded that it most likely corresponds to the @xmath11 resonance [ or , equivalently , @xmath11 rather than @xmath10 was favored to represent a @xmath144 state ] .
the discussion in ref .
@xcite was limited to non - strange mesons . in this work , besides the assignment of @xmath161 , we also have to consider possible assignments for the strange scalar field @xmath131 ; there are also two possibilities : either @xmath13 or @xmath15 . after spontaneous symmetry breaking
, the fields with nonzero vev s are shifted by their expectation values , namely , @xmath162 , where we have introduced @xmath163 . after substituting the shifted fields into the lagrangian , one obtains the tree - level masses by selecting all terms quadratic in the fields , @xmath164 s_{j}\nonumber\\ & -\frac{1}{2}p_{i}\left [ \delta_{ij}\ , \box+(m_{p}^{2})_{ij}\right ] p_{j}\nonumber\\ & -\frac{1}{2}v_{i\mu}\left [ ( -g^{\mu\nu}\,\box+\partial^{\mu}\partial^{\nu } ) \delta_{ij}-g^{\mu\nu}(m_{v}^{2})_{ij}\right ] v_{j\nu}\nonumber\\ & -\frac{1}{2}a_{i\mu}\left [ ( -g^{\mu\nu}\,\box+\partial^{\mu}\partial^{\nu } ) \delta_{ij}-g^{\mu\nu}(m_{a}^{2})_{ij}\right ] a_{j\nu}\nonumber\\ & -\frac{1}{2}v_{i\mu}\left ( ig_{1}f_{ijk}\phi_{k}\partial^{\mu}\right ) s_{j}-\frac{1}{2}s_{i}\left ( ig_{1}f_{ijk}\phi_{k}\partial^{\nu}\right ) v_{j\nu}\nonumber\\ & + \frac{1}{2}a_{i\mu}\left ( g_{1}d_{ijk}\phi_{k}\partial^{\mu}\right ) p_{j}-\frac{1}{2}p_{i}\left ( g_{1}d_{ijk}\phi_{k}\partial^{\nu}\right ) a_{j\nu } , \label{eq : quad}\ ] ] where @xmath165 , and @xmath166 are the scalar , pseudoscalar , vector , and axial - vector ( squared ) mass matrices , respectively , see their explicit expressions in appendix [ app : mass ]
. moreover , @xmath167 and @xmath168 are the antisymmetric and symmetric structure constants of @xmath101 .
the ( squared ) mass matrices are in general non - diagonal due to the mixing among particles sitting in the center of a given nonet , and they can be diagonalized by appropriate orthogonal transformations ( for details see appendix [ app : mass ] ) .
besides the mixing inside the nonets there are other mixing terms , namely the last four terms of eq . , which mix different nonets .
in order to eliminate the latter , one performs the following shifts of the ( axial-)vector fields : @xmath169 these shifts produce additional kinetic terms for the pseudoscalar fields . in order to retain the canonical normalization for the latter
, one has to introduce wavefunction renormalization constants , @xmath170 for the sake of simplicity we have grouped together the isotriplet states with the notation @xmath171 and the isodoublet states with the notation @xmath172 , where @xmath173 refers to @xmath174 .
the coefficients @xmath175 and @xmath176 are determined in order to eliminate the last four mixing terms in eq .
( [ eq : quad ] ) and to obtain the canonical normalization of the @xmath177 , @xmath132 , @xmath133 , @xmath135 , and @xmath131 fields .
after some straightforward calculation one finds the explicit expressions : @xmath178@xmath179 it can be seen from the expressions of the wavefunction renormalization constants that they are always larger than one . finally , using the explicit expressions found in appendix [ app : mass ] the tree - level ( squared ) masses for the different nonets
are obtained as follows : @xmath180 \equiv\frac{z_{\pi}^{2}h_{0n}}{\phi_{n}}\text { , } \label{m_pi}\\ m_{k}^{2 } & = z_{k}^{2}\left [ m_{0}^{2}+\left ( \lambda_{1}+\frac { \lambda_{2}}{2}\right ) \phi_{n}^{2}-\frac{\lambda_{2}}{\sqrt{2}}\phi_{n}\phi_{s}+\left ( \lambda_{1}+\lambda_{2}\right ) \phi_{s}^{2}\right ] \text { , } \label{mkaon}\\ m_{\eta_{n}}^{2 } & = z_{\pi}^{2}\left [ m_{0}^{2}+\left ( \lambda_{1}+\frac{\lambda_{2}}{2}\right ) \phi_{n}^{2}+\lambda_{1}\phi_{s}^{2}+c_{1}\,\phi_{n}^{2}\phi_{s}^{2}\right ] \equiv z_{\pi}^{2 } \left ( \frac{h_{0n}}{\phi_{n}}+c_{1}\,\phi_{n}^{2}\phi_{s}^{2}\right ) \;,\label{eq : etan}\\ m_{\eta_{s}}^{2 } & = z_{\eta_{s}}^{2}\left [ m_{0}^{2}+\lambda_{1}\phi _ { n}^{2}+\left ( \lambda_{1}+\lambda_{2}\right ) \phi_{s}^{2}+\frac{c_{1}}{4}\phi_{n}^{4}\right ] \equiv z_{\eta_{s}}^{2 } \left ( \frac{h_{0s}}{\phi _ { s}}+\frac{c_{1}}{4}\phi_{n}^{4}\right ) \text { , } \label{eq : etas}\\ m_{\eta_{ns}}^{2 } & = z_{\pi}z_{\pi_{s}}\frac{c_{1}}{2}\phi_{n}^{3}\phi _ { s}\text { , } \label{eq : etans}\ ] ] are the ( squared ) pseudoscalar masses , while @xmath181 \text { , } \\
m_{\sigma_{n}}^{2 } & = m_{0}^{2}+3\left ( \lambda_{1}+\frac{\lambda_{2}}{2}\right ) \phi_{n}^{2}+\lambda_{1}\phi_{s}^{2}\text { , } \label{eq : sign}\\ m_{\sigma_{s}}^{2 } & = m_{0}^{2}+\lambda_{1}\phi_{n}^{2}+3\left ( \lambda _ { 1}+\lambda_{2}\right ) \phi_{s}^{2}\text { , } \label{eq : sigs}\\ m_{\sigma_{ns}}^{2 } & = 2\lambda_{1}\phi_{n}\phi_{s}\text { , } \label{eq : signs}\ ] ] are the ( squared ) scalar masses .
the quantities @xmath182 and @xmath183 are mixing terms in the non - strange strange sector .
these mixings can be removed by orthogonal transformations , and the resulting mass eigenstates are found to be @xmath184 \;,\label{m_f0lh}\\ m_{\eta^{\prime}/\eta}^{2 } & = \frac{1}{2}\left [ m_{\eta_{n}}^{2}+m_{\eta_{s}}^{2}\pm\sqrt{(m_{\eta_{n}}^{2}-m_{\eta_{s}}^{2})^{2}+4m_{\eta_{ns}}^{4}}\right ] \;.\end{aligned}\ ] ] moreover , the ( squared ) vector masses are given by @xmath185 while the ( squared ) axial - vector meson masses are @xmath186 it is interesting to note that in case of vectors and axial - vectors there are no mixings in the non - strange strange sector .
the lagrangian ( [ eq : lagrangian ] ) contains 18 parameters ( as mentioned above , we neglect the coupling of the glueball with the other mesons in the present work ) : @xmath187 we make the following observations regarding the model parameters : * the parameters @xmath188 and @xmath189 model the explicit breaking of the chiral symmetry ( esb ) in the ( pseudo)scalar sector via the term @xmath190 $ ] ; they are uniquely determined from the mass terms of the pion , eq .
( [ m_pi ] ) , and of @xmath133 , the strange part of the @xmath145 meson [ see eq .
( [ eq : etas ] ) ] , implying that the masses of @xmath191 and @xmath133 are generated by esb . * the parameters @xmath192 and @xmath193 model the explicit symmetry breaking in the vector and axial - vector channels .
the esb arises from non - vanishing quark masses and therefore we employ the correspondence @xmath194 and @xmath195 .
however , in the vector - meson mass term @xmath196 $ ] only the linear combinations @xmath197 appear . therefore , it is possible to redefine @xmath198 .
then , only the combination @xmath199 appears in the mass formulas .
only this difference will be determined by the fit of the ( axial-)vector masses .
alternately , we may set @xmath200 from the beginning , and determine @xmath193 from the fit .
* the parameters @xmath201 , @xmath202 , @xmath203 , and @xmath204 do not influence any of the decays to be discussed in this work and are therefore not considered in the fit .
consequently , we are left with the following 13 parameters : @xmath205 their large-@xmath57 dependence reads @xcite:@xmath206 we recall that a non - suppressed @xmath88-meson interaction vertex scales as @xmath207 @xcite . in this respect
the parameters @xmath208 and @xmath209 are suppressed in the large-@xmath57 limit : in fact , they scale as @xmath210 and not as @xmath56 .
similarly , the axial - anomaly parameter @xmath122 scales as @xmath211 and not as @xmath210 .
note that the large-@xmath57 behaviour of the model parameters ( [ ncdep ] ) implies that the decay widths ( the formulas for which are given in appendix [ app : decay ] ) decrease with increasing @xmath57 .
for this reason , as already mentioned , the states in our model represent @xmath144 states .
let us now turn to the fit procedure for the parameters discussed above .
our fit aims to ascertain whether ( _ i _ ) it is possible to find a fit containing masses and decay widths for ( pseudo)scalar and ( axial-)vector degrees of freedom present in our model and ( _ ii _ ) which physical scalar states are best described by the @xmath144 states of our model .
all decays considered are two - particle decays , thus they can be calculated from the appropriate tree - level terms of the lagrangian after applying the necessary field shifts and orthogonal transformations . if the decaying particle is denoted by @xmath212 and the decay products by @xmath213 and @xmath46 , respectively , the decay width reads @xmath214 where @xmath215 is the three - momentum of one of the resulting particles in the rest frame of @xmath212 and @xmath216 is the transition matrix element ( decay amplitude ) .
moreover , @xmath217 refers to the so - called isospin factor which is the number of sub - channels in a given decay channel ( for instance , if @xmath218 the @xmath219 decay can have two sub - channels , namely @xmath220 and @xmath221 , which results in @xmath222 ) .
equation ( [ gamma ] ) will be used to calculate all decay widths entering the fit ; details of the calculations can be found in appendix [ app : decay ] .
moreover , when identical particles emerge in the final state , the usual symmetry factors are included .
a brief remark is necessary regarding the errors used in the fit .
the fit will contain input information from experimental data regarding both ( axial-)vector and ( pseudo)scalar states .
the data are very precise for some of the resonances described by our model .
for example , the mass of the @xmath153 resonance ( our @xmath150 state ) is known with @xmath223 accuracy .
our model does not aim to describe hadron vacuum phenomenology with this extreme precision .
the reason is simple : already isospin - breaking effects in the physical hadron mass spectrum are of the order of 5% [ for instance the difference between the charged and neutral pion masses , or the masses of the @xmath32 and the @xmath33 ] , but are completely neglected in our model .
we therefore artificially increase the experimental errors to 5% , if the actual error is smaller , or we use the experimental values , if the error is larger than 5% .
let us now discuss the input information for our fit .
we do this separately for mesons of different spin [ central values from pdg @xcite unless otherwise stated , errors according to the above discussion ] : * _ weak decay constants : _ we use @xmath225 the following formulas relate the decay constants to the vacuum condensates : @xmath226 and @xmath227 @xmath228 * _ pseudoscalar mesons : _ we use @xmath229 with the following notes : ( _ i _ ) @xmath230 and @xmath231 represent isospin - averaged values ; ( _ ii _ ) the relatively large error values come from the criterion @xmath232(5% , experimental error ) , discussed above . * _ vector mesons : _ we use @xmath233 with the following notes : ( _ i _ ) we use the isospin - averaged value for @xmath234 ; ( _ ii _ ) in case of the @xmath235 decay we use the physical mass values in the kinematic factor @xmath236 due to the proximity of @xmath237 to the @xmath238 threshold ( in order to eliminate phase - space effects ) . * _ axial - vector mesons : _ we use @xmath239 with the following notes : ( _ i _ ) the value of @xmath240 is not precisely known ; there are experimental data suggesting this decay channel to be dominant for @xmath241 @xcite and thus we estimate the possible range for @xmath242 from the interval for the full @xmath241 decay width [ @xmath243 mev ] ; ( _ ii _ ) according to pdg the channel @xmath244 is dominant within the channel @xmath245 , with the latter being the overall dominant decay mode for @xmath154 ; we have assumed @xmath246 to be equal to the full decay width @xmath247 mev and determined @xmath248 using an averaged branching ratio @xmath249 from refs .
@xcite and @xcite . * _ isotriplet and isodoublet scalar mesons : _ the observables from eqs .
( [ pi ] ) ( [ avi ] ) will be used with any of the @xmath161 -@xmath131 combinations [ @xmath10/@xmath13 , @xmath250/@xmath15 , @xmath11/@xmath13 , @xmath11/@xmath15 ] , where the data to be used are as follows : @xmath252 we note the following : ( _ i _ ) the interpretation of @xmath13 as a particle is controversial ; we will nonetheless include it into our fits in accordance with the conclusions of ref .
@xcite ; ( _ ii _ ) as in the case of @xmath153 , we will use the decay amplitudes rather than the decay widths for the processes @xmath253 and @xmath254 due to the proximity of @xmath10 to the @xmath255 threshold .
the reader may find it somewhat surprising that we are considering pairs of states above [ @xmath11/@xmath15 ] 1 gev , below [ @xmath256/@xmath13 ] 1 gev , as well as `` mixed '' pairs @xmath256/@xmath15 , @xmath11/@xmath13 .
the reason to consider also mixed pairs is that we want to avoid any kind of prejudice in the assignment of our @xmath1 states , and thus explore all possibilities .
nonetheless , these mixed pairs as well as the pair below 1 gev will be disfavored by our analysis ( see below ) .
mass formulas used in the fit are presented in eqs .
( [ m_pi ] ) ( [ eq : etas ] ) , ( [ m_a_0 ] ) ( [ eq : sigs ] ) , and ( [ m_f0lh ] ) ( [ m_f1_s ] ) , whereas formulas for the decay widths are given in appendix [ app : decay ]
. moreover , the fit shall be constrained by the following conditions : * @xmath258 , @xmath259 , @xmath260 , @xmath261 , due to eqs .
( [ z_pi ] ) ( [ z_k_s ] ) . *
@xmath262 and @xmath263 , i.e. , pure non - strange states should be lighter than pure strange states .
* @xmath264 .
this is a necessary condition for the spontaneous breaking of chiral symmetry . *
@xmath265 and @xmath266 .
this is necessary for the potential in the lagrangian ( [ eq : lagrangian ] ) to be bounded from below .
* @xmath267 .
the reason is that otherwise ( _ i _ ) an instability of the vacuum in the physical @xmath39 direction would occur [ see the lagrangian ( [ eq : lagrangian ] ) ] and ( _ ii _ ) @xmath268 and @xmath269 would become imaginary in the chiral transition , i.e. , once the condensates vanish [ see eqs .
( [ m_rho ] ) and ( [ m_a_1 ] ) ] . *
@xmath270 as otherwise @xmath271 in the @xmath39 mass term ( [ m_rho ] ) would be negative ; this would imply that spontaneous chiral symmetry breaking decreases the @xmath39 mass .
this is clearly unnatural because the breaking of chiral symmetry generates a sizable constituent mass for the light quarks , which is expected to positively contribute to the meson masses .
this positive contribution is a feature of all known models ( such as the nambu jona - lasinio model and constituent quark approaches ) .
indeed , in an important class of hadronic models [ see ref .
@xcite and refs . therein ]
the only and obviously positive contribution to the @xmath39 mass squared is proportional to @xmath272 ( i.e. , @xmath273 ) . before discussing the results of the fit ,
it is important to stress which particles have not been included in the list above and why .
namely , we have omitted experimental information about the scalar - isoscalar and the axial - kaon states .
( _ i _ ) scalar - isoscalar states are not included in the fit because in this first study we have neglected the coupling of the glueball and of additional light scalar states , such as tetraquarks , to the other mesons .
thus , although there are three scalar states @xcite above 1 gev , at most two can be described within our model . below 1 gev
the resonance @xmath3 is still poorly known and @xmath4 is distorted by the nearby @xmath238 threshold .
( _ ii _ ) the axial - kaon state @xmath143 is not included in the fit for a similar reason : the kaonic states from our axial - vector nonet @xmath274 mix with the kaonic states of the nonet @xmath275 which is not part of our lagrangian .
( this is possible , because charge conjugation is not a well - defined quantum number for kaons . ) also in this case we can not assign our theoretical axial kaon field to a specific resonance [ in pdg there are two axial kaons with masses @xmath35 and @xmath36 ] , because the mixing is expected to be large , see e.g. ref
. @xcite .
the experimental quantities discussed in the previous subsection do not depend on all 13 parameters of eq .
( [ 13p ] ) , but on @xmath276 only through the two combinations @xmath277 moreover , instead of the parameters @xmath188 and @xmath189 we use the condensates @xmath278 and @xmath279 : this is equivalent , as @xmath280 are completely determined by the masses of pion and @xmath133 , cf .
( [ m_pi ] ) , ( [ eq : etas ] ) .
summarizing , we have the following eleven parameters entering the fit : @xmath281 we perform our fit using the experimental values for the 17 quantities given in eqs .
( [ eq : weakdecayconstants ] ) ( [ avi ] ) . in addition
, we use the experimental values for the pairs @xmath10 / @xmath13 , @xmath10 / @xmath15 , @xmath11 / @xmath13 , @xmath11 / @xmath15 given in eq .
( [ si1 ] ) .
these are four additional experimental quantities [ except in the case of @xmath10 , where we use the two values for the decay amplitudes instead of one value for the decay width ] . in total
, we therefore fit @xmath282 ( 22 ) experimental quantities to the eleven parameters given in eq .
( [ 11p ] ) .
the corresponding values of @xmath283 are listed in table [ tab : khi ] .
we see that the combination @xmath11/@xmath15 gives the best value for @xmath283 .
in fact , with a @xmath284 of 1.23 this fit is remarkably good , meaning that all physical quantities are reproduced within an average error of @xmath285 . in the framework of an effective model for the strong interaction , which neglects isospin - breaking effects and describes all mesonic resonances up to about 1.8 gev , we perceive this to be a remarkable achievement .
note that also the fit with the pair @xmath10 / @xmath15 has a @xmath286 of the same order of magnitude as the one with the pair @xmath11 / @xmath15 .
nevertheless , we will disregard this fit for two reasons .
firstly , the @xmath286 for the pair @xmath10 / @xmath15 is , although rather small , still larger by a factor of about two than the @xmath286 for the pair @xmath11 / @xmath15 .
secondly , and more importantly , the fit with the pair @xmath10 / @xmath15 produces a scalar - kaon mass of @xmath287 mev , which can not be assigned to any physical resonance as it is much larger than @xmath288 and much smaller than @xmath289 . note that this problem is also present in njl models with mixing between scalar mesons below and above 1 gev @xcite .
[ c]|c|c|c|pair & @xmath283 & @xmath286 + @xmath11/@xmath15 & @xmath290 & @xmath291 + @xmath10/@xmath13 & @xmath292 & @xmath293 + @xmath10/@xmath15 & @xmath294 & @xmath295 + @xmath11/@xmath13 & @xmath296 & @xmath297 + the detailed comparison of theory with data for the pair @xmath11 / @xmath15 is presented in table [ comparison1 ] where the theoretical errors are also shown .
errors for the model parameters ( @xmath298 ) are calculated as the inverse square roots of the eigenvalues of the hessian matrix obtained from @xmath299 , where @xmath300 denotes elements of the parameter set ( [ 11p ] ) and @xmath301 .
theoretical errors @xmath302 for each observable @xmath303 ( mass , decay width ) can be calculated according to the following formula : @xmath304 the remarkable agreement of our results with experimental data in the ( pseudo)scalar and ( axial-)vector sectors can now be explicitly seen for the various quantities shown in table [ comparison1 ] .
we thus conclude that the quark - antiquark states @xmath161 and @xmath305 should be assigned to the resonances @xmath11 and @xmath306 .
moreover , also the axial - vector resonances are well described by the fit : we therefore also interpret them as predominantly quark - antiquark states .
[ c]|c|c|c|observable & fit [ mev ] & experiment [ mev ] + @xmath307 & @xmath308 & @xmath309 + @xmath310 & @xmath311 & @xmath312 + @xmath230 & @xmath313 & @xmath314 + @xmath231 & @xmath315 & @xmath316 + @xmath317 & @xmath318 & @xmath319 + @xmath320 & @xmath321 & @xmath322 + @xmath268 & @xmath323 & @xmath324 + @xmath234 & @xmath325 & @xmath326 + @xmath327 & @xmath328 & @xmath329 + @xmath269 & @xmath330 & @xmath331 + @xmath332 & @xmath333 & @xmath334 + @xmath335 & @xmath336 & @xmath337 + @xmath338 & @xmath339 & @xmath340 + @xmath341 & @xmath342 & @xmath343 + @xmath344 & @xmath345 & @xmath346 + @xmath347 & @xmath348 & @xmath349 + @xmath242 & @xmath350 & @xmath351 + @xmath352 & @xmath353 & @xmath354 + @xmath355 & @xmath356 & @xmath357 + @xmath358 & @xmath359 & @xmath360 + @xmath361 & @xmath362 & @xmath363 + in fig .
[ figure ] we present the results of table [ comparison1 ] in a slightly different way : as the difference of the theoretical and experimental values divided by experimental error ; the error bars correspond to the theoretical error values from our fit .
[ c]cc & in table [ tab : param ] we finally present the values of the parameters of the model together with their errors as obtained from our best fit .
we remark that , since the quantities entering our fit are not affected by interactions with the glueball , this result holds also when including the latter .
[ c]|c|c|parameter & value + @xmath364 [ gev@xmath365 & @xmath366 + @xmath367 [ gev@xmath365 & @xmath368 + @xmath122 [ gev@xmath369 & @xmath370 + @xmath193 [ gev@xmath365 & @xmath371 + @xmath372 & @xmath373 + @xmath374 & @xmath375 + @xmath278 [ gev ] & @xmath376 + @xmath279 [ gev ] & @xmath377 + @xmath378 & @xmath379 + @xmath380 & @xmath381 + @xmath382 & @xmath383 + before moving to the consequences of the fit , we discuss two important aspects of our study : _ ( i ) _ _ mass ordering in the scalar sector : _ as evident from table [ comparison1 ] , our fit yields @xmath384 whereas the experimental mass ordering is opposite .
the reason for our result is the pattern of explicit symmetry breaking implemented in our model , which renders @xmath1 states with a strange quark approximately 100 mev heavier than non - strange states .
however , other mechanisms , as for instance mixing of the currently present @xmath1 states with light tetraquark states , may occur that change this mass ordering @xcite : namely , in this mixing scenario a pure quarkonium and a pure tetraquark in the isovector sector mix to form the resonances @xmath11 and @xmath10 , while a similar mixing in the isodoublet sector leads to the resonances @xmath306 and @xmath385 .
the fact that the mass ordering @xmath386 is realized in nature can be understood from a larger mixing angle , and therefore a larger level repulsion , in the isovector sector .
a detailed study of this scenario necessitates the inclusion of a full nonet of light states , see also the discussion in the conclusions .
an additional point can be raised about the ratio of the full decay widths of the @xmath11 and @xmath15 resonances . from @xmath387 symmetry arguments for @xmath1 states one expects the full decay widths of these resonances to scale as @xcite @xmath388 whereas experimental data suggest @xcite @xmath389 in ref .
@xcite , this problem was also solved by introducing tetraquark fields at the level of an effective lagrangian and by studying tetraquark - quarkonium mixing in the @xmath9 and @xmath12 channels .
contrarily , in our considerations so far only @xmath1 states were taken into account . thus one might expect that our fit results should be closer to the @xmath390 limit rather than to the experimental ratio .
however , the opposite is the case : as evident from table ii , our fit results for @xmath11 and @xmath15 reproduce the experimental ratio @xmath391 .
our analysis shows that the reason is the inclusion of the chiral - anomaly term and , in particular , of ( axial-)vector mesons . artificially decoupling ( axial-)vector states ( i.e. , setting @xmath392 ) and removing the chiral - anomaly term from our lagrangian ( i.e. , setting @xmath393 ) we obtain unphysically large decay widths @xmath394 gev , @xmath395 @xmath396 gev and also @xmath397 @xmath398 , very close to the value obtained from @xmath390 symmetry
. the ratio would be @xmath399 if only ( axial-)vectors were decoupled .
hence , just as in our previous publication @xcite , we emphasize again the crucial importance of the ( axial-)vector degrees of freedom for scalar - meson phenomenology .
however , there still remains an open question how additional light scalar fields will influence the results in the @xmath400-@xmath131 sector .
_ ( ii ) _ _ unitarity corrections : _ our results for masses and decay widths are valid at tree level .
most particles in our model are rather narrow , and thus justify at least in a first approximation this procedure .
this is also in agreement with large-@xmath57 considerations , according to which the role of the contribution of mesonic loops is suppressed .
however , the axial - vector state @xmath401 and the scalar states @xmath11 and @xmath402 as well as the scalar - isoscalar state @xmath5 to be discussed in the next subsection , have a large width ( of the order of @xmath403 mev or more ) .
then , the role of loops and unitarity corrections for both the masses and the decay widths needs to be discussed .
concerning the mass shifts due to mesonic loops , we notice that an indirect indication that their contribution is not too large is the fact that the resonances @xmath404 and @xmath33 have a similar mass , although the former has a much larger decay width than the latter . in a theoretical study of this system
one should determine the pole position of the resonance : the mass is then usually denoted as the real part of the pole . as shown recently in ref .
@xcite , the value of the real part of the pole does not differ too much from the bare ( tree - level ) mass , not even for large coupling constants . concerning the influence of loops on the value of decay widths , we note that ref .
@xcite has shown that the corrections are small as long as the ratio @xmath405 is small .
for instance , for the case of @xmath5 , the average decay width reported by the particle data group is @xmath406 mev whereas the average resonance mass is @xmath407 mev .
this implies @xmath408 .
the ratio is almost as small as in the case of the @xmath39 meson where @xmath409 and it is known from chiral perturbation theory that such resonances obtain only small corrections upon unitarizations [ see , e.g. , ref .
@xcite ] .
we thus conclude that the role of loops should not modify the picture presented here .
it should , however , be stressed that the inclusion of loops is a necessary step for the future : on the one hand we can numerically evaluate the role of mesonic loops and thus quantitatively verify our statements , on the other hand , even if large variations for the states above 1 gev do not occur as we expect , an interesting question is how poles on the complex plane emerge .
namely , these states could arise as tetraquark states as mentioned above , but could also represent dynamically generated states , see alsoe.g.ref
. @xcite .
[ for a detailed discussion of this point including the interrelations of tetraquark , molecular , and dynamically generated states we refer to ref . @xcite .
] we now turn to the scalar - isoscalar mesons .
we shall discuss four different aspects : 1 .
_ results in the large-@xmath57 limit ( @xmath410 ) : _ + from the fit in the previous section we can not immediately obtain the masses of the scalar - isoscalar states of the model because their masses do not depend solely on the combination @xmath411 , but separately on the parameters @xmath412 and @xmath209 .
similarly , the decay rates of the scalar - isoscalar mesons do not depend solely on the combination @xmath413 , but separately on the parameters @xmath414 and @xmath208 .
+ interestingly , the parameters @xmath209 and @xmath208 are large-@xmath57 suppressed . setting them to zero we obtain a prediction for the masses and decay widths of the scalar - isoscalar states in the large-@xmath57 limit . moreover ,
when @xmath415 there is also no mixing between @xmath416 and @xmath67 states : @xmath417 and @xmath418 their masses read : @xmath419 these masses clearly lie above 1 gev . comparing these values to the experimental values for the three isoscalar states above 1 gev @xcite
@xmath420 we see that the mass of @xmath159 is well compatible with that of the resonance @xmath421 [ note that @xmath422 in eq .
( [ above ] ) emulates the pdg mass interval ( 12001500 ) mev . ]
the mass of @xmath160 appears to be close to that of @xmath6 .
remember , though , that we artificially enlarged the experimental error to @xmath285 ; the actual error is only 6 mev , and then the mass of @xmath160 would be ( just ) outside the experimental error .
considering the decays of @xmath160 we shall provide evidence that an assignment of @xmath160 to @xmath7 is also possible ( and even rather likely ) .
+ for @xmath410 the decay rates of @xmath423 into @xmath158 and @xmath424 read @xmath425 for the experimental values we quote the pdg values @xcite : @xmath426 note that for the resonance @xmath5 no branching ratios into @xmath158 and @xmath424 are reported in ref .
the value for @xmath427 is our estimate from pdg and the review @xcite , whereas @xmath428 is our estimate from results presented in refs .
+ our results for @xmath159 are in agreement with the experimental decay widths of @xmath421 our theoretical value for @xmath429 turns out to be too large , while @xmath430 vanishes , because @xmath431 is a pure @xmath67 state .
nevertheless , our model predicts the existence of a scalar - isoscalar state which decays predominantly into kaons ; this is indeed the decay pattern shown by @xmath68 for these reasons we suggest to identify our state @xmath160 as ( predominantly ) @xmath7 .
this identification will be further tested when @xmath432 , @xmath433 below .
+ it should be stressed that a quantitative study of the scalar - isoscalar system can not be performed at present because our model contains only two states , while three resonances appear ( this is also the reason why we did not include the scalar - isoscalar states into the fit ) . as many studies confirm @xcite , the mixing with the scalar glueball can be sizable : a reliable analysis of the scalar - isoscalar states can only be performed when taking the scalar glueball into account . the preliminary results in the @xmath434 sector @xcite have indeed shown that the glueball and quarkonia degrees of freedom interact strongly and that the decay patterns are sizably influenced .
results for @xmath435 , @xmath436 : _ + a nonvanishing value of the parameter @xmath209 induces a mixing of the pure states @xmath18 and @xmath21 , see eq .
( [ eq : signs ] ) .
our fit determines only the value of the linear combination @xmath364 rather than the value of @xmath209 , see discussion before eq .
( [ 11p ] ) .
nonetheless , a range of values for @xmath209 can be estimated using the value of @xmath364 from the fit and the condition @xmath264 .
consequently , we also obtain a range of values for our isoscalar masses @xmath437 and isoscalar decay widths .
the masses vary in the following intervals : @xmath438 mev @xmath439 mev and @xmath440 mev @xmath441 mev . considering these mass values
, @xmath159 may correspond to either @xmath3 , @xmath4 , or @xmath5 and @xmath160 may correspond to either @xmath6 or @xmath7 @xcite .
therefore , a mere calculation of scalar masses does not allow us to assign the scalar states @xmath159 and @xmath160 to physical resonances . in order to resolve this ambiguity
, we will calculate various decay widths of the states @xmath159 and @xmath160 and compare them to data @xcite .
+ the dependence of @xmath442 on @xmath437 is presented in fig .
[ pions ] . +
[ c]cc & + the decay width @xmath443 is consistent with the experimental range @xmath444 mev for the @xmath445 decay width , if 1000 mev @xmath446 1460 mev .
other assignments can be excluded : ( _ i _ ) our @xmath159 state can not be assigned to @xmath3 as @xmath447 mev in the ( new ) pdg mass range @xmath448 mev for @xmath3 .
therefore , this strongly disfavors @xmath3 as a @xmath1 state .
@xmath443 varies between approximately 140 mev and 170 mev in the mass interval of the @xmath4 resonance , i.e. , the interval between 970 mev and 1010 mev .
the pdg result for the full decay width of the @xmath4 resonance is between 40 mev and 100 mev , with the @xmath158 channel being dominant @xcite , i.e. , about a factor two smaller than our theoretical values .
consequently , the assignment of @xmath159 to @xmath4 is also disfavored by our analysis .
therefore , ( although there could still be mixing with other states ) we assign our @xmath159 state to @xmath5 , thus supporting the interpretation of the latter state as a quarkonium .
although experimental data is not conclusive , we mention that the decay width @xmath449 shown in fig .
[ kaons ] is consistent with the data .
+ in the mass interval @xmath450 mev @xmath451 mev , our results for the decay width @xmath452 are consistent with experimental values for both @xmath6 and @xmath7 , see eq .
( [ si2 ] ) .
however , the @xmath424 decay widths present us with a different conclusion : as already indicated in the case @xmath453 , our @xmath160 state decays much more abundantly into kaons than into pions , and the experimental data suggest only one physical resonance with the same feature : @xmath7 .
the other resonance @xmath6 decays preferably into pions .
thus ( and although the mass value for @xmath160 is too small in the range where the decay width into pions agrees with the experimental value ) we assign our @xmath160 state to @xmath7 and inspect in the following whether this assignment is justified by other data .
+ [ c]cc & + we observe from fig . [ kaons ] that @xmath454 rises rapidly and remains above the pdg result @xmath455 mev in the entire mass interval @xmath456 mev .
although the absolute value of the decay width is rather large , several ratios of decay widths can be described correctly by our fit , most notably the @xmath457 decay ratio presented in fig .
[ sppkk2 ] .
+ [ c]cc & + let us first discuss results for @xmath458 ( left panel of fig .
[ sppkk2 ] ) .
we observe that the ratio varies between @xmath459 for @xmath460 mev and @xmath461 for @xmath462 mev .
experimental data regarding this ratio for @xmath5 are unfortunately inconclusive , @xmath463 is quoted by the besii collaboration @xcite and @xmath464 @xmath465 is the value given by the obelix collaboration @xcite ; for the mass interval shown in fig .
[ sppkk2 ] , our result is most consistent with the value of the wa102 collaboration @xcite , @xmath464 @xmath466 . the ambiguities in the experimental value of this ratio
do not allow us to constrain our parameters , although our @xmath159 state is compatible with the @xmath5 data also for this particular case .
+ let us now discuss results for @xmath467 ( right panel of fig .
[ sppkk2 ] ) .
as already mentioned , the analysis of the two - pion decay did not allow for a definitive assignment of our @xmath160 state , as it could correspond either to @xmath6 or to @xmath7 .
the pdg data suggest the ratio @xmath468 or @xmath469 , respectively , whereas , for @xmath7 , we use the wa102 ratio @xmath470 @xcite .
the latter is only marginally consistent with the one preferred by the pdg [ @xmath471 , originally published by the besii collaboration @xcite , that suffers from a large background in the @xmath472 channel ( approximately 50% ) and is therefore omitted from our considerations .
we observe from the right panel of fig .
[ sppkk2 ] that our ratio @xmath473 never corresponds to the one experimentally determined for @xmath6 .
although the ratio shows a strong increase near the left border of the mass interval , decreasing the mass beyond this border would violate the constraint @xmath264 .
thus , the experimental value is out of reach .
conversely , our ratio describes exactly the value @xmath474 if @xmath475 mev .
[ we discard the second possibility @xmath476 mev as the decay width @xmath454 would exceed 600 mev , see fig .
[ kaons ] , i.e. , an order of magnitude larger than the experimental value , see eq .
( [ si2 ] ) . ]
this implies @xmath477 the parameter @xmath209 remains close to the large-@xmath57 limit and is much smaller than @xmath382 , see table [ tab : param ] .
+ the contribution of the pure - strange state @xmath21 to @xmath160 is then approximately 96% , as can be calculated from eqs.([eq : eigen_mass_angle ] ) ( [ a16 ] ) . thus our predominantly strange state @xmath160 describes the @xmath457 ratio of @xmath7 , and not that of @xmath6 , although the correct description requires that @xmath478 corresponds to the mass value of @xmath6 .
the correct description of the decay ratio is an indication that our assignment of @xmath160 to @xmath7 is the correct one , whereas the fact that @xmath478 is smaller than @xmath479 indicates the necessity to include the coupling to the third isoscalar degree of freedom of our model : the glueball . as concluded from the @xmath434 version of our model @xcite
, the @xmath6 is predominantly a glueball state and thus considering the glueball state is expected to induce a level repulsion in the masses .
this may shift the currently too small mass value of the predominantly strange quarkonium from approximately @xmath480 gev to @xmath481 gev , where it is experimentally found .
nonetheless , our results clearly demonstrate that scalar quarkonia are found above , rather than below , @xmath74 gev .
+ until now we have only considered the case where one of our large-@xmath57 suppressed parameters ( @xmath209 ) is nonzero .
we have also investigated the influence of non - vanishing values for the other large-@xmath57 suppressed parameter , @xmath208 , on the decay widths @xmath482 and @xmath483 [ see eqs .
( [ aa ] ) , ( [ bb ] ) , ( [ cc ] ) , and ( [ dd ] ) ] .
consistency with the large-@xmath57 deliberations requires us to keep @xmath208 smaller than , or in the vicinity of , @xmath378 and @xmath380 , see table [ tab : param ] .
we again observe that our ratio @xmath467 never corresponds to the one of the @xmath6 resonance whereas the ratio @xmath474 is correctly described if @xmath484 . in this case
, @xmath478 rises to approximately 1540 mev and is thus outside the pdg result @xmath485 mev but still too small when compared to @xmath486 mev .
thus , it is still necessary to include a glueball degree of freedom into our model .
nonetheless , the qualitative correspondence of our predominantly non - strange quarkonium to @xmath5 and of our predominantly strange quarkonium to @xmath7 , and also the conclusion that scalar @xmath1 states are located in the energy region above 1 gev , remain valid in the case @xmath487 .
+ the assignment of @xmath160 to @xmath7 is further justified considering decays into @xmath145 and @xmath123 mesons .
@xmath488 and @xmath489 decay channels for the scalar - isoscalar mesons : _
+ pdg data suggest the following values of @xmath488 decay widths for @xmath6 and @xmath7 @xcite @xmath490 our analysis always yields @xmath491 mev ; there is consequently no value of @xmath478 at which @xmath492 would be compatible with @xmath493 .
however , @xmath475 mev [ obtained requiring @xmath494 , i.e. , @xmath477 and @xmath495 yields @xmath496 ( where the errors arise from the uncertainty in @xmath209 only ) , i.e. , in agreement with the experimental value for @xmath497 quoted in eq . ( [ gammaetaeta ] ) .
the same parameter set also yields @xmath498 purely a prediction , as @xmath499 has not yet been measured .
+ the choice @xmath500 @xmath501 also yields @xmath502 whereas experimental data read@xmath503 in all cases , the @xmath6 data are off by several standard deviations from our theoretical results , while there is good agreement with the results for @xmath7 .
+ we have also considered decays involving the @xmath123 meson . since the threshold for @xmath489 decays is at approximately 1.5 gev , it suffices to consider decays of @xmath160 only . the value @xmath504 yields
@xmath505 whereas experimental data read @xmath506 the decay ratio @xmath507 is unknown ; there are also no data for the @xmath489 decay channel of @xmath7 .
still we observe that neither @xmath508 nor @xmath509 describe the corresponding experimental results for @xmath6 . indeed the only piece of experimental data regarding @xmath6 that is described by our fit results is the one for the @xmath510 decay ratio .
nonetheless , all the other results regarding decay ratios obtained by our analysis clearly demonstrate the correspondence of our predominantly strange state @xmath160 to @xmath7 ; @xmath159 was found to correspond to @xmath5 already in the discussion of the @xmath158 decay channel .
consequently , we conclude that @xmath5 and @xmath7 are favoured as scalar @xmath1 states .
however , we also stress again that the mass of our @xmath160 remains too low when compared to @xmath479 due to a missing scalar - glueball state expected to shift @xmath478 to @xmath479 by level repulsion .
our lagrangian ( [ eq : lagrangian ] ) implements the mixing of two pure pseudoscalar isosinglet states , @xmath146 and @xmath511 .
the mixing term is presented in eq .
( [ eq : etans ] ) .
the mass terms for @xmath132 ( [ eq : etan ] ) and @xmath133 ( [ eq : etas ] ) are determined by our fit parameters presented in table [ tab : param ] .
the same parameters also determine the @xmath145-@xmath123 mixing angle as [ see eqs .
( [ eq : eigen_mass_angle ] ) ( [ a16 ] ) ] @xmath512 the result is close to maximal mixing , i.e. , our result suggests a slightly larger mixing than those of ref .
@xcite .
the @xmath143 state has not been assigned to a physical resonance and included into our fit , because the pdg listing suggest two distinct assignment candidates : @xmath35 and @xmath36 , expected to mix @xcite .
therefore , mass and decay widths are left as predictions of this work .
there are three decay widths of the @xmath143 state that can be calculated within our model at tree level : @xmath143 @xmath513 , @xmath514 , and @xmath515 .
they account for approximately @xmath516 of the full @xmath35 decay width and almost @xmath517 of the full @xmath36 decay width @xcite . using the parameter values stated in table [ tab : param ]
it is possible to calculate the mass , eq .
( [ m_k_1 ] ) , as well as the decay width [ via the generic decay - width formula ( [ gk1vp ] ) ] of our @xmath143 state .
we obtain the following results : @xmath518 the mass is within @xmath519 of @xmath520 mev and thus rather close to the experimental result . however , the value of the full decay width is @xmath521 mev , while the pdg data read @xmath522 mev and @xmath523 mev .
our result is therefore approximately three times too large when compared to the data for @xmath35 and approximately @xmath524 too large when compared to the data for @xmath36 ; errors have been omitted from the calculation .
these results demonstrate the necessity to include a pseudovector @xmath525 nonet into our model and implement its mixing with the already present axial - vector nonet . a possible mixing term between an axial - vector nonet @xmath526 and a pseudovector nonet @xmath527 reads@xmath528)\ ; , \label{k11}\ ] ] with @xmath110 from eq .
( [ eq : expl_sym_br_delta ] ) .
various studies have indeed found the mixing to be non - negligible @xcite ; see also ref .
@xcite . as a consequence of our fit
we can determine the branching ratio of the resonance @xmath11 into @xmath424 , @xmath529 and @xmath530 we obtain the following values [ see eqs .
( [ b1 ] ) , ( [ b2 ] ) , and ( [ b3])]:@xmath531 this leads to the following branching ratios:@xmath532 which should be compared with the experimental results @xcite @xmath533 our results are , within errors , consistent with the data .
the mass of the @xmath39 meson consists of three terms , @xmath534 the first term @xmath535 is generated by the condensation of the dilaton field @xmath536 @xmath537 the second term is proportional to the chiral condensate @xmath538 in the large-@xmath57 limit the parameter @xmath501 and we can determine the terms from the results of the fit : @xmath539 gev and @xmath540 gev .
it turns out that the glueball - driven term is dominant .
it is interesting to note that models based on qcd sum rules @xcite or brown - rho scaling @xcite predict that @xmath541 , where @xmath542 in the first and @xmath543 in the latter case .
the order parameter for chiral symmetry breaking @xmath278 decreases as a function of nuclear matter density @xmath88 : for small densities it is well known [ see , e.g. ref .
@xcite ] that @xmath544 so these models predict a substantial decrease of @xmath268 already at nuclear matter saturation density , @xmath545 . in this respect , it is interesting to evaluate the value of the @xmath39 mass at @xmath545 according to eq .
( [ rhomass ] ) .
we assume that the glueball - driven term @xmath535 does not vary .
this can be motivated by considering that the glueball is massive and a substantial decrease in its mass may occur only at much higher density . assuming that the linear dependence on the density of the condensate holds up to saturation density , we obtain that @xmath268 decreases by about @xmath546 mev compared to @xmath547 mev predicted by refs . @xcite .
obviously , these considerations are only qualitative but could represent the starting point of interesting studies of vector mesons in the medium , which is an important aim of several experimental investigations @xcite . to this end , one has to calculate the behavior of the dilaton and chiral condensates in the medium within the same theoretical framework .
we have presented a linear sigma model with three quark flavors .
the model implements the symmetries of qcd , the discrete @xmath548 symmetry , the global chiral @xmath549 symmetry , and the breaking mechanisms of the last : spontaneous ( due to the chiral condensate ) , explicit ( due to non - vanishing quark masses ) as well as at the quantum level [ the @xmath50 anomaly ] .
moreover , it implements also dilatation symmetry and its breaking ( the so - called trace anomaly ) in the ym sector of the theory . in this way , besides explicit breaking of dilatation symmetries arising from the nonzero current quark masses and the trace and axial anomalies in the gauge sector , all other interaction terms in our lagrangian carry mass dimension equal to four .
furthermore , requiring analyticity in the fields makes the number of allowed terms finite .
the model has been used to describe meson states up to energies of @xmath550 gev .
this energy region exhibits numerous resonances , related by scattering reactions and decays . for this reason , a realistic model of qcd degrees of freedom in the mentioned energy region should describe as many of the resonances as possible .
thus , we have constructed a linear sigma model that contains scalar ( two isoscalars , @xmath159 and @xmath160 , as well as an isotriplet , @xmath161 , and two isodoublets , @xmath131 ) , pseudoscalar ( @xmath177 , @xmath135 , @xmath145 , @xmath123 ) , vector ( @xmath39 , @xmath551 , @xmath152 , @xmath235 ) , and axial - vector [ @xmath241 , @xmath143 , @xmath33 , @xmath154 ] degrees of freedom .
we have thus constructed a single model that contains ( pseudo)scalars and ( axial-)vectors both in the non - strange and strange channels . to our knowledge , this is the first time that such a comprehensive approach has been presented .
the model , dubbed extended linear sigma model ( elsm ) , has allowed us to study the overall phenomenology of mesons and , in particular , to explore the nature of scalar and axial - vector resonances . in order to test our model we have performed a global fit to @xmath282 experimental quantities involving both the ( pseudo)scalar and the ( axial-)vector masses and decays . due to mixing with the scalar glueball (
not included here explicitly , because its coupling to the other mesons was neglected ) , we did not include the scalar - isoscalar resonances in the fit .
similarly , we have omitted the axial - vector resonance @xmath553 due to the fact that in reality a large mixing of two kaonic fields from the @xmath274 and @xmath275 nonets takes place .
one of the central questions of our discussion has been the assignment of the scalar states : to this end we have tested the possible scenarios for the isotriplet and isodoublet scalar states by assigning our scalar fields @xmath161 to @xmath10 or @xmath11 and @xmath131 to @xmath13 or @xmath15 .
the outcome is univocal : the global fit works well only if the states @xmath11 and @xmath554 are interpreted as quark - antiquark states . on the contrary ,
the other combinations deliver unacceptably large values of @xmath283 , see table [ tab : khi ] .
we thus conclude that the scalar @xmath9 and @xmath12 states lie above 1 gev and have to be identified with the resonances @xmath11 and @xmath29 moreover , the overall phenomenology described by the fit is very good , see table [ comparison1 ] .
it is then possible to properly describe many different mesonic masses and decays within a unified treatment based on the symmetries of qcd .
it should be stressed that the inclusion of the ( axial-)vector mesons has a crucial impact on our results and represents the most important new ingredient of our approach .
the good agreement with data also shows that the axial - vector mesons can be interpreted , just as their vector chiral partners , as quark - antiquark states .
we have then studied the consequences of our fit .
we have primarily concentrated on the scalar - isoscalar sector which was not included in the fit . in the limit @xmath555 it is possible to make clear predictions for the two states @xmath159 and @xmath160 .
their masses lie above 1 gev and their decay patterns have led us to identify @xmath159 with ( predominantly ) @xmath5 and @xmath160 with ( predominantly ) @xmath68 the theoretical decay rates of @xmath159 are in agreement with experiment ; @xmath160 decays only into kaons , but turns out to be too wide . at a qualitative level
, this large-@xmath57 result clearly shows that also the scalar - isoscalar quark - antiquark states lie above 1 gev .
the overall situation in the scalar - isoscalar sector can be slightly improved when adding large-@xmath57 suppressed terms .
some of these terms correspond to interactions between mesons and a glueball state .
these were neglected here , but are necessary for a full quantitative study .
then starting from @xmath556 , @xmath557 , and @xmath558 one aims to describe properly @xmath559 @xmath560 and @xmath7 .
additionally , we have studied other consequences of the fit , such as predictions for the @xmath488 channel of scalar - isoscalar states , the @xmath561-@xmath123 mixing , the prediction for the axial - vector kaonic state , and for the @xmath161 branching ratios .
finally , we have also discussed the origin of the mass of the @xmath39 meson and , at a qualitative level , its possible value at nuclear matter density .
there is , however , one question that remains open .
interpreting resonances above 1 gev as @xmath1 states leads to questions about the nature of @xmath3 , @xmath562 @xmath4 , and @xmath13 .
their presence is necessary for the correct description of @xmath158 scattering lengths [ see , e.g. ref .
there are two possibilities : ( _ i _ ) they can arise as ( quasi-)molecular states of @xmath158 or @xmath424 , respectively , or ( _ ii _ ) as tetraquark states @xcite .
the question is whether the attraction is large enough so that these states are bound , or whether the attraction is not sufficient so that they are ( unstable ) resonances in the continuum .
this question and deciding whether possibility ( _ i _ ) or ( _ ii _ ) is realized in nature represent interesting starting points for future studies , along the lines of the bethe - salpeter approach of ref .
@xcite , the lattice approach of ref .
@xcite , or even holographic approaches such as those presented in ref .
@xcite . in either case
, one may include the low - lying isoscalar states as interpolating fields in our lagrangian , such as has been done in ref .
@xcite .
the present model can be also studied in the baryonic sector , see ref .
@xcite for the two - flavor case .
the very same ideas of chiral symmetry and dilatation invariance can be applied to the baryonic sector as well .
the extension of the model to three flavors in the baryonic sector would surely represent an interesting problem : a multitude of data on decays and masses is available to make a precise test of our approach .
additionally , restoration of chiral symmetry at nonzero temperature and density is one of the fundamental questions of modern hadron physics , see , e.g. refs .
@xcite , in which the two - flavor version of this model has been studied at nonzero density , or ref .
@xcite for alternative approaches to the exploration of nonzero chemical potential .
linear sigma models are appropriate to study chiral symmetry restoration because they contain from the onset not only ( pseudo)scalar and ( axial-)vector mesons but also their chiral partners ; mass degeneration of chiral partners represents a signal for the chiral transition .
therefore , we also plan to apply the model to study chiral symmetry restoration at nonzero temperatures and densities .
wolf and p. kovcs thank the institute for theoretical physics of goethe university for its hospitality , where part of this work was done .
they were partially supported by the hungarian otka funds t71989 and t101438
. the work of d. parganlija and f. giacosa was supported by the foundation of the polytechnical society frankfurt .
this work was financially supported by the helmholtz international center for fair within the framework of the loewe program ( landesoffensive zur entwicklung wissenschaftlich - konomischer exzellenz ) launched by the state of hesse .
rischke thanks r. longacre and r. pisarski for enlightening discussions .
after spontaneous symmetry breaking ( see sec . [ ssec : ssb_and_mass ] ) , from the quadratic terms of the lagrangian the ( squared ) mass matrices for the different fields are read off as @xmath563 where @xmath564 is the @xmath565 transformation matrix to the non strange strange base , @xmath566{cccc}\sqrt{2 } & 0 & \dots & 1\\ 0 & 1 & \dots & 0\\ \vdots & \vdots & \ddots & \vdots\\ 1 & 0 & \dots & -\sqrt{2}\end{array } \right ) , \ ] ] and the @xmath567 coefficient tensors are @xmath568 , \\
j_{ijkl } & = g_{1}^{2}f_{ikm}f_{jlm}+\frac{h_{1}}{2}\delta_{ij}\delta _ { kl}+\frac{h_{2}}{2}d_{ijm}d_{klm}+\frac{h_{3}}{4}\left ( d_{ikm}d_{jlm}+d_{ilm}d_{jkm}-f_{ikm}f_{jlm}-f_{ilm}f_{jkm}\right ) , \\
j_{ijkl}^{\prime } & = g_{1}^{2}d_{ikm}d_{jlm}+\frac{h_{1}}{2}\delta _ { ij}\delta_{kl}+\frac{h_{2}}{2}d_{ijm}d_{klm}-\frac{h_{3}}{4}\left ( d_{ikm}d_{jlm}+d_{ilm}d_{jkm}-f_{ikm}f_{jlm}-f_{ilm}f_{jkm}\right ) .\end{aligned}\ ] ] as can be seen from eqs .
, and eqs .
, there is a mixing in the pseudoscalar and scalar @xmath569 sectors , which can be resolved by the following two - dimensional orthogonal transformations @xmath570{cc}\cos\theta_{\eta/\sigma } & \sin\theta_{\eta/\sigma}\\ -\sin\theta_{\eta/\sigma } & \cos\theta_{\eta/\sigma}\end{array } \right ) \text { , } \label{eq : orthog}\ ] ] where @xmath571 are the pseudoscalar and scalar mixing angles .
the ( squared ) mass eigenvalues can be written with the help of the mixing angles as @xmath572 where , if we require that @xmath573 , the mixing angles are given by @xmath574 it can be seen that , if @xmath575 , then @xmath576 and , if @xmath577 , then @xmath578 . substituting these expression into eq .
it is found that @xmath579 , \\ m_{\eta_{2}/\sigma_{2}}^{2 } & = \frac{1}{2}\left [ m_{\eta_{n}/\sigma_{n}}^{2}+m_{\eta_{s}/\sigma_{s}}^{2}+\sqrt{(m_{\eta_{n}/\sigma_{n}}^{2}-m_{\eta_{s}/\sigma_{s}}^{2})^{2}+4m_{\eta_{ns}/\sigma_{ns}}^{4}}\right ] , \label{a16}\ ] ] from which it is obvious that the condition @xmath573 is fulfilled . in this way we know that @xmath580 must be identified as @xmath320 , and similarly for @xmath581 .
in this section we show the explicit form of some of the most relevant decay widths calculated from our model at tree - level using eqs .
( [ gamma ] ) .
the formulas below are organized according to the type of the decaying particle . at first we are considering the decays of the scalar isotriplet @xmath161 , scalar kaon @xmath131 , and scalar isosinglets @xmath582 .
the @xmath161 state has three relevant decay channels , namely into @xmath583 , @xmath584 , and @xmath424 , the first two of which are strongly connected due to the mixing between @xmath145 and @xmath123 .
the @xmath585 and @xmath586 decay widths , s have the same formulas for the decay widths . ]
considering that in both cases @xmath587 , read @xmath588 ^{1/2}|\mathcal{m}_{a_{0}\rightarrow\eta\pi}|^{2},\label{b1}\\ \gamma_{a_{0}\rightarrow\eta^{\prime}\pi } & = \frac{1}{8m_{a_{0}}\pi}\left [ \frac{(m_{a_{0}}^{2}-m_{\eta^{\prime}}^{2}-m_{\pi}^{2})^{2}-4m_{\eta^{\prime}}^{2}m_{\pi}^{2}}{4m_{a_{0}}^{4}}\right ] ^{1/2}|\mathcal{m}_{a_{0}\rightarrow\eta^{\prime}\pi}|^{2 } , \label{b2}\ ] ] with the following transition matrix elements , @xmath589 where @xmath590 and @xmath591 , \\ c_{a_{0}\eta_{n}\pi } & = g_{1}z_{\pi}^{2}w_{a_{1}},\\ a_{a_{0}\eta_{s}\pi } & = \frac{1}{2}c_{1}z_{\pi}z_{\eta_{s}}\phi_{n}^{2}\phi_{s}.\end{aligned}\ ] ] the @xmath592 decay width includes two subchannels , @xmath593 and @xmath594 , which results in the isospin factor @xmath222 , and accordingly the decay width is found to be@xmath595 where @xmath596 now turning to the scalar kaon there is only one relevant decay channel , @xmath597 , for which the decay width reads @xmath598
^{1/2 } & \left [ \frac{}{}a_{k_{0}k\pi}+(c_{k_{0}k\pi } + d_{k_{0}k\pi}-b_{k_{0}k\pi } ) \right .
\\ & \times\left .
\frac{m_{k_{0}}^{2}-m_{k}^{2}-m_{\pi}^{2}}{2}+c_{k_{0}k\pi } m_{k}^{2}+d_{k_{0}k\pi}m_{\pi}^{2}\right ] \ ; , \end{split}\ ] ] where @xmath599 , \\ c_{k_{0}k\pi } & = \frac{z_{\pi}z_{k}z_{k_{0}}}{2}[-g_{1}(iw_{k^{\star}}+w_{k_{1}})+\sqrt{2}iw_{k^{\star}}w_{k_{1}}(g_{1}^{2}-h_{3})\phi_{s}],\\ d_{k_{0}k\pi } & = \frac{z_{\pi}z_{k}z_{k_{0}}}{4 } \left \ { 2g_{1}(iw_{k^{\star}}-w_{a_{1}})+iw_{k^{\star}}w_{a_{1 } } \left [ ( 2h_{3}-h_{2}-3g_{1}^{2})\phi_{n}+\sqrt{2}(g_{1}^{2}+h_{2})\phi_{s } \right ] \right \}.\end{aligned}\ ] ] finally , for the @xmath600 particles there are two relevant decay channels : the two - pion and the two - kaon channels .
it is important to note that due to the particle mixing between @xmath601 and @xmath602 , the matrix elements are linear combinations that depend on the scalar mixing angle @xmath603 ( see appendix [ app : mass ] ) , as can be seen explicitly below .
the decay widths in the @xmath158 channel are @xmath604 where an isospin factor of @xmath605 was considered and the @xmath606 , which would mean @xmath222 . however , since the two @xmath607 are indistinguishable , the solid - angle integral ( there is no angular dependence at tree - level ) ends up as @xmath608 instead of @xmath609 , which means a factor of @xmath610 in case of @xmath606 compared to @xmath611 , thus @xmath612 . ] and the matrix elements are @xmath613 -\left ( \lambda_{1}+\frac{\lambda_{2}}{2}\right ) \right\ } , \label{aa}\\ \mathcal{m}_{f_{0}\pi}^{{h}}(m ) & = 2z_{\pi}^{2}\phi_{s}\left\ { -\frac { g_{1}^{2}}{4}\frac{m^{2}}{m_{a_{1}}^{2}}\left ( 1-\frac{2m_{\pi}^{2}}{m^{2}}\right ) \frac{h_{1}\phi_{n}^{2}}{m_{a_{1}}^{2}}-\lambda_{1}\right\ } .
\label{bb}\ ] ] in the @xmath424 channel , where @xmath222 , the decay widths read @xmath614 where the matrix elements , using the notations @xmath615 and @xmath616 , are @xmath617 , \label{cc}\\ \mathcal{m}_{f_{0}k}^{h}(m ) & = -z_{k}^{2}\left [ 2(\lambda_{1}+\lambda _ { 2})\phi_{s}-\frac{\lambda_{2}}{\sqrt{2}}\phi_{n}+\sqrt{2}g_{1}w_{k_{1}}(m_{k}^{2}-m^{2})+w_{k_{1}}^{2}\left ( 2h_{s}\phi_{s}-\frac{h_{3}-g_{1}^{2}}{\sqrt{2}}\phi_{n}\right ) \frac{m^{2}-2m_{k}^{2}}{2}\right ] .
\label{dd}\ ] ] in the case of the vector mesons we are considering the decays of the @xmath39 meson , the @xmath152 vector kaon , and the @xmath235 meson .
all these particles have only one relevant decay channel .
the first is the @xmath618 decay which has the following quite simple decay width formula @xmath619 ^{3/2}\left [ g_{1}z_{\pi}^{2}-\frac{g_{2}}{2}\left ( z_{\pi}^{2}-1\right ) \right ] ^{2}.\ ] ] the next one is the @xmath620 decay , in case of which the decay width reads @xmath621
^{3/2}(a_{k^{\star}k\pi}-b_{k^{\star}k\pi } + c_{k^{\star}k\pi})^{2}\;,\ ] ] where the constants are defined as @xmath622 \ ; , { \nonumber}\\ b_{k^{\star}k\pi } & = \frac{1}{4}z_{\pi}z_{k}\left [ 2g_{1}+w_{a_{1}}(-3g_{1}^{2}-h_{2}+2h_{3})\phi_{n}+\sqrt{2}w_{a_{1}}(g_{1}^{2}+h_{2})\phi _ { s}\right ] \ ; , { \nonumber}\\ c_{k^{\star}k\pi } & = \frac{1}{2}z_{\pi}z_{k}w_{a_{1}}w_{k_{1}}g_{2}m_{k^{\star}}^{2}\;.\end{aligned}\ ] ] finally , the @xmath623 decay width reads @xmath624 ^{3/2}\left
[ 2g_{1}z_{k}^{2}\left ( 1+\frac{\delta_{n}-\delta_{s}}{m_{\phi}^{2}}\right ) -g_{2}(z_{k}^{2}-1)\right ] ^{2}.\ ] ] turning to axial - vector mesons , the considered decays are the @xmath241 decay with two relevant channels and the @xmath151 decay with one relevant channel .
since in case of the @xmath625 and @xmath626 decays , the decaying as well as one of the resulting particle are ( axial-)vector mesons , the matrix elements have a more complicated form than in the previous cases , as can be seen below .
the first considered decay width is the one of the @xmath627 process , which takes the following simple form @xmath628 ^{3}\;. \label{eq : a1pigamma}\ ] ] in case of the @xmath625 decay width one has to consider two channels ( @xmath629 ) , thus @xmath222 and the decay width is found to be @xmath630 ^{1/2}\left [ |v_{\mu\nu}|^{2}-\frac{|v_{\mu\nu } k_{\rho}^{\nu}|^{2}}{m_{\rho}^{2}}-\frac{|v_{\mu\nu}k_{a_{1}}^{\mu}|^{2}}{m_{a_{1}}^{2}}+\frac{|v_{\mu\nu}k_{a_{1}}^{\mu}k_{\rho}^{\nu}|^{2}}{m_{\rho } ^{2}m_{a_{1}}^{2}}\right ] \ ; , \label{eq : decay_a1_rho_pi}\ ] ] where @xmath631 is the vertex following from the relevant part of the lagrangian , @xmath632\right\ } \label{eq : v_mu_nu}\ ] ] and @xmath633 , @xmath634 and @xmath635 are the four - momenta of @xmath241 , @xmath39 , and @xmath177 in the rest frame of @xmath241 , respectively . using the following kinematic relations @xmath636 the terms in eq .
are given by @xmath637 -6\frac{g_{1}g_{2}(g_{1}^{2}-h_{3})}{m_{a_{1}}^{2}}(m_{a_{1}}^{2}-m_{\rho}^{2})\right\ } , { \nonumber}\\ & \frac{|v_{\mu\nu}k_{\rho}^{\nu}|^{2}}{m_{\rho}^{2}}=z_{\pi}^{2}\phi_{n}^{2}\left\ { ( g_{1}^{2}-h_{3})^{2}-\frac{g_{1}^{2}g_{2}^{2}}{m_{\rho}^{2}}(e_{\rho}^{2}-m_{\rho}^{2})+2\frac{g_{1}g_{2}(g_{1}^{2}-h_{3})}{m_{\rho}^{2}}(e_{\pi}^{2}-m_{\pi}^{2})\right\ } , \label{eq : v_prod}\\ & \frac{|v_{\mu\nu}k_{a_{1}}^{\nu}|^{2}}{m_{a_{1}}^{2}}=z_{\pi}^{2}\phi _ { n}^{2}\left\ { ( g_{1}^{2}-h_{3})^{2}-\frac{g_{1}^{2}g_{2}^{2}}{m_{a_{1}}^{4}}m_{\rho}^{2}\mathbf{k}^{2}-2\frac{g_{1}g_{2}(g_{1}^{2}-h_{3})}{m_{a_{1}}^{2}}(e_{\rho}^{2}-m_{\rho}^{2})\right\ } , { \nonumber}\\ & \frac{|v_{\mu\nu}k_{a_{1}}^{\mu}k_{\rho}^{\nu}|^{2}}{m_{\rho}^{2}m_{a_{1}}^{2}}=z_{\pi}^{2}\phi_{n}^{2}\frac{(g_{1}^{2}-h_{3})^{2}}{m_{\rho}^{2}}e_{\rho}^{2}.{\nonumber}\ ] ] analogously to the previous case , the width of the @xmath626 decay which includes four sub - channels ( @xmath638 , @xmath639 , @xmath640 , @xmath641 ) , resulting in @xmath642 , becomes @xmath643 ^{1/2}\left [ |\tilde{v}_{\mu\nu}|^{2}-\frac{|\tilde{v}_{\mu\nu}k_{k^{\star}}^{\nu}|^{2}}{m_{k^{\star}}^{2}}-\frac{|\tilde{v}_{\mu\nu}k_{f_{1s}}^{\mu}|^{2}}{m_{f_{1s}}^{2}}+\frac{|\tilde{v}_{\mu\nu}k_{f_{1s}}^{\mu}k_{k^{\star}}^{\nu}|^{2}}{m_{k^{\star}}^{2}m_{f_{1s}}^{2}}\right ] , \label{eq : decay_f1s_k_kstar}\ ] ] where @xmath644 has the same lorentz structure as eq .
and the only difference is in the constants in front of the different terms .
more explicitly , @xmath645\right\ } , \\ & a_{fkk}=\frac{1}{4}\left [ g_{1}^{2}(\sqrt{2}\phi_{n}-6\phi_{s})+\sqrt { 2}h_{2}(\phi_{n}-\sqrt{2}\phi_{s})+4h_{3}\phi_{s}\right ] , \\ & b_{fkk}=-\frac{1}{\sqrt{2}}g_{2}w_{k_{1}}.\end{aligned}\ ] ] the kinematic relations are the same as in eq . with the following substitutions : @xmath646 , while the expressions analogous to eq . are , @xmath647 -6a_{fkk}b_{fkk}(m_{f_{1s}}^{2}-m_{k^{\star}}^{2})\right\ } , { \nonumber}\\ & \frac{|\tilde{v}_{\mu\nu}k_{k^{\star}}^{\nu}|^{2}}{m_{k^{\star}}^{2}}=z_{k}^{2}\left\ { a_{fkk}^{2}-b_{fkk}^{2}(e_{k^{\star}}^{2}-m_{k^{\star}}^{2})+2a_{fkk}b_{fkk}(e_{k}^{2}-m_{k}^{2})\right\ } , \label{eq : v_prod2}\\ & \frac{|\tilde{v}_{\mu\nu}k_{f_{1s}}^{\nu}|^{2}}{m_{f_{1s}}^{2}}=z_{k}^{2}\left\ { a_{fkk}^{2}-b_{fkk}^{2}m_{k^{\star}}^{2}\mathbf{k}^{2}-2a_{fkk}b_{fkk}(e_{k^{\star}}^{2}-m_{k^{\star}}^{2})\right\ } , { \nonumber}\\ & \frac{|\tilde{v}_{\mu\nu}k_{f_{1s}}^{\mu}k_{k^{\star}}^{\nu}|^{2}}{m_{k^{\star}}^{2}m_{f_{1s}}^{2}}=z_{k}^{2}\frac{a_{fkk}^{2}}{m_{k^{\star}}^{2}}e_{k^{\star}}^{2}.{\nonumber}\ ] ] analogously , the width for a generic decay of the form @xmath648 , where @xmath649 denotes a vector and @xmath650 a pseudoscalar state , reads@xmath651 \ ; , \label{gk1vp}\ ] ] where @xmath652 for the @xmath653 and @xmath654 decays and @xmath587 for the @xmath655 decay , @xmath656 ( the theta function ensures that the decay width vanishes below threshold ) and @xmath657\right\ } \text{. } \label{hk1vp}\ ] ] the values of @xmath658 , @xmath659 , and @xmath660 depend on the process considered : @xmath661 and @xmath662 .
the coefficients read @xmath663 \;,\label{ak1rk}\\ b_{k_{1}\rho k } & = \frac{i}{2}z_{k}g_{2}w_{k_{1}}\;,\label{bk1rk}\\ a_{k_{1}\omega_{n}k } & = -\frac{i}{4}z_{k}[g_{1}^{2}(\phi_{n}+\sqrt{2}\phi_{s})-h_{2}(\phi_{n}-\sqrt{2}\phi_{s})-2h_{3}\phi_{n}]\;,\label{ak1ok}\\ b_{k_{1}\omega_{n}k } & = -\frac{i}{2}z_{k}g_{2}w_{k_{1}}\;\text{. } \label{bk1ok}\ ] ] a. v. anisovich _ et al .
_ , phys .
b * 449 * , 154 ( 1999 ) ; m. ablikim _ et al .
_ [ bes collaboration ] , phys .
b * 603 * , 138 ( 2004 ) [ arxiv : hep - ex/0409007 ] ; m. ablikim _ et al . _
[ bes collaboration ] , phys .
b * 607 * , 243 ( 2005 ) [ hep - ex/0411001 ] ; m. ablikim _ et al .
_ [ bes collaboration ] , phys . rev .
d * 72 * , 092002 ( 2005 ) [ arxiv : hep - ex/0508050 ] ; d. v. bugg , arxiv : hep - ph/0603018 ; a. v. anisovich , d. v. bugg , v. a. nikonov , a. v. sarantsev and v. v. sarantsev , phys . rev .
d * 85 * , 014001 ( 2012 ) [ arxiv:1110.4333 [ hep - ex ] ] .
a. v. anisovich and a. v. sarantsev , phys .
b * 413 * , 137 ( 1997 ) [ hep - ph/9705401 ] ; r. delbourgo and m. d. scadron , int .
j. mod .
a * 13 * , 657 ( 1998 ) [ hep - ph/9807504 ] ; d. black , a. h. fariborz , s. moussa , s. nasri and j. schechter , phys .
d * 64 * , 014031 ( 2001 ) [ hep - ph/0012278 ] ; m. d. scadron , f. kleefeld , g. rupp and e. van beveren , nucl .
a * 724 * , 391 ( 2003 ) [ hep - ph/0211275 ] ; d. v. bugg , phys .
b * 572 * , 1 ( 2003 ) [ erratum - ibid .
b * 595 * , 556 ( 2004 ) ] .
i. caprini , g. colangelo and h. leutwyler , phys .
lett . * 96 * , 132001 ( 2006 ) [ hep - ph/0512364 ] ; f. j. yndurain , r. garcia - martin and j. r. pelaez , phys . rev .
d * 76 * , 074034 ( 2007 ) [ hep - ph/0701025 ] ; h. leutwyler , aip conf .
* 1030 * , 46 ( 2008 ) [ arxiv:0804.3182 [ hep - ph ] ] ; r. kaminski , r. garcia - martin , p. grynkiewicz and j. r. pelaez , nucl .
suppl . * 186 * , 318 ( 2009 ) [ arxiv:0811.4510 [ hep - ph ] ] ; r. garcia - martin , r. kaminski , j. r. pelaez and j. ruiz de elvira , phys .
lett . * 107 * , 072001 ( 2011 ) [ arxiv:1107.1635 [ hep - ph ] ] .
m. f. m. lutz and e. e. kolomeitsev , nucl .
phys . a * 730 * , 392 ( 2004 ) [ arxiv : nucl - th/0307039 ] ; m. wagner and s. leupold , phys .
b * 670 * , 22 ( 2008 ) [ arxiv:0708.2223 [ hep - ph ] ] ; m. wagner and s. leupold , phys .
d * 78 * , 053001 ( 2008 ) [ arxiv:0801.0814 [ hep - ph ] ] ; s. leupold and m. wagner , arxiv:0807.2389 [ nucl - th ] .
a. heinz , s. struber , f. giacosa and d. h. rischke , phys .
d * 79 * , 037502 ( 2009 ) [ arxiv:0805.1134 [ hep - ph ] ] ; a. heinz , s. struber , f. giacosa and d. h. rischke , acta phys .
supp . * 3
* , 925 ( 2010 ) [ arxiv:1006.5393 [ hep - ph ] ] .
m. bando , t. kugo and k. yamawaki , phys .
rept . * 164 * , 217 ( 1988 ) ; g. ecker , j. gasser , a. pich and e. de rafael , nucl .
b * 321 * , 311 ( 1989 ) ; e. e. jenkins , a. v. manohar and m. b. wise , phys .
lett . * 75 * , 2272 ( 1995 ) [ arxiv : hep - ph/9506356 ] . c. terschlusen and s. leupold , prog .
phys . * 67 * , 401 ( 2012 ) [ arxiv:1111.4907 [ hep - ph ] ] .
s. janowski , d. parganlija , f. giacosa and d. h. rischke , phys .
d * 84 * , 054007 ( 2011 ) [ arxiv:1103.3238 [ hep - ph ] ] .
d. parganlija , f. giacosa and d. h. rischke , aip conf .
* 1030 * , 160 ( 2008 ) [ arxiv:0804.3949 [ hep - ph ] ] ; d. parganlija , f. giacosa and d. h. rischke , pos confinement * 8 * , 070 ( 2008 ) [ arxiv:0812.2183 [ hep - ph ] ] ; d. parganlija , f. giacosa and d. h. rischke , arxiv:0911.3996 [ nucl - th ] ; d. parganlija , f. giacosa and d. h. rischke , acta phys . polon . supp . * 3 * , 963 ( 2010 ) [ arxiv:1004.4817 [ hep - ph ] ] .
d. parganlija , f. giacosa , d. h. rischke , p. kovacs and g. wolf , int .
j. mod .
a * 26 * , 607 ( 2011 ) [ arxiv:1009.2250 [ hep - ph ] ] ; d. parganlija , f. giacosa , p. kovacs and g. wolf , aip conf .
proc . * 1343 * , 328 ( 2011 ) [ arxiv:1011.6104 [ hep - ph ] ] ; p. kovacs , g. wolf , f. giacosa and d. parganlija , epj web conf .
* 13 * , 02006 ( 2011 ) [ arxiv:1102.4732 [ hep - ph ] ] ; d. parganlija , acta phys .
supp . * 4 * , 727 ( 2011 ) [ arxiv:1105.3647 [ hep - ph ] ] ; d. parganlija , arxiv:1109.4331 [ hep - ph ] . c. morningstar and m. j. peardon , aip conf .
proc . * 688 * , 220 ( 2004 ) [ arxiv : nucl - th/0309068 ] ; m. loan , x. q. luo and z. h. luo , int .
j. mod .
a * 21 * , 2905 ( 2006 ) [ arxiv : hep - lat/0503038 ] ; e. b. gregory , a. c. irving , c. c. mcneile , s. miller and z. sroczynski , pos * lat2005 * , 027 ( 2006 ) [ arxiv : hep - lat/0510066 ] ; y. chen _ et al .
_ , phys .
d * 73 * , 014516 ( 2006 ) [ arxiv : hep - lat/0510074 ] ; c. m. richards _
[ ukqcd collaboration ] , phys .
d * 82 * , 034501 ( 2010 ) [ arxiv:1005.2473 [ hep - lat ] ] . c. amsler and f. e. close , phys .
d * 53 * , 295 ( 1996 ) [ arxiv : hep - ph/9507326 ] ; w. j. lee and d. weingarten , phys .
d * 61 * , 014015 ( 2000 ) [ arxiv : hep - lat/9910008 ] ; f. e. close and a. kirk , eur .
j. c * 21 * , 531 ( 2001 ) [ arxiv : hep - ph/0103173 ] ; f. giacosa , t. gutsche , v. e. lyubovitskij and a. faessler , phys . rev .
d * 72 * , 094006 ( 2005 ) [ arxiv : hep - ph/0509247 ] ; f. giacosa , t. gutsche , v. e. lyubovitskij and a. faessler , phys .
b * 622 * , 277 ( 2005 ) [ arxiv : hep - ph/0504033 ] ; f. giacosa , t. gutsche and a. faessler , phys .
c * 71 * , 025202 ( 2005 ) [ arxiv : hep - ph/0408085 ] ; h. y. cheng , c. k. chua and k. f. liu , phys . rev .
d * 74 * , 094005 ( 2006 ) [ arxiv : hep - ph/0607206 ] ; l. bonanno and a. drago , phys .
c * 79 * , 045801 ( 2009 ) [ arxiv:0805.4188 [ nucl - th ] ] ; v. mathieu , n. kochelev and v. vento , int .
j. mod .
e * 18 * , 1 ( 2009 ) [ arxiv:0810.4453 [ hep - ph ] ] .
c. rosenzweig , a. salomone and j. schechter , phys .
d * 24 * , 2545 ( 1981 ) ; a. salomone , j. schechter and t. tudron , phys . rev .
d * 23 * , 1143 ( 1981 ) ; c. rosenzweig , a. salomone and j. schechter , nucl .
b * 206 * , 12 ( 1982 ) [ erratum - ibid .
b * 207 * , 546 ( 1982 ) ] ; a. a. migdal and m. a. shifman , phys .
b * 114 * , 445 ( 1982 ) ; h. gomm and j. schechter , phys .
b * 158 * , 449 ( 1985 ) ; r. gomm , p. jain , r. johnson and j. schechter , phys .
d * 33 * , 801 ( 1986 ) .
d. black , a. h. fariborz , f. sannino and j. schechter , phys . rev .
d * 59 * , 074026 ( 1999 ) [ hep - ph/9808415 ] ; a. h. fariborz , r. jora and j. schechter , phys .
d * 72 * , 034001 ( 2005 ) [ hep - ph/0506170 ] ; a. h. fariborz , r. jora and j. schechter , phys .
d * 76 * , 014011 ( 2007 ) [ hep - ph/0612200 ] ; a. h. fariborz , r. jora and j. schechter , phys .
d * 77 * , 034006 ( 2008 ) [ arxiv:0707.0843 [ hep - ph ] ] ; a. h. fariborz , r. jora and j. schechter , phys .
d * 79 * , 074014 ( 2009 ) [ arxiv:0902.2825 [ hep - ph ] ] ; a. h. fariborz , r. jora , j. schechter and m. n. shahid , phys .
d * 83 * , 034018 ( 2011 ) [ arxiv:1012.4868 [ hep - ph ] ] ; a. h. fariborz , r. jora , j. schechter and m. n. shahid , phys .
d * 84 * , 113004 ( 2011 ) [ arxiv:1106.4538 [ hep - ph ] ] ; t. k. mukherjee , m. huang and q. -s .
yan , arxiv:1203.5717 [ hep - ph ] . e. van beveren , t. a. rijken , k. metzger , c. dullemond , g. rupp and j. e. ribeiro , z. phys .
c * 30 * ( 1986 ) 615 [ arxiv:0710.4067 [ hep - ph ] ] ; n. a. tornqvist , z. phys .
c * 68 * ( 1995 ) 647 [ hep - ph/9504372 ] ; m. boglione and m. r. pennington , phys .
d * 65 * ( 2002 ) 114010 [ hep - ph/0203149 ] ; e. van beveren , d. v. bugg , f. kleefeld and g. rupp , phys .
b * 641 * ( 2006 ) 265 [ hep - ph/0606022 ] ; j. a. oller , e. oset and j. r. pelaez , phys .
d * 59 * ( 1999 ) 074001 [ erratum - ibid .
d * 60 * ( 1999 errat , d75,099903.2007 ) 099906 ] [ arxiv : hep - ph/9804209 ] .
v. a. polychronakos
_ et al . _ ,
d * 19 * , 1317 ( 1979 ) ; a. b. wicklund , d. s. ayres , d. h. cohen , r. diebold and a. j. pawlicki , phys . rev .
* 45 * , 1469 ( 1980 ) ; a. etkin _ et al .
_ , phys .
d * 25 * , 1786 ( 1982 ) ; b. v. bolonkin _ et al .
_ , yad . fiz .
* 46 * , 799 ( 1987 ) [ nucl . phys .
b * 309 * , 426 ( 1988 ) ] ; g. d. tikhomirov , i. a. erofeev , o. n. erofeeva and v. n. luzin , phys . atom . nucl . * 66 * , 828 ( 2003 ) [ yad .
fiz . * 66 * , 860 ( 2003 ) ] ; v. v. vladimirsky _ et al .
_ , phys . atom
. nucl . * 69 * , 493 ( 2006 ) [ yad .
fiz . * 69 * , 515 ( 2006 ) ] .
m. s. bhagwat , l. chang , y. -x .
liu , c. d. roberts and p. c. tandy , phys .
c * 76 * , 045203 ( 2007 ) [ arxiv:0708.1118 [ nucl - th ] ] ; f. ambrosino , a. antonelli , m. antonelli , f. archilli , p. beltrame , g. bencivenni , s. bertolucci and c. bini _ et al .
_ , jhep * 0907 * , 105 ( 2009 ) [ arxiv:0906.3819 [ hep - ph ] ] ; g. amelino - camelia , f. archilli , d. babusci , d. badoni , g. bencivenni , j. bernabeu , r. a. bertlmann and d. r. boito _ et al . _ ,
j. c * 68 * , 619 ( 2010 ) [ arxiv:1003.3868 [ hep - ex ] ] ; m. c. chang , y. c. duh , j. y. lin , i. adachi , k. adamczyk , h. aihara , d. m. asner and t. aushev _ et al .
_ , phys .
d * 85 * , 091102 ( 2012 ) [ arxiv:1203.3399 [ hep - ex ] ] .
r. k. carnegie , r. j. cashmore , w. m. dunwoodie , t. a. lasinski and d. w. g. leith , phys .
b * 68 * , 287 ( 1977 ) ; j. l. rosner , comments nucl .
* 16 * , 109 ( 1986 ) ; n. isgur and m. b. wise , phys .
b * 232 * , 113 ( 1989 ) ; h. g. blundell , s. godfrey and b. phelps , phys . rev .
d * 53 * , 3712 ( 1996 ) [ arxiv : hep - ph/9510245 ] ; f. e. close and a. kirk , z. phys .
c * 76 * , 469 ( 1997 ) [ arxiv : hep - ph/9706543 ] ; d. m. li , h. yu and q. x. shen , chin .
* 17 * , 558 ( 2000 ) [ arxiv : hep - ph/0001011 ] ; d. m. asner _
[ cleo collaboration ] , phys .
d * 62 * , 072006 ( 2000 ) [ arxiv : hep - ex/0004002 ] ; w. s. carvalho , a. s. de castro and a. c. b. antunes , j. phys .
a * 35 * , 7585 ( 2002 ) [ arxiv : hep - ph/0207372 ] ; h. y. cheng , phys . rev .
d * 67 * , 094007 ( 2003 ) [ arxiv : hep - ph/0301198 ] ; t. barnes , n. black and p. r. page , phys .
d * 68 * , 054014 ( 2003 ) [ arxiv : nucl - th/0208072 ] ; d. m. b. li , b. ma , y. x. li , q. k. yao and h. yu , eur .
j. c * 37 * , 323 ( 2004 ) [ arxiv : hep - ph/0408214 ] ; j. vijande , f. fernandez and a. valcarce , j. phys .
g * 31 * , 481 ( 2005 ) [ arxiv : hep - ph/0411299 ] .
d. m. li , b. ma and h. yu , eur .
j. a * 26 * , 141 ( 2005 ) [ arxiv : hep - ph/0509215 ] ; d. m. li and z. li , eur . phys .
j. a * 28 * , 369 ( 2006 ) [ arxiv : hep - ph/0606297 ] ; h. hatanaka and k. c. yang , phys . rev .
d * 77 * , 094023 ( 2008 ) [ erratum - ibid .
d * 78 * , 059902 ( 2008 ) ] [ arxiv:0804.3198 [ hep - ph ] ] ; h. y. cheng and c. k. chua , phys .
d * 81 * , 114006 ( 2010 ) [ erratum - ibid .
d * 82 * , 059904 ( 2010 ) ] [ arxiv:0909.4627 [ hep - ph ] ] ; k. -c .
yang , phys .
d * 84 * , 034035 ( 2011 ) [ arxiv:1011.6113 [ hep - ph ] ] ; h. -y .
cheng , phys .
b * 707 * , 116 ( 2012 ) [ arxiv:1110.2249 [ hep - ph ] ] .
r. l. jaffe , phys .
d * 15 * , 267 ( 1977 ) ; r. l. jaffe , phys .
d * 15 * , 281 ( 1977 ) ; l. maiani , f. piccinini , a. d. polosa and v. riquer , phys .
93 * ( 2004 ) 212002 [ arxiv : hep - ph/0407017 ] ; m. napsuciale and s. rodriguez , phys . rev .
d * 70 * , 094043 ( 2004 ) [ arxiv : hep - ph/0407037 ] ; f. giacosa , phys .
d * 74 * , 014028 ( 2006 ) [ arxiv : hep - ph/0605191 ] . c. michael _ et al . _
[ etm collaboration ] , jhep * 1008 * , 009 ( 2010 ) [ arxiv:1004.4235 [ hep - lat ] ] ; r. baron , p. .boucaud , j. carbonell , a. deuzeman , v. drach , f. farchioni , v. gimenez and g. herdoiza _ et al .
_ , jhep * 1006 * , 111 ( 2010 ) [ arxiv:1004.5284 [ hep - lat ] ] ; r. baron _ et al . _
[ european twisted mass collaboration ] , comput .
* 182 * , 299 ( 2011 ) [ arxiv:1005.2042 [ hep - lat ] ] ; m. wagner [ etm and y collaborations ] , pos lattice * 2010 * , 162 ( 2010 ) [ arxiv:1008.1538 [ hep - lat ] ] ; r. baron _ et al . _
[ etm collaboration ] , arxiv:1009.2074 [ hep - lat ] .
r. baron _ et al .
_ [ etm collaboration ] , pos lattice * 2010 * , 123 ( 2010 ) [ arxiv:1101.0518 [ hep - lat ] ] ; m. wagner [ etm collaboration ] , acta phys .
supp . * 4 * , 747 ( 2011 ) [ arxiv:1103.5147 [ hep - lat ] ] . t. sakai and s. sugimoto , prog .
phys . * 113 * , 843 ( 2005 ) [ hep - th/0412141 ] ; t. sakai and s. sugimoto , prog .
phys . * 114 * , 1083 ( 2005 ) [ hep - th/0507073 ] ; d. li , m. huang and q. -s .
yan , arxiv:1206.2824 [ hep - th ] .
a. schmitt , s. stetina and m. tachibana , phys .
d * 83 * , 045008 ( 2011 ) [ arxiv:1010.4243 [ hep - ph ] ] ; f. preis , a. rebhan and a. schmitt , jhep * 1103 * , 033 ( 2011 ) [ arxiv:1012.4785 [ hep - th ] ] ; f. preis , a. rebhan and a. schmitt , j. phys .
g g * 39 * , 054006 ( 2012 ) [ arxiv:1109.6904 [ hep - th ] ] . | we study scalar , pseudoscalar , vector , and axial - vector mesons with non - strange and strange quantum numbers in the framework of a linear sigma model with global chiral @xmath0 symmetry .
we perform a global fit of meson masses , decay widths , as well as decay amplitudes .
the quality of the fit is , for a hadronic model that does not consider isospin - breaking effects , surprisingly good .
we also investigate the question whether the scalar @xmath1 states lie below or above 1 gev and find the scalar states above 1 gev to be preferred as @xmath1 states .
additionally , we also describe the axial - vector resonances as @xmath1 states . |
proper interpretation of results obtained from radial velocity ( rv ) survey of gk giants requires detailed knowledge of their physical parameters . since gk - giants are known to exhibit various types of intrinsic variability induced by pulsations and/or surface inhomogeneity connected with rotation these effects have to be ruled out before substellar companion interpretation can be considered .
effective temperatures and gravitational accelerations are needed to obtain estimates of stellar radii and together with the rotational periods address the influence of possible starspots on results .
additional knowledge of metallicities allows for stellar masses estimates that are crucial for interpretation of the observed rv periodical variations and determination of companion s masses .
the observational material and data reduction are described in niedzielski & wolszczan ( this volume ) .
the atmospheric parameters of program stars were obtained with the purely spectroscopic method @xcite based on analysis of fe i and fe ii lines and assumption of lte . to test reliability of our determinations
we used the procedure to obtain parameters of 8 stars of @xcite .
comparison of results shows that @xmath3 agrees within @xmath4 , @xmath5 within @xmath6 and @xmath7 $ ] within @xmath6 .
these values are comparable to our intrinsic error estimates , except for metallicity , which is systematically higher than literature values .
typical intrinsic uncertainties of these parameters are @xmath8 , @xmath9 , @xmath10 and @xmath11 } = 0.06 $ ] . in table 1
we present results of our determinations of atmospheric parameters for 16 stars , targets of our survey .
the metalicities presented in table 1 are corrected for the systematic effect .
stellar masses were obtained by comparing the position of stars on the hr diagram with the theoretical evolutionary tracks of @xcite and @xcite for given metallicity .
stellar radii were determined using the calibration given in @xcite .
the uncertainties of stellar masses obtained by comparison with theoretical evolutionary tracks depend on the accuracy of the position of a star on hr diagram and metalicity . for stars for which the parallax is
precise enough the metallicity may introduce significant uncertainty in mass .
we assume that for an average red giant with known parallax the mass may be estimated within @xmath12 .
l|c|c|c|c|c|c|c name & @xmath3 & @xmath5 & @xmath13 & @xmath7 $ ] & @xmath14 & @xmath15 & @xmath16 + & @xmath17 & & @xmath18 & & @xmath19 & & + hd 17028 & 4246 & 2.43 & 1.52 & 0.11 & -2.89 & 5.5 & 27.4 + hd 17092 & 4649 & 3.00 & 1.27 & 0.22 & 0.80 & 2.3 & 10.9 + hd 77819 & 5007 & 2.92 & 1.19 & -0.11 & 1.01 & 2.5 & 9.4 + hd 95296 & 4603 & 2.23 & 1.36 & -0.08 & -0.28 & 2.7 & 10.9 + hd 96127 & 4319 & 2.65 & 1.23 & 0.35 & -2.46 & 5.0 & 22.1 + hd 102103 & 4565 & 2.54 & 1.44 & -0.04 & -0.63 & 3.0 & 12.8 + hd 102272 & 4908 & 3.07 & 1.30 & -0.26 & 1.43 & 1.7 & 10.1 + hd 109402 & 5036 & 4.03 & 0.80 & -0.11 & 3.34 & 1.3 & 8.6 + hd 112914 & 4975 & 4.86 & 0.01 & -0.25 & 5.22 & 0.8 & - + hd 134901 & 5427 & 3.91 & 0.86 & -0.22 & 3.09 & 1.3 & 3.4 + hd 160723 & 4775 & 2.67 & 1.29 & -0.30 & 0.80 & 2.2 & 10.3 + hd 257498 & 4773 & 2.94 & 1.33 & -0.19 & 5.85 & 0.8 & 10.3 + bd + 03 2562 & 4365 & 2.95 & 1.29 & -0.12 & 0.40 & 1.4 & 22.1 + bd + 20 2457 & 4352 & 2.19 & 1.64 & -0.65 & 0.40 & 0.8 & 22.1 + bd + 57 144 & 5010 & 3.09 & 1.28 & -0.08 & -0.22 & 3.3 & 9.4 + sao 112133 & 4993 & 3.56 & 1.50 & 0.18 & 0.80 & 2.1 & 9.4 + we thank y. takeda for making his code available for us .
the project is supported in part by the polish ministry of science and higher education grant 1p03d 007 30 .
pz was supported by polish ministry of science and higher education grant spb 104/e-337/6 .
alonso , a. , salaris , m. , et al . 2000 , a&a , 355 , 1060 butler , r. p. , et al .
2006 , , 646 , 505 girardi , l. , bressan , a. , bertelli , g. , chiosi , c. 2000 , a&as , 141 , 371 salasnich , b. , girardi , l. , weiss , a. , chiosi , c. 2000 , a&a , 361 , 1023 takeda , y. , ohkubo , m. , sadakane , k. 2002 , pasj , 54 , 451 takeda , y. , ohkubo , m. , sato , b. , kambe , e. , sadakane , k. 2005 , pasj , 57 , 27 | the objective of the psu / tcfa search for planets around evolved stars is to study evolution of planetary systems in the stellar evolution timescale .
for such an analysis precise physical parameters of the hosts of the planetary systems are essential . in this paper
we present an attempt to obtain basic physical parameters for a sample of evolved stars observed within our survey with the high resolution spectrograph of the hobby - eberly telescope . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ pawe zieliski@xmath0 and andrzej niedzielski@xmath1 _ @xmath0torun centre for astronomy , nicolaus copernicus university , + ul .
gagarina 11 , 87 - 100 torun , poland _ _ @xmath2department of astronomy and astrophysics , pennsylvania state university , 525 davey laboratory , university park , pa 16802 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ |
recently , we have published our discoveries of a previously unknown galactic supernova remnant ( snr ) ( aschenbach 1998 , iyudin et al .
the x - ray image obtained in the rosat all - sky survey shows a disk - like , partially limb brightened emission region of 2@xmath7 in diameter , which is the typical appearance of a shell - like snr ( cf .
[ picture ] ) .
the pspc x - ray spectra reveal rather high temperatures of @xmath8 3@xmath510@xmath9 k , which indicate that rx j0852.0 - 4622 is a young object . combining the low age and the 2@xmath7 angular extent it
is concluded that rx j0852.0 - 4622 is relatively close - by .
comparison with historical snrs limits the age to about @xmath0 1500 yrs and the distance to @xmath101 kpc .
the case of rx j0852.0 - 4622 being an snr was clinched by the detection of @xmath1-ray line emission from @xmath2ti , which is a titanium isotope exclusively produced in supernovae .
the centre of the @xmath2ti source , called gro j0852 - 4642 , is off - set from the center of rx j0852.0 - 4622 by 0.4@xmath7 , but this is significantly less than the angular resolution of the comptel instrument , so that rx j0852.0 - 4622 and gro j0852 - 4642 are considered to be the same object . using a weighted mean lifetime of @xmath2ti of 90.4 yrs , the angular diameter and adopting a mean expansion velocity of 5000 km / s as well as a @xmath2ti yield of 5@xmath510@xmath11 m@xmath12 age and distance are uniquely determined to @xmath0680 yrs and @xmath0200 pc , respectively . therefore
, rx j0852.0 - 4622/gro j0852 - 4642 could be the nearest supernova to earth to have occured during recent human history .
the discovery of rx j0852.0 - 4622 and the interpretation as an snr was made by one of us ( ba ) in early 1996 . during the time which followed it
was attempted to associate some fraction of the @xmath3al @xmath1-ray line emission from the vela snr region measured by comptel ( oberlack et al .
1994 , diehl et al .
1995 ) with rx j0852.0 - 4622 .
the results have not been conclusive basically because of the unknown distance of rx j0852.0 - 4622 ( oberlack 1997 ) .
the discovery of @xmath2ti @xmath1-ray line emission , however , made it clear that rx j0852.0 - 4622 is indeed a nearby object , so that we could take up again the discussion of the association of @xmath3al emission with rx j0852.0 - 4622 .
for example the combination of just the @xmath3al and @xmath2ti data allow to derive a distance independent estimate of the age of gro j0852 - 4642 .
furthermore , if a major fraction of the vela @xmath3al mass would be associated with the snr , a type ia supernova is excluded within the framework of current explosion models . under the assumption of adiabatic expansion ( sedov - like ) of the snr we give an estimate of the supernova explosion energy e@xmath13 related to the progenitor star and the ambient matter density n@xmath13 .
the uncertainties in the determination of age and distance by exploiting the x - ray spectra are discussed to come up with a time span in which to search for the historical supernova event . in a recent paper chen & gehrels ( 1999 )
conclude that rx j0852.0 - 4622 was created by a core - collapse supernova of a massive star .
their analysis is based on the x - ray data and @xmath1-ray data published published earlier by us ( aschenbach 1998 , iyudin et al . 1998 ) .
we discuss their approach and conclusions in the relevant section .
basis for the determination of the age and distance of the new snr is the law of radioactive chain - decay @xmath2ti @xmath14sc @xmath14ca , for which we have @xmath15 with the definitions : _ f _ photon flux density , _ d _ distance , _
y@xmath16 _ mass yield of the element a , _
m@xmath16 _ its atomic mass , @xmath17 its mean life time and _ t _ the age . for the @xmath2ti decay chain @xmath18 , so that we can neglect @xmath19 and the second exponential term of eq .
1 . the data available suggest @xmath20 90 yrs , which we adopt for the following .
this value is also close to the result of ( 87.7 @xmath41.7 ) yrs recently published by ahmad et al .
_ f _ is the flux of the 1.157 mev line which has been measured by iyudin et al .
( 1998 ) to ( [email protected])@xmath2110@xmath11 photons @xmath22 s@xmath23 .
apart from the statistical error a systematic error should be added , which is estimated to @xmath4 20% . the range of @xmath24 , @xmath25 and _ y@xmath16 _ of eq .
1 can be constrained by introducing the x - results .
the angular radius @xmath26 = 1@xmath7 is related to @xmath24 and @xmath25 by @xmath26 = @xmath27 / _ d _ , with @xmath28 the mean expansion velocity of the snr . by substituting @xmath24 or @xmath25 of eq .
1 by @xmath28 a quantitative relation between @xmath29 and @xmath25 or @xmath29 and @xmath24 can be derived with @xmath28 as a parameter .
an estimate of @xmath28 can be obtained from the x - ray spectra .
the analysis of the rosat x - ray spectra is affected by the presence of the low energy emission of the vela snr , which is aggravated by the large size of rx j0852.0 - 4622 .
but an archival rosat pspc pointing observation centred on the southeastern limb of the vela snr with an exposure of 11000 s happens to contain the northern limb section of rx j0852.0 - 4622 .
the number of counts is sufficient to extract a small section of the limb as well as a small section of the vela snr offset by just 10 arcmin to create both uniform source and background spectra .
limiting the analysis to these small regions , each 10 arcmin @xmath5 10 arcmin in size , reduces the impact of any spectral and spatial non - uniformity across the source and the background
. fits to the residual northern limb spectrum were performed with a two component , optically thin thermal emission equilibrium model ( raymond - smith model ) with kt@xmath30 = 0.21@xmath31 kev , kt@xmath32 = 4.7@xmath33 kev and an absorption column density of n@xmath34 = @xmath352.3@xmath3610@xmath37 @xmath22 .
we note that the data can be fit equally well ( @xmath38 1 ) with a straight power law with index @xmath39 = 2.6@xmath40 and absorption column density n@xmath41 = @xmath35 1.2@xmath4210@xmath43 @xmath22 .
the spectrum of the rest of the snr can be obtained only from the rosat all - sky survey data and because of the relatively low exposure the spatial region selected for analysis needs to be large which increases the uncertainty in assessing the background level from the vela snr . for the full remnant excluding the bright northern limb an acceptable fit with @xmath38 1
is obtained with a two temperature model with kt@xmath44 = 0.14@xmath45 kev , kt@xmath46 = 2.5@xmath33 kev and n@xmath47 = @xmath354.0@xmath4810@xmath37 @xmath22 .
the matter density of the shock - wave heated snr plasma can be derived from the observed x - ray flux @xmath49 via the relation @xmath49 = @xmath50 .
@xmath51 is the cooling function for the best - fit values of @xmath52 ; @xmath53 is the electron number density and @xmath54 is the number density of the un - shocked matter , intitially uniformly distributed in a sphere .
furthermore , a factor of four has been used for the density jump at the shock .
the low and high temperature components are associated with densities of @xmath55 = 0.6@xmath5d@xmath56 @xmath57 and @xmath58 = 0.06@xmath5d@xmath56 @xmath57 , respectively , with d@xmath59 measured in units of 200 pc . despite the acceptable spectral
fit the value of @xmath55 and the column density are quite uncertain , as the low temperature component could be significantly affected by the vela snr radiation , which is , however , not the case for @xmath58 . as usual for thermal snrs
two components with different temperatures are needed for equilibrium models to fit the observed spectrum .
if the plasma is far from ionization equilibrium the low temperature component appears as an artifact because of the under - ionization .
the time - scale to reach ionization equilibrium is about 10@xmath60 s @xmath57/n@xmath61 , with n@xmath61 the electron density of the radiating plasma in units of @xmath57 . with @xmath25 = 680 yrs and the densities given above 1.5@xmath510@xmath62 s @xmath63 n@xmath6410@xmath65 s @xmath57 , which demonstrates that rx j0852.0 - 4622 departs significantly from ionization equilibrium .
clearly the high temperatures observed are closer to the real electron temperature .
but even the high temperatures may underestimate the average temperature of the electrons and ions if the electrons are heated mainly by coulomb collisons with the ions , which occurs on a timescale similar to that of reaching ionization equlibrium .
the x - ray temperatures which have been produced by shock wave heating can be used to estimate the velocity @xmath66 of the shock wave : @xmath67@xmath68 ; @xmath69 is the proton mass and @xmath70 is the mean molecular weight , which is 0.6 for a fully ionized plasma of cosmic abundances . again , this relation is for a density jump of a factor of four at the shock . discarding the low temperature components as argued above the x - ray temperatures stretch from 1.8 kev @xmath71 kt@xmath59 @xmath71 9.2 kev including @xmath41-@xmath72 errors .
the mean kt@xmath59 , which is consistent at the same significance level with both the radiation from the bulk of the snr and its northern limb section , taking the thermal option , is kt@xmath59 = 4.4 kev .
the corresponding best estimates of the minimal and maximal shock velocities using the relation above are 1940 km / s , 1240 km / s and 2800 km / s , respectively .
the current shock velocity @xmath66 is related to the mean expansion velocity @xmath28 by the past temporal evolution of the snr . with the limited observations available
we are forced to rely on what is known about historical remnants .
the compilation of strom ( 1994 ) provides both maximal internal shock velocity and mean expansion velocity and the ratio @xmath73 is 1.5 for the crab nebula and cas a , 2.5 for sn 1006 and 3.5 for the kepler and tycho snrs . for a purely adiabatic expansion in a uniform medium of constant matter density ( the sedov description ) @xmath73 = 2.5 .
as strom has pointed out the observed maximal internal velocities may not be strictly related to @xmath66 but they provide a reasonable estimate .
more recently measurements of the expansion rate in the x - ray band have become available by comparing images obtained with the einstein and rosat observatories or even just the rosat images taken at different epochs . both koralesky et al .
( 1998 ) and vink et al .
( 1998 ) have found an expansion rate of cas a of 0.002% yr@xmath23 which corresponds to a factor of @xmath0 1.55 for the ratio of mean expansion rate over current expansion rate .
hughes ( 1996 ) has found a similar value for the tycho snr . in each of these cases , however , the current expansion velocities , using reasonable distance estimates , are significantly larger than the x - ray spectra and temperatures indicate .
therefore a factor of 1.5 for @xmath73 is a very conservative lower limit to estimate @xmath28 from x - ray spectra . for a worst case estimate
we define a velocity range for rx j0852.0 - 4622 applying factors of 1.5 , 2.5 , 3.5 to the minimal , best - estimate and maximal @xmath66 , respectively , which leads to a best - estimate expansion velocity @xmath74 = 5000 km / s bracketed by a minimal expansion velocity of @xmath75 = 2000 km / s and a maximal expansion velocity @xmath76 = 10000 km / s ; the values have been rounded off slightly .
since we do nt know whether the x - ray temperatures are associated with either the blast wave heated ambient medium or the progenitor ejecta heated by reverse shocks , the expansion velocities derived may even be lower limits .
similarly , the values are too low if the electrons have not reached thermal equilibrium with the ions . for the discussion of the impact of the expansion velocity on age and distance we note that the @xmath2ti line appears to be broadened ( iyudin et al .
1998 ) , the origin of which is unknown . in the most extreme case that the width is exclusively attributed to doppler broadening
the associated velocity is ( 15300@xmath43700 ) km / s with an upper limit of @xmath77 = 19000 km / s .
we include @xmath77 for the sake of completeness , but we stress that the use of @xmath77 for constraining the type of progenitor is overinterpretating the @xmath1-ray data and it is essentially misleading .
nevertheless , we add that nagataki ( 1999 ) quotes @xmath78 expansion velocities @xmath7912000 km / s for sub - chandrasekhar mass models of sne ia .
[ age ] shows the relation between @xmath29 and the age @xmath25 of rx j0852.0 - 4622 parametrized by @xmath28 .
despite the large uncertainty of @xmath28 , @xmath25 is determined to within @xmath4 100 yrs for fixed @xmath29 using @xmath28 of the x - ray data . for @xmath29 = 5@xmath510@xmath11 m@xmath12 and @xmath28 = @xmath80 , @xmath25 = 680 yrs and @xmath24 = 200 pc ( c.f .
[ distance ] ) .
age and distance are rather insensitive to the exact value of the @xmath2ti @xmath1-ray line flux . a total error of the flux of @xmath440% , which is the sum of the statistical error and the systematic error , broadens the range of @xmath25 by @xmath440 yrs and that of @xmath24 by @xmath4 10 pc .
if rx j0852.0 - 4622 is expanding as fast as @xmath77 indicates the age would be as low as @xmath0500 yrs .
model calculations provide a range for @xmath29 , which runs for symmetric core - collapse supernovae from 1.4@xmath510@xmath11 m@xmath12 to 2.3@xmath510@xmath81 m@xmath12 , depending on progenitor mass ( woosley & weaver 1995 , thielemann et al .
1996 ) . within this range ( @xmath82 ) is within ( 500 yrs , 400 pc ) and ( 950 yrs , 80 pc ) .
nagataki et al .
( 1998 ) have pointed out that @xmath29 could be much higher in an axisymmetric collapse - driven supernova .
for example , in their model with the highest degree of asymmetry they obtain @xmath29 = 5.1@xmath510@xmath81 for @xmath83 = 0.07 m@xmath12 .
this value of @xmath29 would allow an age of up to 1000 yrs . for type ia supernovae
we find 7.9@xmath510@xmath84 m@xmath12 @xmath85 @xmath71 4.7@xmath510@xmath11 m@xmath12 for carbon deflagration models ( nomoto et al .
1984 , iwamoto et al . 1999 ) , which do not significantly differ from the core - collapse sne , and accordingly the range for ( @xmath82 ) is not affected .
values of @xmath29 as high as 2@xmath510@xmath86 m@xmath12 are obtained in he - detonation models ( woosley & weaver 1994 ) .
this would allow ( @xmath82 ) to lie between ( 850 yrs , 500 pc ) and ( 1100 yrs , 100 pc ) .
in summary , for any @xmath29 given by current models the upper limit of the distance of rx j0852.0 - 4622 is 500 pc and 1100 years for the age .
chen & gehrels ( 1999 ) have also used the x - ray temperature to derive age and distance , although in a slightly different manner .
they use the temperature derived from the rosat data for the central region ( aschenbach 1998 ) which might not be representative for the current expansion velocity at the rim , why we prefer a somewhat higher velocity consistent with the apparently higher temperature observed at the limb . for the mean expansion velocity , which should be higher than the current expansion velocity by some factor ,
chen & gehrels derive a range of 2000 5000 km / s , whereas we propose a range of 2000 km / s to 10000 km / s by comparison with observational data obtained for the historical remnants . for given velocity and @xmath29
our results agree with those obtained by chen & gehrels , but our estimates allow a wider range of age and distance .
@xmath2ti decays by electron capture , which means that lifetime depends on ionization stage , in particular to what extent the k shell is populated .
the @xmath2ti lifetime of @xmath87 90 yrs is the mean lifetime for two electrons in the k - shell irrespective of the number of electrons in the higher shells . for just one electron in the k - shell , the hydrogen - like state @xmath2ti@xmath88
, the lifetime is expected to be about twice as long , and for the fully ionized atom @xmath2ti@xmath89 the lifetime is @xmath90 .
1 gives the decay rate for the ionic fraction @xmath91ti@xmath92 = 0 ( full ionization ) , @xmath91ti@xmath93 = 0 ( one electron in the k - shell ) and @xmath91ti@xmath94 = 1 . for @xmath91ti@xmath94@xmath10 1 eq .
1 is modified by introducing the ionic fraction @xmath91ti@xmath94 with @xmath95 = 90 yrs and @xmath91ti@xmath93 with @xmath95 = 2 @xmath5 90 yrs ; the impact of @xmath91ti@xmath92 has been neglected because of its comparatively low contribution to @xmath96 . the solution for @xmath25 of eq . 1 for either the ionization or
the no - ionization case is done with the same @xmath96 .
as before also @xmath24 and @xmath25 are not independent of each other but constrained by @xmath26 and @xmath28 , which means that not only @xmath25 but also @xmath24 is to change for the ionization case compared to the no - ionization case .
so the comparison is done with the same @xmath28 but not with the same @xmath24 .
furthermore @xmath28 is constrained by the x - ray spectra and the impact of the uncertainty of @xmath28 on @xmath24 and @xmath25 has been given in the previous section . if @xmath25 = @xmath97 for @xmath98 = 1 and @xmath25 = @xmath99 for @xmath1001 eq .
2 describes the change of the age in terms of @xmath101 .
@xmath102\ ] ] @xmath103\ ] ] the ionization stage of @xmath2ti of gro j0852 - 4642 is not yet known , but a case study is useful to demonstrate quantitatively the impact of the ionization on the estimate of @xmath25 and @xmath24 .
if @xmath2ti would have been heated to around k@xmath104 = 4.4 kev like the x - ray emitting plasma , e.g. by a reverse shock propagating in the ejecta and if @xmath2ti is in ionization equilibrium the ionic fractions can be extracted from literature .
titanium has not been tabulated so far but the distributions of the ionic fraction of calcium and iron are available , which are taken as case representative examples .
arnaud & rothenflug ( 1985 ) , for instance , computed @xmath105 = 0.086 , @xmath106 = 0.339 and 57.5% of ca completely ionized for log t = 7.8 . using eq .
2 @xmath97 = 680 yrs increases to @xmath99 = 930 yrs as does @xmath24 by the same factor of @xmath107 . for iron , which has @xmath108 = 0.686 and @xmath109 = 0.269 at log t = 7.8 , @xmath99 = 900 yrs , which is very close to the result obtained for ca , despite a significantly different distribution of the ionic fractions . for higher temperatures , e.g. log t = 8.5 , @xmath105 = 0
, @xmath106 = 0.0465 and 95.4% of ca is completely ionized . for this ionic fraction distribution @xmath107 = 1 and @xmath25 and @xmath24
are unchanged although only 4.65% of the total @xmath110 decays radioactively .
for even lower values of @xmath111 , i.e. a larger fraction of totally ionized @xmath2ti , @xmath112 or the age becomes even lower . using the distribution of the ionic fractions of iron at log t = 8.5 , @xmath107 = 1.28 .
clearly , the ionization of ti has an impact but of moderate size .
values of @xmath25 and @xmath24 may be underestimated by some 30% when the ionization starts to affect the k - shell population and they may be even unchanged if only some 10% or less of the ti has just one electron in the k - shell but is otherwise completely ionized .
quite recently mochizuki et al .
( 1999 ) have modelled the heating and ionization of @xmath2ti by the reverse shock in cas - a , for which they report the possibility of a currently increased @xmath2ti activity .
with respect to rx j0852.0 - 4622 / gro j0852 - 4642 they find that the reverse shock does not heat the ejecta to sufficiently high temperatures to ionize @xmath2ti because of the low ambient matter density .
future x - ray spectroscopy measurements may answer the question of ionization .
but independent of the outcome this section shows that even if ionization were significant it does not change the conclusion that rx j0852.0 - 4622 / gro j0852 - 4642 is a young nearby snr .
the sedov relation @xmath114 , which has been adopted for describing the adiabatic expansion of an snr of radius r in a homogenous medium of matter density @xmath115 , has been used quite often in the past to estimate the explosion energy e@xmath113 associated with the supernova ( winkler & clark 1974 , pfeffermann et al .
the limitations of this approach are well known .
the x - ray spectra provide @xmath116 , from which @xmath28 is derived , the x - ray flux is proportional to @xmath117 via @xmath118 ( cf .
2.2 ) and the x - ray image shows the angular extent @xmath26 . with the sedov relation @xmath119 is not yet uniquely determined but can then be expressed as a function of a single variable , for instance @xmath25 . since for rx j0852.0 - 4622 @xmath25
can be related to @xmath29 via eq .
( 1 ) , @xmath119 is a function of @xmath29 .
in contrast to the relation @xmath120 , @xmath121 is constrained because of the limited range of @xmath29 , at least towards the higher end . fig .
[ energy ] shows @xmath121 for various @xmath28 , because @xmath28 is not uniquely determined by the x - ray spectra . for the reference values of @xmath29
= 5@xmath510@xmath11 m@xmath12 and @xmath28 = @xmath80 = 5000 km / s , @xmath119 = 2.6@xmath510@xmath122 , which is a factor of about 40 less than the canonical @xmath119 = 10@xmath123 erg . with @xmath28 = 5000 km
/ s this value can not be reached with a realistic @xmath29 ; even a value of @xmath119 = 10@xmath124 erg is hardly consistent with reasonable @xmath29 values at @xmath28 = 5000 km / s .
it is interesting to note that a similarly low value of @xmath119 , i.e. @xmath119 @xmath8 4.4@xmath510@xmath122 erg s@xmath23 have been derived by willingale et al .
( 1996 ) for the snr of sn 1006 , with which rx j0852.0 - 4622 shares a number of other similarities like the x - ray appearance and the ratio of radio to x - ray surface brightness ( aschenbach 1998 ) .
for the reference value of @xmath24 = 200 pc the total swept - up mass of rx j0852.0 - 4622 is less than one solar mass , which means that the slow - down of the remnant expansion may not be dominated by @xmath115 , so that the applicability of the sedov relation may be questioned .
the radial evolution depends then on the details of the explosion rather than just on @xmath119 .
for instance , most of the kinetic energy of the sn may be in matter which does not radiate in x - rays .
@xmath119 could be raised by increasing @xmath115 .
the slow down could have occurred at times @xmath125 when regions of higher density might have been passed by the shock wave , e.g. if the progenitor star had produced a strong stellar wind . for a mass loss rate of 10@xmath11 m@xmath126yr@xmath23 and a wind velocity of 1000 km / s the wind number density would exceed 10@xmath127@xmath57 within a radius of about 6@xmath510@xmath128 cm and x - rays would have been emitted from this region .
lower wind velocities like those typical of red supergiants would increase the size accordingly .
the emission region would expand to a measurable size over the 700 yrs but both radiative cooling and adiabatic expansion are likely to have reduced the flux below the detection limit .
nevertheless , we point out that the rosat image shows weak but enhanced emission from the central 15@xmath129 diameter region ( c.f . fig .
[ picture ] ) . for higher values of @xmath28 , e.g. for @xmath28 = @xmath76 = 10000 km / s and @xmath29 = 5@xmath510@xmath11m@xmath12 , @xmath119 = 2.9@xmath510@xmath124
erg is relatively close to the canonical e@xmath113 , but the swept - up , x - ray radiating mass is still just 1 m@xmath12 . basically , because of the low @xmath49 and the maximal value of @xmath24 consistent with @xmath29 the swept - up , x - ray radiating mass never exceeds a few solar masses . in summary , e@xmath113 is not very sensitive to @xmath29 ( c.f . fig .
[ energy ] ) but instead to the mean expansion velocity @xmath28 . taking the full range of @xmath28 indicated by the x - ray spectra
it follows that 10@xmath122 erg @xmath10 e@xmath113 @xmath10 3@xmath510@xmath124 erg for a sedov - type expansion . the energy budget made of e@xmath113 and the kinetic and thermal energy observed can be used to constrain the mass of the progenitor .
the total energy e@xmath130 of the x - ray radiating mass , i.e. the sum of the kinetic energy and the thermal energy , amounts to e@xmath130 = 4@xmath510@xmath131 erg / s@xmath132 with @xmath133 in units of 1000 km / s . since the maximum velocity of the ejecta should not exceed the uniform expansion velocity _ v _ , which for the adiabatic case is 2.5@xmath134 , a lower limit of the ejecta mass m@xmath135 is @xmath136 with m@xmath135 in m@xmath12 and e@xmath137 in 10@xmath123 erg . for @xmath28 = @xmath80 = 5000
km / s @xmath138 4 m@xmath12 , i.e. a massive progenitor is required for e@xmath137 = 1 , whereas a low mass progenitor with m@xmath139 0.9 m@xmath12 is consistent with the data for @xmath28 = @xmath76 = 10000 km / s .
a more massive progenitor is required if the bulk of the ejecta mass is moving at significantly lower velocities .
another approach to constrain the progenitor and the supervova type has been taken by chen & gehrels ( 1999 ) .
they have used the shock wave velocity indicated by the x - ray temperature observed in the central region of the snr and used this as the current expansion velocity . by comparison of this velocity with
that predicted by sn explosion models and their subsequent evolution into an ambient medium of constant matter density , they conclude that the likely progenitor of the snr was a massive star of 15 m@xmath12 with a type ii explosion , solely based on the relatively low value of the current expansion velocity v@xmath140 inferred from the x - ray temperature .
lower mass progenitors like those leading to a sn of type ia are supposed to have significantly higher ejecta velocities and according to chen & gehrels an ambient matter density @xmath79 500 @xmath57 is required to decelerate the explosion wave from initially 11000 km / s to the current value of 1300 km / s , using the relation v@xmath141 t@xmath142 ( chen & gehrels , 1999 ) .
if we use the standard sedov - taylor relation of v@xmath141 t@xmath143 instead , i.e. the asymptotic limit of the evolution into a uniform medium of constant matter density , a much lower ambient density of 1.4 @xmath57 is sufficient to reduce the ejecta speed from 11000 km / s to v@xmath140 = 3900 km / s ( the upper limit of v@xmath140 estimated by chen & gehrels ) in 1000 years for an ejecta mass of one solar mass and e@xmath113 = 10@xmath123 erg .
although this ambient density still exceeds the observed value by a factor of @xmath030 it is not unreasonable in comparison with other snrs and ism densities .
different ejecta mass , explosion energy and non constant density distributions , in particular , might reduce the required matter density further .
in contrast to chen & gehrels we are therefore very reluctant to rule out a snia for rx j0852.0 - 4622 based on just the x - ray temperature . after the discovery of its x - ray emission in early 1996 it was attempted to identify rx j0852.0 - 4622 as a source contributing to the 1.8 mev @xmath3al @xmath1-ray line emission from the vela region , which had been mapped with the comptel instrument ( oberlack et al .
1994 , diehl et al .
because of its identifaction as an snr and because of its apparently low distance rx j0852.0 - 4622 was considered a good candidate to provide a measurable amount of the @xmath3al @xmath1-ray line emission .
1.8 mev @xmath1-ray lines are emitted in the radioactive decay of @xmath3al , which is processed and released in supernovae but in other sources as well .
the 1.8 mev vela source appears to be extended with a significant peak at about lii = 267.4@xmath7 , bii = -0.7@xmath7 .
oberlack ( 1997 ) has used the rosat x - ray map of the vela region to model the 1.8 mev @xmath1-ray map , taking into account the full size of the vela snr , the vela snr explosion fragments ( aschenbach et al .
1995 ) , rx j0852.0 - 4622 and other potential sources .
he found two `` comptel point - like '' sources which could contribute significantly to the @xmath1-ray peak , which are the vela snr fragment d / d@xmath129 and rx j0852.0 - 4622 .
the peak position and the center position of rx j0852.0 - 4622 agree within the 2-@xmath72 localization accuracy of comptel , and the 1.8 mev point source flux is f@xmath144 = ( [email protected])@xmath2110@xmath11 photons @xmath22 s@xmath23 out of the total vela flux of f@xmath145 = ( [email protected])@xmath2110@xmath11 photons @xmath22 s@xmath23 .
recently , diehl et al . ( 1999 ) reported a 2-@xmath72 upper limit of 2@xmath2110@xmath11 photons @xmath22 s@xmath23 for a contribution of rx j0852.0 - 4622 to the overall vela emisson .
this result is not really in conflict with the result of oberlack , which we are going to use in the present paper . as we show below most of the our conclusions
do not depend on the precise value of f@xmath146 anyway . if f@xmath144 is to be attributed to a single snr with a representative yield of @xmath147 = 5@xmath510@xmath11 m@xmath12 it follows from eq .
( 1 ) that the distance of the source would be ( 160 @xmath4 20 ) pc using a mean lifetime of @xmath148 = 1.07@xmath510@xmath149 yrs . because this excitingly low distance for an snr was not supported by any other measurements at that time and
because of other competing @xmath3al sources like the vela snr fragment d / d@xmath129 the results were not published .
but after the discovery of the @xmath150 emission , which immediately implies a low age because of its short lifetime and a correspondingly low distance because of the x - ray angular diameter , the situation has changed and both the @xmath2ti and the @xmath3al flux may indeed come from a single supernova now visible as the rx j0852.0 - 4622 snr . because of the uncertainty of the amount of f@xmath146 actually to be attributed to rx j0852.0 - 4622 we discuss two cases in the following chapters : a. ) f@xmath144 is entirely from rx j0852.0 - 4622 ; b. ) f@xmath144 is not entirely associated with rx j0852.0 - 4622 but then the comptel data provide a firm upper limit of f@xmath151 = 3.5@xmath510@xmath11 photons @xmath22 s@xmath23 , which is the total flux observed for the entire vela region . eq .
1 can be used to compute the age @xmath152 of rx j0852.0 - 4622 by using the fluxes of just the two radionuclides , making use of @xmath153 : @xmath154 interestingly , the age determination does not require knowledge of the distance , and it depends only on the ratio of the mass yields of the two elements considered , which might be useful for further searches for young snrs . for f@xmath146 = f@xmath144 and @xmath155 = 1 , @xmath152 = ( 750 @xmath4 25 ) yrs .
this age agrees remarkably well with the age @xmath25 = @xmath97 = 680 yrs which has been derived from the @xmath2ti data and the x - ray measurements , and it appears to support the identification of rx j0852.0 - 4622 being the source of both the @xmath2ti and the @xmath3al emission .
furthermore the value of @xmath152 is not very sensitive to the precise value of @xmath156 ; even if only one fifth of @xmath156 , e.g. , is actually associated with rx j0852.0 - 4622 , @xmath152 is reduced by just 145 yrs . some interesting conclusions can be drawn about the type of the supernova by making use of model produced values of ( @xmath157 . the core - collapse models of woosley & weaver ( 1995 ) give 0.1 @xmath71 @xmath155 @xmath71 4.1 for progenitor masses between 11 m@xmath12 and 40 m@xmath12 for initial solar metallicity , excluding their models with @xmath158 m@xmath12 .
this leads to @xmath152 = ( 540 880 ) yrs @xmath430 yrs .
[ limit ] shows the ( @xmath159 ) plane of the core - collapse model data ( s - sequence of solar metallicity ) of woosley & weaver ( 1995 ) ; pairs of ( @xmath159 ) with @xmath147 greater than the values cut by the line of fixed @xmath28 are not consistent with f@xmath151 .
[ limit ] demonstrates that the models of woosley & weaver ( 1995 ) are consistent with relatively low expansion velocities , most of them with @xmath160 5000 km / s , which fits nicely the expansion velocity estimated from the x - ray temperature . for increasingly lower metallicity , @xmath147 of the woosley & weaver computations decreases and eventually the data of all the z = 0 models are above the @xmath28 = 5000 km / s cut , except the models for which @xmath16110@xmath162 m@xmath12 .
we note that the data of the models s18a , s19a and s25a describing the explosion of the progenitor with a mass of 18 m@xmath12 , 19 m@xmath12 and 25 m@xmath12 , respectively , are closest to the @xmath28 = 5000 km / s line .
this appears to be in rather good agreement with the conclusion which has been derived from the energy balance described in sect .
the core - collapse models of thielemann et al .
( 1996 ) for masses between 13 m@xmath12 and 25 m@xmath12 show similar values of @xmath29 but significantly lower values of @xmath147 because only the yields of the explosively produced elements are given ( thielemann , private communication , 1999 ) .
therefore @xmath147 is to be treated as lower limit , and the applicability to rx j0852.0 - 4622 remains unanswered at this stage .
interestingly , woosley & weaver ( 1995 ) have also calculated the yields of models with very little output of @xmath163 , from which the supernova power is being drawn after a possible plateau phase .
models with a small yield of @xmath163 , which may explain the sub - luminous supernovae after the early phase , also have low @xmath29 but relatively high values of @xmath147 , which is produced predominantly in the upper envelope by ordinary burning .
the yields predicted by woosley & weaver for solar metallicity are shown in fig .
[ limit ] as well .
clearly , the ratio of @xmath164 is very high and it is too high to be consistent with the observations . given the low value of @xmath29 , @xmath147 is simply too large .
such large values have to be checked against f@xmath151 .
f@xmath151 requires a minimal @xmath24 for a given @xmath147 , which in turn requires a maximal @xmath25 to be consistent with @xmath29 .
minimal @xmath24 and maximal @xmath25 define a minimal expansion velocity for the angular diameter not to exceed @xmath26 .
[ limit ] shows that the woosley & weaver models require very large values of @xmath28 such high mean expansion velocities after some 700 yrs are unlikely and it is evident that these models can not explain the rx j0852.0 - 4622 measurements primarily because they are inconsistent with the upper limit of the @xmath3al flux .
furthermore , the models of woosley & weaver show a signficant gap for @xmath29 , which covers the range 3@xmath510@xmath162 m@xmath165 10@xmath11 m@xmath12 . this gap might be artificial and further model calculations are required to check , whether the fall - back of matter towards the center of the explosion chokes the production of the high - z elements to the extent shown by the current explosion models .
models with somewhat lower values of both @xmath29 and @xmath147 would be consistent with the observations , and a lower value of @xmath29 might mean a low value of @xmath163 as well , which allows for a sub - luminous supernova although the connection between low y@xmath166 and low kinetic energy and luminosity is not yet well established . models for type ia supernovae predict a much higher ratio of ( @xmath155 ) ; iwamoto et al .
( 1999 ) predict 16@xmath167 470 and the sub - chandrasekhar models of woosley & weaver have 280@xmath167 930 .
these values correspond to a relatively large @xmath25 ( eq . 3 ) and
a low @xmath24 with @xmath168 = @xmath156 , which results in a relatively low value of the mean expansion velocity , i.e. 50 km / s @xmath169 275 km / s for the models of iwamoto et al . and 180 km / s @xmath170 1060 km / s for the models of woosley & weaver .
these values are well below the lower limit velocity of @xmath66 = 1240 km / s and are therefore inconsistent with the x - ray temperature measurements .
it appears that the type ia model predictions are in serious conflict with the measurements , thus excluding type ia models from explaining rx j0852.0 - 4622 .
but this conclusion hinges on the assumption that @xmath168 = @xmath156 is actually associated with rx j0852.0 - 4622 .
if only a minor fraction of @xmath10 1 % of @xmath156 is due to rx j0852.0 - 4622 , also type ia models may be reconsidered . for this case @xmath24 and @xmath25 are given in sect .
in contrast to type ia supernovae core - collapse supernovae are expected to leave a neutron star or a black hole initially close to the explosion center . here
we restrict the discussion to a neutron star .
if borne with a significant kick - velocity the neutron star will travel a distance from the centre given by the kick - velocity .
kick - velocities as large as 1000 km / s have been reported for pulsars , and for rx j0852.0 - 4622 with a mean expansion velocity of @xmath28 = 5000 km / s any putative neutron star should be within a radius of about 12@xmath129 around the center .
the rosat all - sky survey data of this area have been searched for point sources and two candidate sources have been found .
excess emission has been detected at ra(2000 ) = 8@xmath171 52@xmath129 3 `` , dec(2000 ) = 46@xmath7 18@xmath129 36 '' , which is off - set from the explosion center by 3.4@xmath129 . with the nominal value of @xmath28 and @xmath25 derived above
the separation corresponds to 283 km / s for the transverse component of the kick - velocity or a proper motion of 0.3 `` yr@xmath23 .
the center of the explosion has been determined by the circle matching best the snr outer boundary . the uncertainty in the center position is estimated to be about @xmath41.5@xmath129 .
this point - like source and the implications concerning a compact remnant have already been reported and discussed ( aschenbach , 1998 ) . here
we report excess emission from a second point - like source inside the suspected area at ra(2000 ) = 8@xmath171 51@xmath129 58 '' , dec(2000 ) = 46@xmath7 21@xmath129 33 `` .
this source has been detected in the low energy rosat image of the vela snr created from the counts which have been recorded in the central 40@xmath129 diameter field of the pspc .
compared to the full field of 2@xmath7 this procedure improves the spatial resolution considerably and thereby the sensitivity of detecting point sources above the diffuse background .
the 17 source counts per 40''@xmath540 `` pixel exceed the mean background level of 4.4 counts ( 40''@xmath540")@xmath23 by 6-@xmath72 .
the source count rate is 0.12 counts / s .
no spectrum and no information about interstellar absorption is available .
but the flux can be used to estimate the size of the x - ray emitting area as a function of temperature @xmath172 for a black - body with the interstellar absorption as parameter , which is shown in fig.[blackbody ] .
if the source is a black - body radiating neutron star of 10 km radius with emission from the entire surface , there is an upper limit of @xmath172 = 1.2@xmath510@xmath173k imposed by the 5-@xmath72 upper limit of n@xmath174 = 10@xmath175 @xmath22 .
but more realistical is a much lower value of n@xmath176 10@xmath43 @xmath22 , which is typical for the southeastern section of the vela snr .
then , fig.[blackbody ] implies that @xmath177 3@xmath510@xmath6k .
if just a fraction of the full surface is radiating the temperature may be slightly higher by about a factor of two , for instance , if the area of the radiating spot is about 1% of the full neutron star surface area . for a full size area radiating neutron star , which is just 700 yrs old
, @xmath172 would be surprisingly low .
furthermore , if the supernova left a neutron star somewhere else the black - body surface temperature would be even lower because of a lower x - ray count rate , unless the column density to the putative neutron star should be even higher .
a supernova going off at a distance of 200 pc should have been a spectacular sight for the contemporaries in the 13th or 14th century .
( if we take into account the full band of our age determination also the 12th and the 15th century should not be excluded . )
just how spectacular this event was depends on the absolute visual magnitude @xmath178 . for a bright supernova of type ia with @xmath179 19.0
the burst of light would have been as bright as the full moon . for a type ii , ib or ic which are intrinsically fainter the light output can be orders of magnitude less . for
the sub - luminous or ultra - dim sne recently discussed ( schaefer 1996 , hatano et al .
1997 , woltjer 1997 ) @xmath178 = 13.0
@xmath4 2 .
examples quoted are sn 1181 with @xmath178 = 12.68 @xmath4 1.41 and cas a with @xmath180 13.03 @xmath4 2.69 . taking a somwewhat extreme position with @xmath178 = 11.5 the sn associated with rx j0852.0 - 4622 would still have been brighter than venus .
taking the other extreme position of a type ia the sn is a candidate to be recognized even in daylight . in any case
it should have been seen .
records of astronomical events including the epoch of the supernova proposed by us were taken by the far - east astronomers of china , japan and corea ( clark & stevenson 1977 , ho peng yoke 1962 ) .
their observatories were typically located at a geographical latitude of @xmath035@xmath7 north ( clark & stevenson 1977 ) , within @xmath45@xmath7 north or south , so that the sn of rx j0852.0 - 4622 would have risen above the horizon after sun - set by up to 11@xmath7 from middle of december to end of march .
if the sn of rx j0852.0 - 4622 would have exploded say in late march it would have re - appeared in the northern hemisphere during night after more than 250 days with significantly lower brightness .
the light curve of the sub - luminous sn 1997d ( turatto et al .
1998 ) provides an estimate of @xmath181 5 at @xmath0 250 days after outburst , which appears rather little compared with other sne which show @xmath181 8 like sn 1994w ( sollerman et al . 1998 ) . with @xmath178 = 11.5 ,
@xmath182 = 5 and a distance modulus of 6.5 the sn of rx j0852.0 - 4622 then had @xmath183 = 0 and it is not unlikely to have escaped the attention of the medieval far - east observers . even with the @xmath178 of sn 1181 the chances to miss it would not have been low .
furthermore , within this scenario , that only the tail of the light curve had been caught , the sn may have not been noticed as a `` guest star '' because the change was only against a pattern observed more than 200 days before .
the light curve of sn 1997d also demonstrates that even if the sn went off in the december
march time frame the detection might have been prevented by a rather short peak / plateau period , and sky visibility conditions become important . for sn 1997d
this period probably lasted for @xmath71 60 days ( turatto et al . ) , over which the brightness decreased by @xmath181 3 .
the above exercise demonstrates with realistic data that it is indeed possible that the sn of rx j0852.0 - 4622 was not bright enough , despite its proximity .
this sort of physical explanation requires a sub - luminous sn .
the peak luminosity and the early lightcurve are determined by the ejecta mass , e@xmath113 , pre - sn radius , the structure of the outer layers , @xmath83 and its distribution . as shown by chugai & utrobin ( 1999 ) in their model for sn 1997d
it appears that @xmath83 is rather low for this class of sub - luminous sne , and it remains to be seen whether these sne can actually produce enough @xmath2ti ( but c.f .
[ age ] , [ distance ] for the minimum amount ) .
we stress that we can not exclude that the sn was indeed much brighter and even a daylight object .
observers located much further south , like the people of the incas , the aztecs or in middle- and south - africa , should have had better visibility and their traditions are recommended to be searched for an event pointing to a sn . in this context
it is interesting to note that the records of the far - east observers as published by ho peng yoke ( 1962 ) appear to be incomplete .
the compilation shows three gaps , which are suspiciously long and statistically inconsistent with the mean rate of entries .
these periods include the years of 773 814 , 1245 1264 and 1277 1293 , the latter two of which are relevant for rx j0852.0 - 4622 . finally , we point out that there is a chance that the progenitor star of rx j0852.0 - 4622 is shown in ancient star charts if it happened to be a massive star . up to now
just one progenitor star of a supernova has been identified , which is the progenitor of sn 1987a , the b3 ia blue supergiant sanduleak 69 202 with m@xmath184 = 6.8 ( west et al .
1987 ) . at a distance of 200 pc
the apparent unreddened visual magnitude would have been @xmath183
= 0.3 , which would have made the star the brightest star in the vela constellation located between @xmath1 vel and @xmath185 vel .
a wolf - rayet type progenitor star would have been less bright with @xmath186 + 2.5 , but still comparable with the other bright stars in vela .
this opens up an interesting explanation for the apparent absence of a historical record , which admittedly is a speculation .
if the progenitor star had been so bright the supernova might not have been noted down as a `` guest star '' .
the existing star would just have become brighter . and if the short peak of the outburst had been missed of whatsoever reason and only the tail of the supernova lightcurve has been observed the change of brightness might not have been spectacular , and the star would have disappeared slowly over a couple of years .
an estimate of the age @xmath25 and distance @xmath24 of the supernova remnant rx j0852.0 - 4622 / gro j0852 - 4642 can be obtained by combining the rosat x - ray and comptel @xmath1-ray data . assuming a @xmath187ti yield of the supenova of 5 @xmath5 10@xmath188 m@xmath189 and an expansion velocity of 5000 km / sec @xmath25 = 680 yrs and @xmath24 = 200 pc are obtained .
actually , the expansion velocity is constrained by the x - ray data to lie in the range of 2000 km / sec @xmath10 v @xmath10 10 000 km / sec yielding an uncertainty of the age of @xmath4 100 years for a fixed @xmath187ti yield . for the highest @xmath187ti yield given by current supernova models ,
a firm upper limit of the distance is 500 pc and 1100 years for the age .
the determination of the age depends to some extent on the ionization state of @xmath187ti because @xmath187ti decays by electron capture .
the values quoted above have been obtained under the assumption that the k - shell is fully populated . if the k - shell contains only one electron , the @xmath187ti mean lifetime
is estimated to increase by a factor of two .
but the age of the snr will not increase by the same factor because of the angular diameter , and therefore distance constraint .
adopting the same mean expansion velocity @xmath25 and @xmath24 can change by about 35% at most and for a very strong ionization @xmath25 and @xmath24 may be even lower than the `` nominal '' estimate .
future x - ray spectroscopy measurements are needed to search for ti x - ray emission lines to determine the ionization state of @xmath187ti and further constrain @xmath25 and @xmath24 . the x - ray surface brightness of rx j0852.0 - 4622
is rather low and implies a rather low matter density of the shock wave heated plasma if the radiation is thermal . a formal analysis of the x - ray data in terms of a sedov - type evolution of the snr using the standard conversion of x - ray temperature in shock velocity turns out a rather low value of a few times 10@xmath122 erg for the explosion energy @xmath119 , which can be raised only significantly if the mean expansion velocity @xmath28 would exceed 10000 km / s .
but if @xmath28 is closer to 5000 km / s as the x - ray data indicate then the bulk of @xmath119 resides still in kinetic energy of the ejecta , not radiating in x - rays , which means that any reverse shock has not yet penetrated deep into the ejecta , and that the titanium is not highly ionized . in this case a lower limit for the mass of the progenitor star of 25 m@xmath12 is estimated from the energy balance .
there is evidence for @xmath190al 1.809 mev line emission from rx j0852.0 - 4622 , which has been measured by comptel towards the vela region .
admittedly this has still to be confirmed .
but if a non - neglible part of this 1.809 mev line flux is coming from rx j0852.0 - 4622 a similar age of 600 - 750 yrs for the snr is obtained for similar yields of @xmath190al and @xmath187ti .
since @xmath25 depends only on the logarithm of the yields and the fluxes @xmath25 will not change significantly even for large changes of the @xmath190al flux .
it is more a matter of whether or not there is @xmath190al emission .
if the @xmath190al line flux is about what is indicated by the comptel data the existing type ia supernova models can be ruled out for the progenitor explosion because they predict a ratio of the @xmath190al and @xmath187ti yield which is by far too large .
they could be reconsidered only if less than 1% of the @xmath190al line flux from vela in total is associated with rx j0852.0 - 4622 .
explosion models of core - collapse supernovae ( woosley & weaver , 1995 ) are in general in agreement with the observations , i.e. the measurements of @xmath28 , @xmath187ti line flux and @xmath190al line flux , judging from their prediction of the yields of @xmath187ti and @xmath190al . their models with @xmath191 10@xmath162 m@xmath12 and @xmath192
10@xmath81 m@xmath12 can definitely be excluded because the yield of @xmath190al predicted is inconsistent with even the upper limit of the @xmath190al line flux measured for the vela region in total . in summary , both the energy balance and the yield predictions of the currently available explosion models point towards a core - collapse event . within the vicinity of the explosion center
two point - like x - ray emission regions have been found , either of which could be the manifestation of a neutron star .
spectra are not available , but if the radiation is assumed to be black - body emisson , the x - ray flux indicates a surface temperature of @xmath873@xmath510@xmath6 k , which is surprisingly low for a 700 years old neutron star .
if neither one of these two sources is a neutron star and the neutron star hides somewhere else with an even lower x - ray count rate or if only a fraction of the radiation observed from the point - like objects is due to thermal radiation the surface temperature of the neutron star could be even lower . in principle
, the supernova could have been seen from the far - east astronomers of china , corea or japan or from geographical latitudes further south .
the supernova could have been very bright and then there are records expected to exist , which should be searched for .
if the supernova would have been of the sub - luminous class with a brightness as low as that of sn 1181 and a short peak - plateau duration it could have been missed .
there is some chance that the progenitor star was sufficiently bright and could have been seen by the naked eye .
then the stellar pattern of vela was different in ancient times and the astronomers who monitored the sky might not have noted the supernova as a `` guest star '' or a `` new '' star , because the star was existing and just brightening and eventually fading away
. ahmad i. , bonino g. , cini castagnoli g. , et al . , 1998 , phys .
80 , 2550 arnaud m. , rothenflug r. , 1985 , a&as 60 , 425 aschenbach b. , 1998 , nature 396 , 141 aschenbach b. , egger r. , trmper j. , 1995 , nature 373 , 587 chen w. , gehrels n. , 1999 , apj 514 , l103 chugai n.n .
, utrobin v.p . , 1999 ,
astro - ph/9906190 10 jun 1999 clark d.h . ,
stephenson f.r . , 1977 , `` the historical supernovae '' , pergamon , oxford diehl r. , bennet k. , bloemen h. , et al . , 1995 ,
a&a 298 , l25 diehl r. , oberlack u. , plschke s. , et al . , 1999 , proc .
3rd integral workshop , taormina 1998 , eds .
g. palumbo , h. bazzano , c. winkler , astrophys .
lett . & comm . , 1999 , in press hatano k. , fisher a. , branch d. , 1997 , mnras 290 , 360 ho peng yoke , 1962 , vistas in astron . 5 , 127 hughes j. , 1996 , baas 189 , 4608 iwamoto k. , brachwitz f. , nomoto k. , et al . , 1999 ,
iyudin a.f . , schnfelder v. , bennett k. , et al . , 1998 ,
nature 396 , 142 koralesky b. , rudnick l. , gotthelf e.v .
, keohane j.w . , 1998 ,
apj 505 , l27 mochizuki y. , takahashi k. , janka h .- th .
, hillebrandt w. , diehl r. , 1999 , a&a 346 , 831 nagataki s. , 1999 , apj 511 , 341 nagataki s. , hashimoto m .- a . , sato k. , yamada s. , mochizuki y. , 1998 , apj 492 , l45 nomoto k. , thielemann f .- k . , yokoi k. , 1984 , apj 286 , 644 oberlack u. , 1997 , phd thesis , ber die natur der galaktischen @xmath3al - quellen untersuchung des 1.8 mev himmels mit comptel , technische universitt mnchen oberlack u. , diehl r. , montmerle t. , et al .
, 1994 , apjs 92 , 433 pfeffermann e. , aschenbach b. , predehl p. , 1991
, a&a 246 , l28 schaefer b.e , 1996 , apj 464 , 404 sollerman j. , cumming r.j . , lundqvist p. , 1998
, apj 493 , 933 strom r.g . , 1994 , a&a 288 , l1 thielemann f
.- k . , nomoto k. , hashimoto m .- a . , 1996 ,
apj 460 , 408 turatto m. , mazzali p.a.,young t.r . , et al . , 1998 ,
apj 498 , l129 vink j. , bloemen h. , kaastra j.s .
, bleeker j.a.m . , 1998 , a&a 339 , 201 west r.m . , lauberts a. , jrgensen h.e . , schuster h .- e . , 1987 , a&a 177 , l1 willingale r. , west r.g . , pye j.p . , stewart g.c . , 1996 , mnras 278 , 749 winkler jr . ,
clark g.w .
, 1974 , apj 191 , l67 woltjer l. , 1997 , a&a 328 , l29 woosley s.e . ,
weaver t.a . , 1994 ,
apj 423 , 371 woosley s.e . ,
weaver t.a . , 1995 ,
apjs 101 , 181 | the discovery of the nearest young supernova remnant rx j0852.0 - 4622 / gro j0852 - 4642 in the galaxy by rosat and comptel has been reported recently . age and distance
are determined to @xmath0680 years and @xmath0200 pc by the x - ray diameter and the @xmath1-ray line flux of radioactive @xmath2ti . here
we discuss the implications of the x - ray spectra and of the fact that 1.8 mev @xmath1-ray line emission from the decay of @xmath3al has been measured from the vela region with a certain fraction possibly associated with the new snr .
we estimate an uncertainty of the age of @xmath4 100 yrs for a fixed yield of @xmath2ti .
the highest values of @xmath2ti yield provided by current supernova explosion models give worst case upper limits of 1100 yrs for the age and of 500 pc for the distance .
also the unknown ionization stage of @xmath2ti adds to the uncertainty of age and distance which is at most another 35% on top .
both the energy balance compiled for the remnant and yield predictions for @xmath2ti and @xmath3al by supernova models favour a core - collapse event .
two point sources have been found in the vicinity of the explosion center , either one of these might be the neutron star left by the supernova .
if there is a neutron star the x - ray count rates of the two point sources provide an upper limit of the blackbody surface temperature , which is very unlikely to exceed 3@xmath510@xmath6 k. the supernova might have been observed some 700 @xmath4 150 yrs ago , but based on the data of sn 1181 , e.g. , there is a realistic chance that it has been missed if the supernova was sub - luminous . |
the max - plus or tropical semiring @xmath0 is the set @xmath1 , equipped with the addition @xmath2 and the multiplication @xmath3 .
we refer the reader for instance to @xcite for introductory materials on max - plus or tropical algebra .
we denote by @xmath4 the @xmath5th - fold cartesian product of @xmath0 , which can be thought of as the tropical analogue of a finite dimensional vector space .
tropical hyperplane _ is a subset of @xmath4 of the form @xmath6 } ( { a}_i+x_i ) \text { is attained at least twice}\ } \enspace , \label{e - def - h}\end{aligned}\ ] ] where @xmath7 is a vector of @xmath4 , not identically @xmath8 , and @xmath9:=\{1,\dots , n\}$ ] .
this definition is motivated by non - archimedean geometry .
indeed , let @xmath10 denote the field of complex puiseux series in the variable @xmath11 , and let @xmath12 denote the valuation which associates to a series the opposite of its smallest exponent .
consider now a hyperplane of @xmath13 , @xmath14 } \mathbf{{a}}_i\mathbf{x}_i = 0 \ } \enspace , \ ] ] where @xmath15 is a vector of @xmath13 lifting @xmath16 , meaning that @xmath17 , for all @xmath18 $ ] .
then it is easily checked that the image of @xmath19 by the map which applies the valuation @xmath12 entrywise is precisely the set of vectors of @xmath20 with rational coordinates .
this is actually a special case of a result of kapranov characterizing the non - archimedean amoeba of a hypersurface @xcite . in the present work ,
we will extend or refine a series of basic results concerning the intersections of tropical hyperplanes , that we now review .
one of these results was established by richter - gebert , sturmfels and theobald .
[ th-1 ] any @xmath21 vectors of @xmath22 in general position are contained in a unique tropical hyperplane .
this statement has also an equivalent dual form : _ the intersection of @xmath21 tropical hyperplanes in general position contains a unique vector up to an additive constant_. the parameters @xmath23 of the hyperplane @xmath20 arising in theorem [ th-1 ] can be obtained by solving the tropical analogue of a square linear system .
its solution turns out to be determined by the tropical analogues of cramer determinants , which in this context are merely the value of optimal assignment problems .
then the data are said to be in _ general position _ if each of the optimal assignment problems arising in this way has a unique optimal solution .
the role of the general position notion is made clear by the following result .
[ th-2 ] a collection of @xmath5 vectors of @xmath22 is contained in a tropical hyperplane if and only if the matrix having these vectors as columns is _ tropically singular _ , meaning that the assignment problem associated with this matrix has at least two optimal solutions
. more information on tropical singularity and related rank notions can be found in @xcite .
different but related results were obtained previously by considering tropical numbers with signs .
indeed , in the above results , the tropical semiring is essentially thought of as the image of the field of complex puiseux series by the valuation .
alternatively , a tropical number with sign may be thought of as the image of a _ real _ puiseux series , i.e. , of an element of @xmath24 .
ideas of this nature were indeed essential in the development by viro of the patchworking method ( see the references in @xcite ) , as well as in the construction by plus @xcite of the symmetrized tropical ( or max - plus ) semiring , @xmath25 .
when considering tropical numbers with signs , the notion of equation has to be replaced by a notion of _ balance _
@xcite : a tropical sum of terms equipped with signs is said to be balanced if the maximum of terms with positive signs coincides with the maximum of the terms with negative signs .
some results of @xcite concerning systems of linear balances can be interpreted geometrically using the signed variant of tropical hyperplanes , considered by joswig in @xcite .
a _ signed tropical hyperplane _ is a subset of @xmath4 of the form @xmath26 where @xmath9= i\cup j$ ] is a non - trivial partition , and @xmath27 is a vector non - identically @xmath8 .
observe that @xmath28 .
consider now a hyperplane of @xmath29 , @xmath30 where @xmath31 is any vector of @xmath29 lifting @xmath16 , meaning now that @xmath17 and that @xmath32 is nonnegative ( recall that a real puiseux series is nonnegative if it is zero or if its leading coefficient is positive ) , for all @xmath18 $ ] .
then it can be easily checked that the vectors with rational entries of a signed tropical hyperplane are precisely the images by the valuation of the nonnegative vectors of the associated hyperplane over the field of _ real _ puiseux series . with these observations in mind ,
the following result established by plus appears to be a `` real '' analogue of theorem [ th-1 ] .
[ th-3 ] any @xmath21 vectors of @xmath4 in sign - general position are contained in a unique signed tropical hyperplane . by comparison with theorem [ th-1 ] ,
we use here a milder notion of general position .
indeed , the `` real '' tropical analogue of a determinant consists of the value of an optimal assignment problem , together with the information of all the possible signs of optimal permutations .
then , a tropical determinant is said to be _ sign - nonsingular _ if all optimal permutations have the same sign .
finally , @xmath21 vectors of @xmath4 are said to be in _ sign - general position _ if all the associated tropical cramer determinants are sign - nonsingular .
the vector @xmath16 defining the signed hyperplane @xmath33 of theorem [ th-3 ] is determined by the tropical cramer determinants , the signs of which provide the sets @xmath34 in the partition .
actually , a more general result was stated in @xcite in the language of systems of balances over the symmetrized tropical semiring @xmath25 .
theorem [ th-3 ] covers a special case with a more straightforward geometric interpretation .
the details of the derivation of the latter theorem from theorem 6.1 of @xcite will be given in section [ sec - geom ] , together with the dual result of theorem [ th-3 ] , concerning intersections of signed tropical hyperplanes .
a result of gondran and minoux , that we restate as follows in terms of signed tropical hyperplanes , may be thought of as a `` real '' analogue of theorem [ th-2 ] .
[ th-4 ] a collection of @xmath5 vectors of @xmath4 is contained in a signed tropical hyperplane if and only if the matrix having these vectors as columns has a sign - singular tropical determinant .
a dual result , concerning the intersection of @xmath5 signed tropical hyperplanes , was stated by plus @xcite and proved by gaubert @xcite , we shall discuss it in section [ sec - geom ] . as pointed out in @xcite , the notion of sign - nonsingularity arising here
is an extension of the notion with the same name arising in combinatorial matrix theory , in particular in the study of the permanent problem of plya , see @xcite for more information .
whereas the results concerning `` balances '' involve the extension of the tropical numbers with signs , we note that other extensions have been used more recently .
in particular , the incorporation of `` phase '' ( instead of sign ) information in tropical constructions has played an important role in the arguments of mikhalkin @xcite .
moreover , viro introduced a general notion of _ hyperfield _
@xcite with a multivalued addition , which he used in particular to capture the phase information .
the semirings with symmetry introduced by the authors in @xcite provide another way to encode the sign or phase .
also , different extensions have been provided by the `` supertropical '' structures of izhakian and rowen @xcite extending the bi - valued tropical semiring introduced by izhakian @xcite ; in the latter , the goal is not to encode `` sign '' or `` phase '' , but the fact that the maximum in an expression is achieved twice at least .
given the analogy between the `` complex '' and `` real '' versions of the tropical cramer theorem ( theorems [ th-1 ] and [ th-3 ] above ) , as well as between the unsigned and signed notions of tropical singularity of matrices ( theorems [ th-2 ] and [ th-4 ] ) , one may ask whether all these results may be derived from common principles .
one may also ask whether results of this kind are valid for more general tropical structures , encoding different kinds of sign or phase information , in the light of recent constructions of extended tropical semirings by viro @xcite , izhakian and rowen @xcite , and the authors @xcite . in this paper , we answer these questions and deal with related algorithmic issues , by developing a theory of elimination of linear systems over semirings , building on ideas and results of @xcite .
this will allow us to show that the earlier results are indeed special instances of general cramer type theorems , which apply to various extensions of tropical semirings .
these theorems are established using axioms allowing one to perform `` elimination '' of balances , in a way similar to gaussian elimination . in this way
, we will generalize and sometimes refine ( handling degenerate cases ) earlier results .
also , some of our proofs are based on jacobi or gauss - seidel type iterative schemes , and lead to efficient algorithms . in passing ,
we revisit some of the results of @xcite , giving their geometric interpretation in terms of signed tropical hyperplanes .
an ingredient of our approach is the introduction ( in section [ sec - prelim ] ) of a rather general notion of _ extension _ of the tropical semiring , together with a general `` balance '' relation @xmath35 , which , depending on the details of the extension , expresses the fact that the maximum is attained at least twice in an expression , or that the maxima of two collections of terms coincide .
our constructions include as a special case the symmetrized tropical semiring of plus @xcite and the bi - valued tropical semiring of izhakian @xcite , but also a certain `` phase extension '' of the tropical semiring , which is a variant of the complex tropical hyperfield of viro .
our notion also includes certain `` supertropical semifields '' in the sense of izhakian and rowen @xcite .
then auxiliary combinatorial results are presented in section [ sec - comb ] .
the general affine cramer system reads @xmath36 , whereas the homogeneous system reads @xmath37 , where @xmath38 is an @xmath39 matrix and @xmath40 is a vector of dimension @xmath5 , both with entries in an extension of the tropical semiring . to be interpreted geometrically , the vector @xmath41 which is searched
will be required to satisfy certain non - degeneracy conditions , typically that the coordinate of @xmath41 do not belong to the set of balanced non - zero elements of the extension . in this way , by developing methods of @xcite , we obtain in section [ sec - elim ] a general result , theorem [ th - cramer ] , concerning the unique solvability of non - singular cramer system , which includes theorems [ th-1 ] and [ th-3 ] as special cases
. then we study in section [ sec - exists ] the existence problem for the solution of the affine cramer system @xmath42 , without making the non - singularity assumptions needed in the previous uniqueness results .
theorem [ theo - jacobi ] below gives a general existence theorem , with a constructive proof based on the idea of the jacobi algorithm in @xcite .
theorem [ theo - gauss - seidel ] gives an alternative gauss - seidel type algorithm .
these results are valid in a large enough class of semirings , including not only the symmetrized tropical semiring as in @xcite , but also the phase extension of the tropical semiring . in section
[ sec - homogeneous ] , we deal with the generalization of theorems [ th-2 ] and [ th-4 ] which concern singular linear systems of @xmath5 equations in @xmath5 variables . in theorem
[ newtgma ] , we characterize the existence of non - degenerate solutions of @xmath43 , recovering theorems [ th-2 ] and [ th-4 ] as special cases .
this extends to more general semirings a theorem of gaubert @xcite dealing with the case of the symmetrized tropical semiring @xmath25 .
we note however that by comparison with the jacobi / gauss - seidel type results of section [ sec - exists ] the results of this section hold under more restrictive assumptions on the semiring .
a geometrical interpretation of the previous results , in the case of the symmetrized tropical semiring , is presented in section [ sec - geom ] . in section [ s_ac ] , we start to address computational issues .
the @xmath44 cramer determinants of the system @xmath36 correspond to @xmath44 optimal assignment problems . in section [ sec - perall ] , we show that our approach based on a jacobi - type iterative method leads to an algorithm to compute the solution , as well as the cramer determinants ( up to signs ) , by solving a single ( rather than @xmath44 ) optimal assignment problem , followed by a single destination shortest path problem . for the sake of comparison , we revisit in section [ sec - transport ] the approach of richter - gebert , sturmfels and theobald @xcite , building on results of sturmfels and zelevinsky dealing with minkowski sums of birkhoff polytopes
: it gives a reduction to a different transportation problem .
although the original exposition of @xcite is limited to instances in general position , we show that some of their results remain valid even without such an assumption .
the computation of the signs of tropical cramer determinants is finally briefly discussed in section [ sec - compute - det ] .
a _ semiring _ is a set @xmath45 with two binary operations , addition , denoted by @xmath46 , and multiplication , denoted by @xmath47 or by concatenation , such that : * @xmath45 is an abelian monoid under addition ( with neutral element denoted by @xmath48 and called zero ) ; * @xmath45 is a monoid under multiplication ( with neutral element denoted by @xmath49 and called unit ) ;
* multiplication is distributive over addition on both sides ; * @xmath50 for all @xmath51 . in the sequel
, a semiring will mean a non - trivial semiring ( different from @xmath52 ) .
briefly , a semiring differs from a ring by the fact that an element may not have an additive inverse .
the first examples of semirings which are not rings that come to mind are non - negative integers @xmath53 , non - negative rationals @xmath54 and non - negative reals @xmath55 with the usual addition and multiplication .
there are classical examples of non - numerical semirings as well .
probably the first such example appeared in the work of dedekind @xcite in connection with the algebra of ideals of a commutative ring ( one can add and multiply ideals but it is not possible to subtract them ) .
a semiring or an abelian monoid @xmath45 is called _ idempotent _ if @xmath56 for all @xmath57 , @xmath45 is called _ zero - sum free _ or _ antinegative _ if @xmath58 implies @xmath59 for all @xmath60 , and @xmath45 is called _ commutative _ if @xmath61 for all @xmath60 .
an idempotent semiring is necessarily zero - sum free .
we shall always assume that the semiring @xmath45 is commutative .
an interesting example of an idempotent semiring is the max - plus semiring @xmath62 where @xmath63 and @xmath3 . here
the zero element of the semiring is @xmath8 , denoted by @xmath64 , and the unit of the semiring is 0 , denoted by @xmath65 .
the usual definition of matrix operations carries over to an arbitrary semiring .
we denote the set of @xmath66 matrices over @xmath45 by @xmath67 . also we denote @xmath68 and we identify @xmath69 with @xmath70 .
note that @xmath71 is a semiring .
some of the following notions were introduced in @xcite , where details and additional properties can be found .
let @xmath45 be a semiring .
a map @xmath72 is a symmetry if [ sym ] @xmath73 a trivial example of a symmetry is the identity map @xmath74 . of course
, in a ring , we may take @xmath75 .
[ car - sym ] a map @xmath76 is a symmetry of the semiring @xmath45 if and only if there exists @xmath77 such that @xmath78 , and @xmath79 for all @xmath57 ( hence @xmath80 commutes with all elements of @xmath45 ) .
let @xmath76 be a symmetry of @xmath45 , and denote @xmath81 . by
, we get that @xmath82 , and @xmath83 for all @xmath57 . by , we deduce that @xmath84 , and since @xmath85 , we get @xmath78 . conversely , if @xmath86 for all @xmath57 , with @xmath78 , then @xmath76 satisfies all the conditions in . in the rest of the paper we shall write @xmath87 for @xmath88 .
so , @xmath89 is not zero generally speaking , but is a formal sentence meaning @xmath90 .
moreover , for any integer @xmath91 , @xmath92 will mean the @xmath5th power of @xmath93 , hence the product of @xmath5 copies of @xmath94 .
also @xmath95 will mean @xmath96 , in particular in the formula @xmath97 .
if the addition of @xmath45 is denoted by @xmath98 instead of @xmath46 , then @xmath95 and @xmath87 will be replaced by @xmath99 and @xmath100 .
[ circ ] for any @xmath101 , we set @xmath102 , thus @xmath103 , and we denote @xmath104 the elements of this set will be called _ balanced elements _ of @xmath45 .
moreover , we define the _ balance _ relation @xmath105 on @xmath45 by @xmath106 if @xmath107 .
note that @xmath108 is an ideal , hence the relation @xmath105 is reflexive and symmetric .
it may not be transitive .
since @xmath108 is an ideal , it contains an invertible element of @xmath45 if and only if it coincides with @xmath45 .
we shall only consider in the sequel symmetries such that @xmath109 .
this permits the following definition .
when @xmath109 , we say that a subset @xmath110 of @xmath45 is _ thin _ if @xmath111 and if it contains @xmath48 and all invertible elements of @xmath45 .
when such a set @xmath110 is fixed , its elements will be called _ thin elements_. note that @xmath112 is both a balanced and a thin element .
in the sequel , we shall consider systems of linear `` equations '' ( the equality relation will be replaced by balance ) , and we will require the variables to be thin . by choosing appropriately the set of thin elements , we shall see that standard gauss type elimination algorithms carry over .
also , in most applications , only the thin solutions will have simple geometrical interpretations .
note that in @xcite , we only used the notation @xmath110 for the maximal possible thin set , that is @xmath113 ; however , it will be useful to consider also for instance the smallest possible thin set , which is the set of invertible elements completed with @xmath112 .
recall that @xmath114 is an _ ordered semigroup _ if @xmath115 is a semigroup , and @xmath116 is an order on @xmath117 such that for all @xmath118 , @xmath119 implies @xmath120 and @xmath121 .
an _ ordered monoid _ is a monoid that is an ordered semigroup .
an _ ordered semiring _ is a semiring @xmath122 endowed with an order relation @xmath116 such that @xmath123 and @xmath124 are ordered semigroups . for any subsemiring @xmath125 of @xmath45
, one defines the relations : @xmath126 we shall write simply @xmath127 instead of @xmath128 , and @xmath129 instead of @xmath130 .
these relations are preorders ( reflexive and transitive ) , compatible with the laws of @xmath45
. they may not be antisymmetric .
the preorder @xmath127 is called the _ natural preorder _ on @xmath45 .
a semiring @xmath45 is said to be _ naturally ordered _ when @xmath127 ( or equivalently @xmath131 ) is an order relation , and in that case @xmath127 is called the _ natural order _ on @xmath45 , and @xmath131 is its opposite order .
the notation @xmath132 and @xmath133 will be used for the corresponding strict relations . when @xmath127 ( or @xmath131 ) is an order relation ,
so are @xmath134 , @xmath129 , @xmath135 , and @xmath136 .
an idempotent semiring is necessarily naturally ordered , and a naturally ordered semiring is necessarily zero - sum free .
we also have : @xmath137 the converse is false in general .
let @xmath45 be a semiring and @xmath138 be an idempotent semiring in which the natural order is total , or for short , a totally ordered idempotent semiring .
we say that a map @xmath139 is a _ modulus _ if it is a surjective morphism of semirings . in this case , we denote @xmath140 by @xmath141 for all @xmath101 .
we shall apply the notations @xmath105 , @xmath134 , @xmath135 ( so @xmath127 , @xmath131 , @xmath129 , and @xmath136 ) , @xmath142 to matrices and vectors , understanding that the relation holds entrywise .
we shall do the same for the notions of `` balanced '' or `` thin '' elements .
[ sym - tot ] if @xmath143 is a totally ordered idempotent semiring , then the only symmetry on @xmath138 is the identity .
moreover , any semiring @xmath144 with a symmetry and a modulus @xmath145 satisfies @xmath146 .
let @xmath76 be a symmetry of @xmath138 . then by proposition [ car - sym ]
, there exists @xmath147 such that @xmath148 , and @xmath79 for all @xmath149 . since @xmath138 is totally ordered , either @xmath150 or @xmath151 .
assuming that the inequality @xmath150 holds , then multiplying it by @xmath80 , we get @xmath152 , and since @xmath148 and @xmath116 is an order relation , we deduce that @xmath153 .
the same is true when the inequality @xmath154 holds .
this shows that @xmath153 , hence @xmath76 is the identity map . if now @xmath45 is a semiring with a symmetry and a modulus , then @xmath155 , and by the above arguments , we get that @xmath156 . in the sequel , we shall consider semirings with a symmetry and a modulus .
the following construction allows one to obtain easily such semirings .
[ extension ] let @xmath144 be a semiring and let @xmath157 be a totally ordered idempotent semiring
. then the semiring @xmath158 is defined as the set @xmath159 endowed with the operations @xmath160 its zero element is @xmath161 and its unit is @xmath162 . for any subset @xmath163 of a semiring @xmath45 , we denote @xmath164 in particular , @xmath165
. then the extension of @xmath138 by a subset @xmath163 of @xmath45 is defined by @xmath166 if @xmath163 is a subsemiring of @xmath45 , then @xmath167 , and @xmath168 are subsemirings of @xmath158 . if in addition @xmath163 is zero - sum free and without zero divisors , then @xmath169 is a subsemiring of @xmath168
. this construction bears some similarity with a semidirect product , which motivates the notations `` @xmath170 '' , @xmath171 and `` @xmath172 '' .
we shall denote by @xmath173 and @xmath174 , instead of @xmath161 and @xmath162 , the zero and unit of @xmath158 .
if @xmath175 is a totally ordered monoid , completing @xmath117 with a bottom element , denoted by @xmath176 , we get the totally ordered idempotent semiring @xmath177 .
all the semirings @xmath138 satisfying the assumptions of proposition [ extension ] are of this form .
when @xmath178 is equipped with its usual order and addition , we recover the max - plus semiring @xmath0 .
we may take more generally for @xmath117 any submonoid of @xmath179 , or take @xmath180 equipped with the lexicographic order and entrywise addition .
the intuition of the construction of proposition [ extension ] is best explained by the following example .
[ ex - best ] let @xmath181 denote the field of complex numbers .
then the semiring @xmath182 will be called `` complex extension of the tropical semiring '' . an element @xmath183
encodes the asymptotic expansion @xmath184 , when @xmath185 goes to @xmath186 ( when @xmath187 , this is the identically @xmath112 expansion ) .
indeed , the `` lexicographic '' rule in the addition of @xmath182 corresponds precisely to the addition of asymptotic expansions , and the entrywise product of @xmath182 corresponds to the product of asymptotic expansions . by taking the zero - sum free subsemiring @xmath188 consisting of the real nonnegative numbers , we end up with the subsemiring @xmath189 , which encodes the asymptotic expansions @xmath190 with @xmath191 and @xmath192 , together with the identically @xmath112 expansion .
the latter semiring was used in @xcite under the name of semiring of ( first order ) jets , to study eigenvalue perturbation problems .
we next list some simple facts concerning extensions of semirings .
[ remark - extension0 ] let @xmath193 be the boolean semiring , i.e. , the idempotent semiring with two elements , let @xmath45 be zero - sum free and without zero divisors .
then @xmath194 is isomorphic to @xmath45 . in general @xmath45
is a subsemiring of @xmath195 since the injective map @xmath196 defined by @xmath197 is a morphism of semirings .
so the semirings @xmath158 and @xmath195 are semiring extensions of @xmath45 .
the same is true for @xmath198 as soon as @xmath45 is zero - sum free without zero divisors .
note that one can also consider the map @xmath199 , such that @xmath200 , for all @xmath201 .
this map yields a surjective multiplicative morphism from @xmath158 ( or from @xmath195 or @xmath198 ) to @xmath45 , such that the composition @xmath202 equals the identity of @xmath45 .
if @xmath45 and @xmath138 are commutative then so is @xmath158 .
if @xmath45 is idempotent ( resp . , naturally ordered ) then so is @xmath158 , and consequently @xmath195 and @xmath198 .
the natural preorder on @xmath158 is the lexicographic preorder : @xmath203 if and only if either @xmath204 , or @xmath205 and @xmath206 .
[ remark - extension ] let @xmath76 be a symmetry of @xmath45 .
we define the map @xmath207 from @xmath158 to itself by @xmath208 for all @xmath57 and @xmath209 .
then @xmath207 is a symmetry of @xmath158 , which preserves @xmath195 and @xmath198 .
we shall call @xmath207 the extension of @xmath76 .
conversely , if @xmath207 is a symmetry of @xmath158 , then it is the extension of some symmetry @xmath76 of @xmath45 . for these symmetries
, we have @xmath210 for all @xmath57 and @xmath209 , hence @xmath211 , @xmath212 , @xmath213 .
if @xmath110 is a thin set of @xmath45 , then @xmath214 is a thin set of @xmath158 , and @xmath215 is a thin set of @xmath195 , @xmath198 , and also of @xmath158 .
we only need to show that any symmetry @xmath207 of @xmath158 is the extension of some symmetry of @xmath45 , the other properties are immediate . by proposition [ car - sym ] , @xmath216 , for all @xmath217 , for some @xmath218 such that @xmath219 . denoting @xmath220 and @xmath221
, we get that @xmath222 .
the properties of @xmath223 imply that @xmath76 is a symmetry of @xmath45 and @xmath224 is a symmetry of @xmath138 .
moreover , since @xmath138 is totally ordered , then by proposition [ sym - tot ] , @xmath224 is the identity map , or @xmath225 .
this shows that @xmath207 is necessarily the extension of the symmetry @xmath76 of @xmath45 .
[ remark - extension3 ] let @xmath45 and @xmath138 be as in proposition [ extension ] .
then the map @xmath226 , @xmath227 is a modulus .
also its restriction to @xmath195 or @xmath198 is a modulus .
if @xmath45 is naturally ordered , then @xmath228 is order preserving .
moreover , the map @xmath229 defined by @xmath230 is an injective and multiplicative morphism , and its image is included in @xmath231 .
the composition @xmath232 equals the identity on @xmath138 , thus the image of @xmath233 is the set of fixed points of @xmath234 .
the map @xmath233 is a semiring morphism if and only if @xmath45 is idempotent .
more generally , if @xmath45 is naturally ordered , then @xmath233 is an order preserving map from @xmath235 to @xmath158 endowed with the natural order @xmath127 , and it satisfies : @xmath236 where @xmath237 is the supremum in the natural order of @xmath158 . the _ symmetrized max - plus semiring _ , which is useful to deal with systems of linear equations over @xmath0 , was introduced in @xcite .
it is also discussed in @xcite , @xcite and @xcite . here
we shall adopt a construction of this semiring using the above extension .
the resulting semiring is isomorphic to the one of @xcite , as shown in @xcite .
we start by defining the symmetrized boolean semiring .
an alternative construction was given in ( * ? ? ?
* section 5.1 ) .
[ def - symb ] let @xmath238 be a set with four elements denoted @xmath239 and @xmath240 .
define the laws @xmath241 on @xmath238 by : @xmath242 then @xmath243 is an idempotent semiring with zero element @xmath176 and unit element @xmath65 , and the map @xmath244 , such that @xmath245 , for all @xmath246 , is a symmetry of @xmath238 , such that @xmath247 .
since @xmath238 is idempotent , it is naturally ordered .
the order of @xmath238 satisfies : ( 100,110 ) ( 45,102)@xmath240 ( 45,100)(-1,-1)38 ( 55,100)(1,-1)38 ( 1,48)@xmath65 ( 90,48)@xmath248 ( 5,43)(1,-1)38 ( 95,43)(-1,-1)38 ( 45,0)@xmath176 the above properties imply that the notation @xmath240 is coherent with the one of definition [ circ ] .
we have @xmath249 , and since @xmath65 and @xmath248 are invertible , the only possible thin set of @xmath238 is @xmath250 .
[ def - smax ] the symmetrized max - plus semiring , @xmath25 , is defined to be @xmath251 , according to proposition [ extension ] . indeed ,
since @xmath238 is zero - sum free , with no zero divisors , the extension @xmath251 is a semiring .
by fact [ remark - extension ] , the symmetry @xmath76 of @xmath238 is extended into the symmetry @xmath207 of @xmath251 .
the definition of the symmetrized max - plus semiring given in @xcite leads to a structure isomorphic to @xmath251 , which was denoted by @xmath25 in this reference . in the present paper
, @xmath25 is directly defined as @xmath251 .
by fact [ remark - extension ] , we have @xmath252 .
moreover @xmath253 is a thin set of @xmath254 , and it is indeed the only possible one .
we shall also use the notations @xmath255 and @xmath256 , thus @xmath257 .
by fact [ remark - extension3 ] , @xmath233 is a morphism , so that we can identify @xmath0 with @xmath258 .
we have @xmath259 . in @xcite
, the elements of @xmath260 are called _ signed _ , thus `` signed '' in this particular semiring is equivalent to `` thin '' . in the idempotent semiring @xmath243 ,
the elements of @xmath261 are not comparable in the natural order , and @xmath173 and @xmath240 are respectively the minimal and maximal elements of @xmath238 .
it is natural to extend the tropical semiring by capturing the phase information , rather than the sign .
the next construction yields a coarse way to do so .
[ groupext ] let @xmath262 be a group , equip it with the trivial order @xmath116 such that every two elements are incomparable ( i.e. , @xmath119 if and only if @xmath263 ) , and add a minimal and maximal element to @xmath264 , denoted respectively @xmath173 and @xmath240 , such that @xmath173 is absorbing for the multiplication in @xmath265 , and @xmath240 is absorbing for the multiplication in @xmath266 . then @xmath267 is an idempotent semiring in which @xmath268 for all @xmath269 , such that @xmath270 .
assume that there exists @xmath271 , such that @xmath272 and @xmath80 commutes with all elements of @xmath264 .
then the map @xmath273 is a non trivial symmetry of @xmath274 , and since @xmath275 for all @xmath276 , we get that @xmath277 and that @xmath278 is the only thin set of @xmath274 .
since @xmath274 is zero - sum free without zero divisors , one can then construct @xmath279 with the thin set @xmath280 .
when @xmath264 is the group with two elements ( of order @xmath281 , so isomorphic to the additive group @xmath282 ) , we recover the semirings @xmath238 and @xmath254 .
when @xmath264 is the unit circle @xmath283 of @xmath181 , we obtain a semiring @xmath284 , with only one possible non trivial symmetry obtained with @xmath285 .
a more powerful semiring than @xmath284 is obtained by the following construction which is a variant of the one of the complex tropical hyperfield that viro @xcite made , using a different set of axioms , see remark [ viro - remark ] below .
[ ex - viro ] let @xmath286 ( for `` phases '' ) denote the set of closed convex cones of @xmath181 seen as a real 2-dimensional space , that is the set of angular sectors of @xmath181 between two half - lines with angle less or equal to @xmath287 or equal to @xmath288 , together with the singleton @xmath289 ( the trivial cone ) .
consider the following laws on @xmath286 : the sum @xmath290 of two elements @xmath291 is the closed convex hull of @xmath292 , and the product @xmath293 is the closed convex hull of the set of complex numbers @xmath294 with @xmath295 and @xmath296 .
then @xmath286 is an idempotent semiring .
the zero is @xmath297 , the unit is the half - line of positive reals , and the invertible elements are the half - lines . taking @xmath298 equal to the set of @xmath87 for @xmath295 , we obtain a symmetry of @xmath286 , which is the only symmetry of @xmath286 different from identity . in that case
, @xmath299 is the subset of @xmath286 consisting of @xmath173 , all lines , and the plane , and we can consider for the thin set @xmath300 the set of half - lines and @xmath173 . we can then construct the semiring @xmath301 equipped with the thin set @xmath302 .
we call this semiring the _ phase extension _ of the tropical semiring . similarly to example
[ ex - best ] , an element @xmath303 of @xmath301 may be thought of as an abstraction of the set of asymptotic expansions of the form @xmath184 , when @xmath185 goes to @xmath186 , where @xmath96 is required to belong to the relative interior of @xmath304 , denoted by @xmath305 .
recall that the relative interior of a convex set is the interior of this set with respect to the topology of the affine space that it generates .
for instance , the relative interior of a closed half - line is an open half - line .
if @xmath306 , and if @xmath307 for some @xmath308 , then , it can readily be checked that @xmath309 , where @xmath310 and @xmath311 . similarly , the product of the semiring @xmath286 is consistent with the one of asymptotic expansions .
note that when the cone @xmath304 is either a line or the whole set @xmath312 , @xmath112 is in the relative interior of @xmath304 .
then the corresponding asymptotic expansion @xmath184 may reduce to @xmath313 , as @xmath314 is allowed .
the elements of @xmath302 correspond to asymptotic expansions with a well defined information on the angle , whereas an element @xmath303 such that @xmath304 is a pointed cone ( a sector of angle strictly inferior to @xmath287 ) correspond to asymptotic expansions having their leading term in a given angular sector .
[ viro - remark ] a related encoding was proposed by viro in @xcite in a different setting , with his _ complex tropical hyperfield _ @xmath315 .
a hyperfield is a set endowed with a multivalued addition and univalued multiplication , that satisfy distributivity and invertibility properties similar to those of semifields .
the hyperfield @xmath315 is the set of complex numbers @xmath181 endowed with a multivalued addition and the usual multiplication .
this allows one to see a non zero complex number @xmath316 as an encoding of asymptotic expansions of the form @xmath317 , when @xmath185 goes to @xmath186 , with @xmath318 .
hence , the phase extension of the tropical semiring and the complex tropical hyperfield provide two abstractions of the arithmetics of asymptotic expansions ( or of puiseux series ) .
the two abstractions differ , however , in the handling of the element @xmath319 . indeed ,
if @xmath320 , then , @xmath319 is defined to be @xmath321 in @xmath315 .
this set may be thought of as an encoding of all the expansions in @xmath185 that are @xmath322 . if @xmath323 and @xmath324 is the angle of the half - line @xmath304 , then @xmath319 encodes all the asymptotic expansions @xmath317 with @xmath325 ( so we get an extra bit of information by comparison with @xmath326 ) .
note that we may also identify @xmath302 with @xmath181 and @xmath301 with subsets of @xmath181 , by means of the bijective map @xmath327 where @xmath324 is as above , and @xmath328 .
again , in this identification , the multiplication and addition of @xmath315 and @xmath302 coincide except for @xmath319 .
moreover , in this way , @xmath302 is not a hyperfield ( since @xmath329 ) , and @xmath315 can not be put in the form of @xmath330 for some semiring @xmath45 .
izhakian introduced in @xcite an extension of the tropical semiring , which can be cast in the previous general construction .
we shall also see that some of the supertropical semifields of izhakian and rowen @xcite can be reduced to the previous construction .
the following presentation is a simplified version of @xcite .
[ def - ti ] let @xmath331 be the semiring which is the quotient of the semiring @xmath53 of non - negative integers by the equivalence relation which identifies all numbers greater than or equal to @xmath281 .
the _ bi - valued tropical semiring _ is @xmath332 in the sense of proposition [ extension ] .
indeed , @xmath333 is zero - sum free and without zero divisors , thus @xmath334 is a subsemiring of @xmath335 .
moreover @xmath336 is isomorphic to the extended tropical semiring defined in @xcite , see @xcite for details .
( in the present paper , we prefer to use the term _ bi - valued _ rather than _ extended _ since other extensions of the tropical semiring are considered . )
recall that this algebraic structure encodes whether the maximum in an expression is attained once or at least twice .
the semiring @xmath333 is not idempotent so the injection @xmath233 is not a morphism .
however , @xmath333 is naturally ordered ( by the usual order of @xmath53 ) , so is the semiring @xmath336 , and @xmath233 satisfies .
the only element @xmath80 of @xmath333 such that @xmath78 is equal to @xmath337 , thus the only symmetry of @xmath333 is the identity map .
since by fact [ remark - extension ] , a symmetry of @xmath336 is the extension of a symmetry of @xmath333 , the only symmetry of @xmath336 is the identity map . in @xmath333 , we have @xmath338 , @xmath339 and the only possible thin set is @xmath340 .
then @xmath341 and @xmath342 is a thin set of @xmath336 , which is the only possible one .
[ rem - supertropical ] in @xcite , in the particular context of the semiring @xmath336 , the names `` reals '' and `` ghosts '' were given to what we call here `` thin '' and `` balanced '' elements .
the construction of the bi - valued tropical semiring has been generalized to the notion of _ supertropical _ semifield or semiring in @xcite and of _ layered _ semiring in @xcite .
supertropical semirings are special cases of semirings with a symmetry and a modulus .
indeed , one can show that the triple @xmath343 is a _ supertropical semiring _ in the sense of @xcite if and only if the following conditions hold .
( i ) @xmath45 is a naturally ordered semiring , endowed with the identity symmetry , such that @xmath108 is a totally ordered idempotent semiring , the map @xmath344 is a modulus , and @xmath45 satisfies the additional properties : @xmath345 if @xmath346 and @xmath347 if @xmath348 .
( ii ) @xmath349 is an ideal of @xmath45 containing @xmath108 , and @xmath350 is the map from @xmath45 to @xmath349 such that @xmath351 .
note that the idempotency of @xmath350 follows from the fact that @xmath228 is a morphism .
similarly , a supertropical semiring is a _ supertropical semifield _ if the following additional conditions hold .
( iii ) @xmath352 is a multiplicative commutative group such that the map @xmath228 is onto from @xmath353 to @xmath354 .
the latter properties mean that @xmath355 is the unique thin set @xmath110 of @xmath45 , and that any element of @xmath108 can be written as @xmath356 with @xmath357 .
( iv ) @xmath358 and @xmath359 .
since conditions ( ii ) and ( iv ) concern only @xmath349 and @xmath350 and do not affect @xmath45 , we shall use in the sequel the name `` supertropical semifield '' for any semiring @xmath45 satisfying the above conditions ( i ) and ( iii ) .
[ ex - supertropical ] the following construction gives the main example of a supertropical semifield .
consider a group @xmath264 and its extension @xmath274 defined as in example [ groupext ] .
let us consider on @xmath274 the additive law @xmath98 such that @xmath360 for all @xmath361 .
then @xmath277 and @xmath279 is a supertropical semifield .
more generally , if @xmath138 is a totally ordered idempotent semifield , then @xmath362 is a supertropical semifield .
we next recall or establish some properties of a combinatorial nature , which will be useful when studying cramer systems over extended tropical semirings . [ de : dets ] let @xmath363 be a semiring with symmetry and @xmath364 .
we define the _ determinant _ @xmath365 of @xmath38 to be the element of @xmath45 given by the usual formula @xmath366 understanding that @xmath367 depending on the even or odd parity of @xmath224 . in a general semiring ,
the determinant is not a multiplicative morphism , however the following identities hold .
let @xmath45 be a semiring with symmetry , then for all @xmath368 , we have : @xmath369 or more precisely : @xmath370 let @xmath45 be a semiring .
a matrix @xmath371 is _ monomial _ if it can be written as @xmath372 where @xmath373 is a diagonal matrix ( @xmath374 for @xmath375 ) , and @xmath376 is the matrix of the permutation @xmath377 ( that is @xmath378 if @xmath379 and @xmath380 otherwise ) . note that , in any semiring @xmath45 , a permutation matrix is invertible since @xmath381 .
hence @xmath382 can be written as @xmath383 if and only if it can be written as @xmath384 , by taking @xmath385 .
this also implies that @xmath372 is invertible in @xmath386 if and only if all diagonal entries of @xmath373 are invertible in @xmath45 .
the following property is easy to check .
[ prop - det - mult ] let @xmath45 be a semiring with symmetry , then for all @xmath368 , such that @xmath38 or @xmath387 is monomial , we have : @xmath388 we introduce now the concept of adjugate matrices . [ def - adjugate ]
let @xmath45 be a semiring with symmetry .
if @xmath389 , we denote by @xmath390 the @xmath391 submatrix in which row @xmath392 and column @xmath393 are suppressed .
define the _
@xmath394-cofactor _ of @xmath38 to be @xmath395 and the _ adjugate matrix _ of @xmath38 to be the @xmath39 matrix @xmath396 with @xmath397-entry : @xmath398 using proposition [ prop - det - mult ] we also obtain the following identities .
[ llem - util07 ] let @xmath45 be a semiring with symmetry , then for all @xmath368 , such that @xmath38 or @xmath387 is monomial , we have : @xmath399 [ lhabitude ] let @xmath45 be a semiring with symmetry , and let @xmath400 be an invertible monomial matrix .
then @xmath401 is invertible and @xmath402 we next recall some properties concerning the idempotent semiring @xmath138 arising as an ingredient of the semiring extension .
although we shall only apply these properties when @xmath138 is totally ordered , we state the properties in their full generality as far as possible .
let @xmath403 be an idempotent semiring equipped with the trivial symmetry and with the natural order .
then @xmath404 , so the determinant of a matrix @xmath38 , @xmath365 coincides with its permanent , denoted by @xmath405 .
the adjugate matrix @xmath396 has a useful interpretation in terms of maximal weights of paths . to see this ,
let us first recall the definition and basic properties of the _ kleene star _ of a matrix in an idempotent semiring .
recall that to an @xmath39 matrix @xmath38 with entries in a semiring @xmath403 , one associates a digraph @xmath406 with nodes @xmath407 and an arc @xmath408 if @xmath409 .
the _ weight _ of a path @xmath410 in @xmath406 is defined as @xmath411 , and its length @xmath412 is equal to @xmath413 ( the number of arcs ) .
this definition applies in particular to _ circuits _ , which are closed paths , meaning that @xmath414 .
we denote by @xmath415 the identity matrix ( with diagonal elements equal to @xmath337 and non - diagonal elements equal to @xmath112 ) .
[ propdef ] let @xmath403 be any idempotent semiring equipped with the natural order .
let @xmath416 be a matrix such that every circuit of @xmath38 has a weight less than or equal to @xmath337
. then @xmath417 when @xmath138 is idempotent , the sum is equivalent to the supremum for the natural order of @xmath138 , hence @xmath418 represents the maximal weight of a path from @xmath392 to @xmath393 . moreover , when every circuit of @xmath38 has a weight less than or equal to @xmath337 , @xmath418 is also equal to the maximal weight of an elementary path from @xmath392 to @xmath393 ( when @xmath419 , this means a path with length @xmath112 ) .
the following result generalizes a theorem of yoeli @xcite , which was stated when the idempotent semiring has a maximal element equal to @xmath337 .
[ th - yoeli ] let @xmath403 be an idempotent semiring equipped with the trivial symmetry and the natural order denoted by @xmath116 .
let @xmath420 be a matrix such that @xmath421 and @xmath422 .
then every circuit of @xmath38 has a weight less than or equal to @xmath337 , and @xmath423 let @xmath420 be a matrix such that @xmath421 and @xmath422 .
let @xmath424 be a circuit of @xmath406 .
one can construct the permutation @xmath425 containing this circuit and all the circuits with one element not in @xmath426 , that is @xmath427 for @xmath428 , and @xmath429 for @xmath430 .
since @xmath431 , the weight of this permutation @xmath432 is equal to the weight @xmath433 of the circuit @xmath426 . since @xmath422 is the sum , thus the supremum for the natural order , of the weights of all permutations , we get that @xmath434 , which shows that every circuit of @xmath38 has a weight less than or equal to @xmath337 to prove the last assertion of the theorem , we need to show that @xmath435 for all @xmath436 $ ] . as remarked after proposition - definition [ propdef ] , since all circuits have a weight less than or equal to @xmath337 , @xmath418 is equal to the maximal weight of an elementary path from @xmath392 to @xmath393 . in particular , @xmath437 , and
since @xmath438 has the same properties as @xmath38 ( the diagonal coefficients are equal to @xmath337 and all circuits have a weight less than or equal to @xmath337 ) , we deduce that @xmath439 .
assume now that @xmath375 .
we have @xmath440\setminus \{j\ } } a_{l\sigma(l)}\enspace,\ ] ] where the sum is taken over all bijections from @xmath9\setminus \{j\}$ ] to @xmath9\setminus \{i\}$ ] .
since a map @xmath441\setminus \{j\}\to [ n]\setminus \{i\}$ ] is a bijection if and only if it can be completed into a permutation of @xmath9 $ ] by taking @xmath442 , the above sum can be taken equivalently over all @xmath425 such that @xmath442 .
we say that a circuit @xmath443 of @xmath406 contains the arc @xmath397 if there exists @xmath428 such that @xmath444 , @xmath445 .
then @xmath446 is an elementary path from @xmath447 to @xmath448 if and only if @xmath449 is an elementary circuit , containing the arc @xmath450 .
let @xmath451 be such a path and @xmath452 .
completing @xmath426 as above to a permutation @xmath425 , containing all circuits with one element not in @xmath426 , we get that @xmath442 , and that @xmath453\setminus \{j\ } } a_{l\sigma(l)}$ ] . since holds with a sum over all @xmath425 such that @xmath442
, we obtain that @xmath454 , and since this holds for all elementary paths from @xmath392 to @xmath393 , we deduce that @xmath455 . to show the reverse inequality , let @xmath456 be such that @xmath442 .
decomposing @xmath224 into elementary cycles , we get in particular a cycle @xmath426 containing the arc @xmath450 .
let @xmath451 be the elementary path from @xmath392 to @xmath393 such that @xmath452 , we deduce that @xmath457\setminus \{j\ } } a_{l\sigma(l)}= { \mathrm{w}}_a(p)\cdot \prod_{c ' } { \mathrm{w}}_a(c')$ ] , where the last product is taken over all cycles @xmath458 of @xmath224 different from @xmath426 .
since all cycles have weights less than or equal to @xmath337 , we get that @xmath457\setminus \{j\ } } a_{l\sigma(l)}{\leqslant}{\mathrm{w}}_a(p){\leqslant}a^*_{ij}$ ] . by applying with a sum over all @xmath425 such that @xmath442
, we obtain that @xmath459 , and so the equality holds .
this finishes the proof of the theorem .
the following proposition gives a semifield version of a well known duality result concerning the optimal assignment problem .
it will allow convenient normalizations .
[ prop - hungarian ] let @xmath382 be an @xmath39 matrix with entries in a totally ordered idempotent semifield @xmath138 , and assume that @xmath460 .
then there exist two @xmath5-dimensional vectors @xmath461 with entries in @xmath462 such that @xmath463\ ] ] and @xmath464 for all @xmath397 such that @xmath379 for every optimal permutation @xmath224 , i.e. , for every permutation @xmath224 such that @xmath465 } c_{i\sigma ( i ) } \enspace .\ ] ] in particular , @xmath465 } u_i \prod_{j\in [ n ] } v_j\ ] ] this is a byproduct of the termination of the hungarian algorithm .
we refer the reader to @xcite for more information on this algorithm .
the latter is usually stated for matrices with entries in the ordered group @xmath466 completed by the @xmath8 element , or equivalently for matrices with entries in the ordered group @xmath467 of strictly positive real numbers , completed by the @xmath112 element . in the latter context
, it allows one to compute @xmath468 .
let us call a _ row _ ( resp .
_ column _ ) _ scaling _ the operation of multiplying by the inverse of a non - zero number a given row ( resp .
column ) of a matrix .
the hungarian algorithm performs a finite number of row and column scalings , reaching eventually a matrix @xmath469 such that @xmath470 for all @xmath471 and @xmath472 for all @xmath397 in a collection of couples of indices among which @xmath5 are _ independent _ , meaning that none of them belong to the same row or column .
then these independent @xmath397 define an optimal permutation , and all the conclusions of the proposition are valid .
the algorithm can be readily checked to be valid when the entries of @xmath382 belong to any totally ordered semifield .
we note that some generalizations of network flow problems to ordered algebraic structures were studied in ( * ? ? ?
12 ) , the previous result could also be derived from results there .
a variant of the following result has appeared in the work of butkovic @xcite .
it allows one to reduce matrices to a `` normal form '' in which the diagonal consists of unit elements and all other elements are not greater than the unit . [ cor - but ] let @xmath382 be an @xmath39 matrix with entries in a totally ordered idempotent semifield @xmath138 , and assume that @xmath460 .
then there exist two diagonal matrices @xmath373 and @xmath473 with invertible diagonal entries , and a permutation matrix @xmath474 ,
such that @xmath475 satisfies @xmath476,\qquad b_{ii}=1 , \qquad \forall i\in [ n ] \enspace .\ ] ] corollary [ cor - but ] shows that in the special case of a totally ordered idempotent semifield , as soon as @xmath477 , we may reduce @xmath382 by diagonal scaling and permutation to a matrix @xmath38 satisfying the assumptions of yoeli s theorem .
in the sequel , we shall consider a semiring @xmath45 with symmetry and a thin set @xmath110 satisfying the following properties , which will allow us to eliminate variables in order to solve tropical linear systems .
[ lrrr ] for @xmath478 , we have that @xmath479 implies @xmath480 .
[ lrrr1 ] the set of non - zero thin elements @xmath481 is closed under multiplication .
so is @xmath110 , a fortiori .
[ pty - weaktrans ] for all @xmath482 , @xmath357 , @xmath483 , @xmath484 , and @xmath485 , we have @xmath486 [ def - tropextension ] if a semiring @xmath45 with a symmetry and a thin set @xmath110 satisfies properties [ lrrr1 ] and [ pty - weaktrans ] , we will say that it _
allows weak balance elimination_. if it satisfies also property [ lrrr ] , then we will say that it _ allows strong balance elimination_. it was pointed out in @xcite ( see also ( * ? ? ? * section 6 ) ) that the symmetrized max - plus semiring @xmath25 satisfies properties [ lrrr ] [ pty - weaktrans ] .
it was also observed in @xcite that so does the bi - valued tropical semiring @xmath336 .
note that in the latter reference , property [ lrrr1 ] was replaced by the stronger property that the set @xmath487 is exactly the set of all invertible elements in @xmath45 , but this stronger property will not always be needed . in
@xcite a proof of property [ pty - weaktrans ] was given specially for @xmath25 , we shall give now some sufficient conditions for the above properties to hold , which allow one to check them easily for the semirings @xmath238 and @xmath333 and deduce them for @xmath25 and @xmath336 .
in particular , we shall also consider the following properties : [ pty - weaktrans0 ] for all @xmath488 , we have @xmath489 [ pty - weaktrans1 ] for all @xmath490 , we have @xmath491 which is for @xmath492 and @xmath493 .
[ pty - thinexact ] @xmath487 .
[ lrrr2 ] the set @xmath45 is additively generated by @xmath110 , which means that any element of @xmath45 is the sum of a finite number of elements of @xmath110 .
[ lemma - pty - weaktrans1 ] @xmath45 allows weak balance elimination if and only if properties [ lrrr1 ] and [ pty - weaktrans1 ] hold together . since property [ pty - weaktrans ] implies in particular property [ pty - weaktrans1 ]
, we get the `` only if '' part of the assertion of the lemma . for the `` if '' part ,
let us assume that properties [ lrrr1 ] and [ pty - weaktrans1 ] hold , and show that property [ pty - weaktrans ] holds .
let @xmath482 , @xmath357 , @xmath483 , @xmath484 , @xmath485 , and @xmath494 be such that @xmath495 and @xmath496 .
let us show that @xmath497 .
since , by property [ lrrr1 ] , @xmath110 is stable under product , @xmath498}\in ( \ss^\vee)^p$ ] .
since @xmath108 is an ideal of @xmath45 , multiplying the equation @xmath496 by @xmath96 , we get that @xmath499 .
then it remains to show the above implication for @xmath337 , @xmath500 and @xmath501 instead of @xmath96 , @xmath41 and @xmath502 respectively . without loss of generality
we can assume that @xmath493 .
moreover , since @xmath496 is equivalent to @xmath503 for all @xmath18 $ ] , where @xmath504 denotes the @xmath392th row of @xmath382 , it is sufficient to prove the above implication for each row of @xmath382 instead of @xmath382 .
we can thus assume that @xmath505 .
let @xmath506 , and assume that @xmath507 and @xmath496 .
the relation @xmath508 is equivalent to @xmath509 . from , which holds by property [ pty - weaktrans1 ] , and @xmath510 , we deduce that @xmath511 . now exchanging the sides of @xmath512 and @xmath513 , and applying with @xmath514
, we can replace @xmath515 by @xmath516 in the previous balance equation .
doing this inductively on all @xmath517 , we obtain @xmath518 , which concludes the proof .
[ lemma - pty - weaktrans2 ] properties [ lrrr ] and [ pty - thinexact ] together imply property [ pty - weaktrans0 ] .
assume property [ lrrr ] holds .
let @xmath488 , @xmath519 be such that @xmath520 and @xmath521 .
we need to show that @xmath522 .
since @xmath519 , if @xmath523 then property [ lrrr ] implies that @xmath524 , so that @xmath522 .
similarly , @xmath522 if @xmath525 . otherwise , @xmath40 and @xmath526 , which implies that @xmath40 and @xmath527 by property [ pty - thinexact ] .
hence @xmath528 , which means that @xmath522 .
[ lemma - balance ] properties [ pty - weaktrans0 ] , [ lrrr1 ] , and [ lrrr2 ] all together imply property [ pty - weaktrans1 ] , hence , property [ pty - weaktrans ] and that @xmath45 allows weak balance elimination .
assume properties [ pty - weaktrans0 ] , [ lrrr1 ] , and [ lrrr2 ] hold .
then holds when @xmath492 and @xmath529 .
let us show that also holds when @xmath492 , @xmath530 , and @xmath531 .
this will mean that property [ pty - weaktrans1 ] holds and will imply by lemma [ lemma - pty - weaktrans1 ] that property [ pty - weaktrans ] holds .
so let @xmath532 , and @xmath519 be such that @xmath507 and @xmath496 , and let us show that @xmath518 . if @xmath533 , then by property [ lrrr1 ] , @xmath534 .
multiplying @xmath507 by @xmath382 , we get that @xmath535 , and since @xmath496 and @xmath536 , property [ pty - weaktrans0 ] implies @xmath518 .
this shows that holds when @xmath492 , @xmath530 , and @xmath533 .
assume now that @xmath537 .
then by property [ lrrr2 ] , there exist @xmath538 such that @xmath539 .
the relation @xmath508 is then equivalent to @xmath540 .
since @xmath541 , @xmath542 and @xmath507 , applying with @xmath492 , @xmath530 , @xmath541 instead of @xmath382 and @xmath543 instead of @xmath502 ( the implication is already known to hold in that case ) , we get that @xmath544 . now exchanging the sides of @xmath545 and @xmath546 , and applying with @xmath492 , @xmath530 , @xmath547 instead of @xmath382
, we can replace @xmath548 by @xmath549 in the previous balance equation .
doing this inductively , we obtain @xmath518 , hence holds when @xmath492 , @xmath530 , and for all @xmath531 .
[ cor - balance ] properties [ lrrr ] , [ lrrr1 ] , [ lrrr2 ] , together with either property [ pty - weaktrans0 ] or [ pty - thinexact ] , imply that @xmath45 allows strong balance elimination .
[ extension - balance ] let @xmath363 be a semiring with symmetry , let @xmath110 be a thin set of @xmath45 , and let @xmath157 be a totally ordered idempotent semiring
. denote by @xmath550 any of the semiring extensions @xmath158 , @xmath195 or @xmath198 defined in proposition [ extension ] endowed with the extension of the symmetry of @xmath45 defined in fact [ remark - extension ] .
here , @xmath550 is assumed to be a subsemiring of @xmath158 .
consider the thin set @xmath551 if @xmath552 or @xmath553 , and @xmath554 if @xmath555 .
then the following properties hold : * @xmath550 satisfies property [ lrrr ] ( resp .
[ lrrr1 ] , resp .
[ pty - weaktrans1 ] ) if and only if @xmath45 does .
* @xmath550 allows weak balance elimination if and only if @xmath45 does . *
@xmath550 allows strong balance elimination if and only if @xmath45 does .
assertions ( b ) and ( c ) follow from assertion ( a ) , lemma [ lemma - pty - weaktrans1 ] , and the definition of strong balance elimination .
let us first prove the `` only if '' part of assertion ( a ) of the proposition .
assume that @xmath45 , @xmath110 , @xmath550 , @xmath556 are as in the statement .
by fact [ remark - extension0 ] , @xmath45 is isomorphic by @xmath557 to a subsemiring of @xmath550 .
moreover , by definition , the map @xmath557 is compatible with the symmetries of @xmath45 and @xmath550 , hence @xmath558 then on @xmath45 the balance relation of @xmath45 coincides with the one of @xmath550 .
also , by definition of @xmath556 , we have @xmath559 , and @xmath560 . from this
, we deduce that property [ lrrr ] ( resp .
[ lrrr1 ] , resp .
[ pty - weaktrans1 ] ) for @xmath550 implies the same for @xmath45 .
let us now show the `` if '' part of the assertion ( a ) . by definition , @xmath561 or @xmath562 . since @xmath563 and @xmath138 are closed by multiplication , it is clear that property [ lrrr1 ] for @xmath45 implies that the same property is valid for @xmath550 .
assume now that property [ lrrr ] holds for @xmath45 and let us prove it for @xmath550 .
remark that for any semiring @xmath45 with symmetry , the assertion of property [ lrrr ] is equivalent to the same assertion with @xmath110 replaced by @xmath481 . indeed , when @xmath564 with @xmath565 or @xmath566 , the equation @xmath479 implies that @xmath567 .
thus @xmath568 .
hence @xmath569 .
then it is sufficient to prove the assertion of property [ lrrr ] for @xmath570 instead of @xmath550 .
let @xmath571 and @xmath572 be such that @xmath479 . since the equation @xmath479 is equivalent to @xmath573 and since @xmath574 if @xmath575 , @xmath576 if @xmath577 , and @xmath578 if @xmath205 , we deduce that @xmath205 and @xmath579 . since @xmath580 , we deduce that @xmath581 .
since @xmath582 , we also get that @xmath583 .
hence if property [ lrrr ] holds for @xmath45 , the above properties imply @xmath584 , so @xmath480 , which shows that property [ lrrr ] also holds for @xmath550 .
assume now that property [ pty - weaktrans1 ] holds for @xmath45 and let us prove it for @xmath550 .
let @xmath585 , and @xmath586 be such that @xmath479 and @xmath587 .
we need to show that @xmath588 .
it is easy to show that this holds when @xmath589 , since then @xmath590 so that @xmath591 . also this holds when @xmath592 , since then @xmath593 and thus again @xmath589 ( since @xmath594 ) , and when @xmath595 since then @xmath596 .
so we can assume that @xmath597 , and @xmath598 .
recall that @xmath599 when @xmath555 and @xmath600 otherwise .
it follows that @xmath601 .
moreover , we have @xmath602 when @xmath552 or @xmath198 . now
since @xmath604 , and @xmath605 , we get that @xmath606 when @xmath552 or @xmath198 .
then in all cases @xmath607 .
multiplying both terms of the relation @xmath479 by @xmath608 , and replacing @xmath41 by @xmath609 , @xmath610 by @xmath611 and @xmath612 by @xmath613 , we are reduced to the case where @xmath614 . then by the same arguments as above we obtain from @xmath479 that either @xmath577 and @xmath615 , or @xmath616 and @xmath617 .
similarly , from @xmath587 , we obtain that either @xmath618 and @xmath619 , or @xmath620 and @xmath621 , or @xmath622 and @xmath623 so that @xmath624 . when @xmath577 and @xmath618 , we get that @xmath625 , so @xmath626 and @xmath588 .
when @xmath577 and @xmath627 , we get that @xmath628 , and again @xmath588 .
when @xmath205 and @xmath618 , we get that @xmath629 , and again @xmath588 . when @xmath205 and @xmath622 , we get that @xmath630 .
since @xmath601 , @xmath617 and @xmath631 , we deduce from property [ pty - weaktrans1 ] for @xmath45 , that @xmath632 , hence @xmath591 again . when @xmath633 , we get that @xmath617 and @xmath621 .
since we also have @xmath634 , we deduce from property [ pty - weaktrans0 ] for @xmath45 , that @xmath635 .
this implies again @xmath588 .
since in all cases @xmath588 , this shows property [ pty - weaktrans1 ] for @xmath550 . note that one can also prove similar equivalences to the ones of item ( a ) of proposition [ extension - balance ] for properties [ pty - weaktrans0 ] and [ pty - thinexact ] . also , one can do the same for property [ lrrr2 ] if @xmath553 .
we do not detail these equivalences here , since we are interested mostly in `` balance elimination '' .
we are now able to give examples of semirings allowing balance elimination , as a consequence of proposition [ extension - balance ] , lemmas [ lemma - pty - weaktrans2 ] and [ lemma - balance ] , and corollary [ cor - balance ] .
the semirings @xmath636 and @xmath637 satisfy properties [ pty - thinexact ] , [ lrrr ] , [ lrrr1 ] and [ lrrr2 ] .
then these semirings allow strong balance elimination .
hence so do @xmath638 and @xmath334 .
the semiring @xmath639 of example [ ex - viro ] , i.e. , the set of closed convex cones of @xmath181 , does not satisfy property [ pty - thinexact ] , but it does satisfy properties [ pty - weaktrans0 ] , [ lrrr ] , [ lrrr1 ] , and [ lrrr2 ] .
so , @xmath286 allows strong balance elimination , and so does @xmath301 ( the phase extension of the tropical semiring ) .
note that for the same semirings @xmath45 , the semirings @xmath640 and @xmath641 also allow strong balance elimination .
let @xmath45 be an integral domain ( a ring without zero divisors ) .
since @xmath642 , the balance relation reduces to the equality relation . taking @xmath643
, we get that @xmath45 allows trivially strong balance elimination .
hence , the semirings @xmath158 and @xmath195 also allow strong balance elimination .
however @xmath198 is not a semiring , since @xmath45 is not zero sum free .
recall that when @xmath644 or @xmath645 , an element @xmath646 is equivalent to the asymptotic expansion @xmath184 , when @xmath185 goes to @xmath186 ( example [ ex - best ] ) .
let @xmath45 be a supertropical semifield ( see remark [ rem - supertropical ] )
. then properties [ pty - weaktrans0 ] , [ lrrr1 ] and [ lrrr2 ] hold , thus @xmath45 allows weak balance elimination , by lemma [ lemma - balance ] .
if the map @xmath228 from @xmath110 to @xmath108 is not injective , then property [ lrrr ] does not hold , thus @xmath45 does not allow strong balance elimination .
the first assertion can be checked easily . for the second one , if @xmath228 is not injective , there exist @xmath647 , @xmath270 such that @xmath348 .
since @xmath648 , we get that @xmath649 , thus property [ lrrr ] does not hold .
we state here a general cramer theorem . in the special case of the symmetrized max - plus semiring , this was established by plus @xcite ,
see also @xcite and ( * ? ? ?
* theorem 6.4 ) .
it is remarked in the latter reference ( see the paragraph before theorem 6.6 of @xcite ) that the proof of ( * ? ? ?
* theorem 6.4 ) is valid in any semiring satisfying properties [ lrrr ] , [ pty - weaktrans ] , and the property that @xmath487 is the set of invertible elements .
however looking at the proof of ( * ? ? ?
* theorem 6.4 ) more carefully , one see that the latter property can be replaced by property [ lrrr1 ] , and that the first part does not use property [ lrrr ] .
this leads to the following general result .
[ th - cramer ] let @xmath45 be a semiring with a symmetry and a thin set @xmath110 , allowing weak balance elimination ( definition [ def - tropextension ] ) .
let @xmath650 and @xmath651 , then 1 .
[ cramer1 ] every thin solution @xmath41 ( such that @xmath652 ) of the linear system @xmath653 satisfies the relation @xmath654 2 .
assume also that @xmath45 allows strong balance elimination ( definition [ def - tropextension ] ) , that the vector @xmath655 is thin and that @xmath656 is invertible in @xmath45 .
then @xmath657 is the unique thin solution of @xmath658 .
the proof follows exactly the same lines as the proof of ( * ? ? ?
* theorem 6.4 ) , so we do not reproduce it .
let us just remark that it relies on an elimination argument , in which `` equations '' involving balances rather than equalities are considered .
the above uniqueness part can be reformulated equivalently in the following homogeneous form . for a matrix @xmath659 over a semiring @xmath45 , and @xmath660 $ ]
, we denote by @xmath661 the @xmath662 matrix obtained from @xmath38 by deleting the @xmath413th column , and by @xmath663 the @xmath413th column of @xmath38 .
[ cor - cramer ] let @xmath45 be a semiring allowing strong balance elimination .
let @xmath664 , and let @xmath665 be such that @xmath666 , for all @xmath667 $ ]
. then if @xmath668 is thin and has at least one invertible entry ( which is the case when @xmath668 is non - zero and @xmath481 is exactly the set of invertible elements of @xmath45 ) , then any thin solution of @xmath37 is a thin multiple of @xmath668 .
let @xmath668 be as in the statement of the theorem and assume that @xmath668 is thin and has at least one invertible entry , for instance @xmath669 is invertible .
let @xmath41 be a thin solution of @xmath670 .
then taking @xmath671 and @xmath672 , we get that @xmath673 is invertible and @xmath674 is thin for all @xmath18 $ ] .
let @xmath675 be such that @xmath676 for @xmath18 $ ] .
then @xmath610 is a thin solution of @xmath677 , and applying the uniqueness part of theorem [ th - cramer ] to this system , we get that @xmath678 , for all @xmath679 $ ] .
hence @xmath680 , with @xmath681 .
we can show that the same conclusion holds for any @xmath682 $ ] , such that @xmath683 is invertible , which implies the corollary .
in this section we consider a semiring @xmath684 with a symmetry , a thin set @xmath685 , and a modulus taking its values in a totally ordered semiring @xmath138 .
for instance @xmath686 where @xmath45 is a zero - sum free semiring with a symmetry and without zero divisors , and @xmath687 . we shall study the square affine systems @xmath653 where @xmath38 is an @xmath39 matrix and @xmath40 is a vector of dimension @xmath5 , all with entries in @xmath125 .
we shall look for the solutions @xmath688 with _ thin _ entries .
the existence results that we shall state in the next sections extend the ones proved in @xcite for @xmath25 .
as the latter results , they will be derived as a byproduct of the convergence of an iterative jacobi - type algorithm to solve the system @xmath658 . recall that the usual jacobi algorithm constructs a sequence which is known to converge to the solution of a linear system under a strict diagonal dominance property . here
we shall first transform the initial system to meet a diagonal dominance property .
then we shall construct a monotone sequence of thin vectors satisfying balance relations which are similar to the equations used in the definition of the usual jacobi algorithm . to show that these thin vectors do exist and
that the resulting sequence does converge , we shall however need some new definitions and properties concerning the semiring @xmath125 , which are somehow more technical than the properties used to establish the cramer theorem , theorem [ th - cramer ] .
we define on @xmath125 the relation @xmath689 , which is finer than the balance relation @xmath35 : latexmath:[\[x{\;|\!\nabla\!|\;}y\quad\leftrightarrow \quad x{{\,\nabla\ , } } y\ ; \text{and } note that when @xmath125 is a semiring extension as in proposition [ extension ] and @xmath691 is as in fact [ remark - extension0 ] , we have : @xmath692 [ pty_order ] @xmath125 is naturally ordered and for all @xmath693 we have : @xmath694 [ pty_order - finite ] @xmath125 is naturally ordered and for all @xmath695 any totally ordered subset of @xmath696 is finite . the following property is stronger .
[ pty_order - equal ] @xmath125 is naturally ordered and for all @xmath693 we have : @xmath697 [ pty_inv ] for all @xmath698 such that @xmath699 is invertible in @xmath138 , there exists @xmath700 such that for all @xmath701 , @xmath702 [ monotone - def ] let @xmath125 be a semiring with a symmetry , a thin set @xmath685 , and a modulus taking its values in a totally ordered semiring @xmath138 .
we shall say that @xmath125 _ allows the construction of monotone algorithms _ if it satisfies properties [ pty_order ] and [ pty_inv ] .
if it satisfies in addition property [ pty_order - finite ] , we shall say that it _ allows the convergence of monotone algorithms_. from and the property that @xmath45 can be seen as a subsemiring of @xmath198 , we obtain easily the following results .
[ l_balance ] let @xmath45 be a naturally ordered semiring with a symmetry and without zero - divisors .
then @xmath198 satisfies property [ pty_order ] if and only if the following holds for all @xmath703 @xmath704 moreover , @xmath198 satisfies property [ pty_order - equal ] if and only if the following holds for all @xmath703 @xmath705 also @xmath198 satisfies property [ pty_order - finite ] if and only if any totally ordered subset of @xmath110 is finite . [ l3.6 ]
let @xmath45 be as in lemma [ l_balance ]
. then @xmath198 satisfies property [ pty_inv ] if and only if for all @xmath706 there exists @xmath707 such that for all @xmath708 @xmath709 the conditions of lemma [ l_balance ] are easily satisfied when @xmath636 and @xmath637 .
for instance , for the first condition of lemma [ l_balance ] one can take @xmath710 if @xmath711 , @xmath712 if @xmath713 and @xmath714 , and any @xmath715 otherwise .
then by lemma [ l_balance ] , @xmath716 and @xmath717 satisfy properties [ pty_order ] [ pty_order - equal ] . the same holds for @xmath286 and @xmath301 as in example [ ex - viro ] ( the phase extension of the tropical semiring ) .
the condition of lemma [ l3.6 ] holds as soon as @xmath481 is the set of invertible elements of @xmath45 , since then @xmath718 is a solution ( @xmath719 is an ideal ) .
this is the case for @xmath636 , @xmath637 , and for the semiring @xmath286 of example [ ex - viro ] .
then by lemma [ l3.6 ] the semirings @xmath716 , @xmath717 , and @xmath301 satisfy property [ pty_inv ] .
note that similarly property [ pty_inv ] holds as soon as @xmath720 , or at least the set of elements @xmath502 of @xmath685 such that @xmath699 is invertible in @xmath138 , is the set of invertible elements of @xmath125 .
the above properties also hold for each of the extensions @xmath274 of a group @xmath264 defined in examples [ groupext ] and [ ex - supertropical ] , ( with a nontrivial symmetry in the first case and the identity symmetry in the second one ) . thus for each of these examples
, @xmath279 satisfies properties [ pty_order][pty_inv ] .
this is in particular the case for the tropical extension of the torus @xmath284 of example [ groupext ] and the supertropical semifield of example [ ex - supertropical ] .
we can also prove directly that properties [ pty_order][pty_inv ] hold for any supertropical semifield .
these examples can be summarized as follows .
all the following semirings allow the convergence of monotone algorithms : the bi - valued tropical semiring @xmath336 , the symmetrized max - plus semiring @xmath254 , the phase extension of the tropical semiring @xmath301 ( example [ ex - viro ] ) , the tropical extension of the torus @xmath284 or that of any group with a non trivial symmetry @xmath279 ( example [ groupext ] ) , and any supertropical semifield ( see remark [ rem - supertropical ] ) .
we also have : [ prop - equal ] let @xmath125 satisfy properties [ pty_order ] and [ pty_order - equal ] .
then for all @xmath693 we have : @xmath721 let @xmath693 be such that @xmath722 , @xmath723 and @xmath724 .
then by property [ pty_order ] applied to @xmath173 and @xmath41 , there exists @xmath725 , such that @xmath726 , @xmath727 and @xmath728 . then @xmath729 and @xmath730 . by property [ pty_order - equal ]
we get that @xmath731 .
since @xmath732 , we deduce that @xmath480 .
the following result shows that the existence part of theorem [ th - cramer ] does not require the condition that all the cramer determinants are thin .
[ th - jac ] let @xmath125 be a semiring allowing the convergence of monotone algorithms , see definition [ monotone - def ] , let @xmath733 , and assume that @xmath734 is invertible in @xmath138 ( but possibly @xmath735 ) .
then for every @xmath736 there exists a thin solution @xmath41 of @xmath42 , which can be chosen in such a way that @xmath737 .
this result will be proved in section [ subsec - stationn ] as a corollary of theorem [ theo - jacobi ] below , which builds the solution using a jacobi - type algorithm .
applying theorem [ th - jac ] to the tropical extension @xmath254 ( in which @xmath738 is invertible if and only if @xmath739 or equivalently @xmath740 ) , we recover the statement of ( * ? ? ?
[ th - jac - smax ] let @xmath741 .
assume that @xmath742 ( but possibly @xmath735 ) .
then for every @xmath743 there exists a thin solution @xmath41 of @xmath42 , which can be chosen in such a way that @xmath737 . a sketch of the proof of the latter result appeared in @xcite ; the complete proof appeared in @xcite . the proof of theorem [ th - jac ] that we next give generalizes the former proof to the present setting .
we also derive as a corollary the following analogous result in the bi - valued tropical semiring @xmath336 .
[ new - th ] let @xmath744 , and assume that @xmath745 ( but @xmath746 may be balanced ) . then for every @xmath747 the thin vector ( `` cramer solution '' ) @xmath748 satisfies @xmath42 .
an element of @xmath336 is such that @xmath738 is invertible if and only if @xmath739 or equivalently @xmath740 .
it is easy to see that an element @xmath749 is thin if and only if @xmath750 .
hence , the only possible thin vector @xmath41 such that @xmath737 is given by @xmath751 .
applying theorem [ th - jac ] to the tropical extension @xmath336 we get the corollary .
the previous results can be reformulated equivalently in the following homogeneous forms .
[ th - jac - homogeneous ] let @xmath125 be a semiring allowing the convergence of monotone algorithms , see definition [ monotone - def ] , and @xmath752 .
let @xmath753 be such that @xmath754 for all @xmath667 $ ] .
assume that either @xmath755 , or at least one entry of @xmath756 is invertible in @xmath138 .
then there exists a thin solution @xmath41 of @xmath37 such that @xmath757 .
let @xmath668 be as in the statement of the theorem and let us show that there exists @xmath758 , such that @xmath37 and @xmath757 .
if all the entries of @xmath668 are @xmath173 , then @xmath589 satisfies trivially the above conditions .
hence , assume without loss of generality that @xmath759 is invertible in @xmath138 .
we set @xmath671 and @xmath760 . by applying theorem [ th - jac ] to the system @xmath677
, we get a thin solution @xmath761 of @xmath677 such that @xmath762 for all @xmath18 $ ] .
let @xmath763 be such that @xmath764 and @xmath765 for @xmath18 $ ] .
then @xmath766 for all @xmath682 $ ] and multiplying the equation @xmath677 by @xmath767 , we get that @xmath670 .
note that when @xmath138 is a semifield , @xmath668 always satisfies the condition of theorem [ th - jac - homogeneous ] .
the uniqueness of the solution is obtained from corollary [ cor - cramer ] . in the particular case of the bi - valued tropical semiring @xmath336
we obtain more precise result .
[ th - new - homogeneous ] let @xmath768 .
let @xmath769 be such that @xmath770 for all @xmath667 $ ] .
then the thin vector @xmath771 satisfies @xmath37 . moreover , if @xmath668 is thin and non - zero , then for any thin vector @xmath41 which is a solution of @xmath37 , we have that @xmath41 is a thin multiple of @xmath668 .
an element @xmath749 is thin if and only if @xmath750 and the set @xmath772 is exactly the set of invertible elements of @xmath336 .
then applying theorem [ th - jac - homogeneous ] to @xmath336 , we get the first assertion .
the second assertion follows similarly from corollary [ cor - cramer ] .
the special case of the existence result , theorem [ th - jac ] , concerning @xmath773 or @xmath25 could be derived alternatively from the existence and uniqueness result in the cramer theorem , theorem [ th - cramer ] . to do
this a perturbation argument can be used since the matrix @xmath38 and the vector @xmath40 can always be `` approximated '' by matrices satisfying the condition of item ( 2 ) of the latter theorem .
the thin solution @xmath41 in theorem [ th - jac ] will be established constructively by means of the jacobi algorithm of @xcite .
the following notion of diagonal dominance is inspired by the notion with the same name which is classically used in numerical analysis .
a real matrix @xmath774 is classically said to have a dominant diagonal if @xmath775 holds for all @xmath776 $ ] .
the tropical analogue of this condition is @xmath777 \enspace .\label{e - tropdiag}\end{aligned}\ ] ] we shall use the following related condition .
[ def - domdiag ] we shall say that @xmath778 has a _
dominant diagonal _ if @xmath779 corollary [ cor - but ] implies that , when @xmath138 is a semifield , any matrix which has a dominant diagonal in this sense is diagonally similar to a matrix satisfying the tropical analogue of the usual condition of diagonal dominance , in particular , definition [ def - domdiag ] is weaker than . the following decomposition is similar to the one used in the classical ( relaxed ) jacobi algorithm .
[ exist - jacobi - dec ] let @xmath125 be a semiring satisfying property [ pty_order ] .
then any matrix @xmath780 with a dominant diagonal can be decomposed into the sum @xmath781 of matrices @xmath373 and @xmath782 such that @xmath373 is a diagonal matrix with diagonal entries in @xmath685 , and @xmath783 .
the latter decomposition will be called a _
jacobi - decomposition _ of @xmath38 .
let @xmath778 be a matrix with a dominant diagonal .
let @xmath18 $ ]
. from property [ pty_order ] applied to @xmath784 and @xmath785 , there exists @xmath786 such that @xmath787 and @xmath788 . since @xmath789 , there exists @xmath790 such that @xmath791 . taking for @xmath373 the diagonal matrix such that @xmath792 and for @xmath793 the matrix such that @xmath794 and @xmath795 for all @xmath375 , we get that @xmath796 .
moreover , @xmath797 for all @xmath18 $ ] , so latexmath:[$|\det d|= we obtain @xmath783 .
[ rem - dom ] we may always reduce the problem @xmath799 to the case where @xmath800 . indeed , since @xmath801 is a morphism and @xmath802 .
we have @xmath803 for all @xmath804 .
since @xmath138 is a totally ordered idempotent semiring , computing the permanent of the matrix @xmath805 is equivalent to solving an optimal assignment problem , which furnishes an optimal permutation . permuting the rows of the matrix @xmath38
, we can transform the system into a system @xmath806 such that the optimal permutation of @xmath807 is the identity , which implies @xmath808 .
since @xmath809 , then @xmath807 has a dominant diagonal as soon as @xmath734 is invertible in @xmath138 .
moreover , since @xmath228 is a morphism , we get that @xmath810 and @xmath811 . using lemmas [ lhabitude ] and [ llem - util07 ]
, we deduce that proving theorem [ th - jac ] for @xmath799 is equivalent to proving it for @xmath806 .
[ theo - jacobi ] let @xmath125 be a semiring allowing the construction of monotone algorithms , see definition [ monotone - def ] .
let @xmath780 have a dominant diagonal , and let @xmath796 be a jacobi - decomposition
. then 1 .
[ jacobi-1 ] one can construct a sequence @xmath812 of thin vectors satisfying : 1 .
[ jacobi - i]@xmath813 ; 2 .
[ jacobi - ii]@xmath814 .
[ jacobi-2 ] the sequence @xmath815 is stationary after at most @xmath5 iterations , meaning that @xmath816 for all @xmath817 , and we have @xmath818 3 . [ jacobi-4 ] when @xmath125 allows the convergence of monotone algorithms ( definition [ monotone - def ] ) , the sequence @xmath819 is stationary , meaning that there exists @xmath820 such that @xmath821 for all @xmath822 .
moreover , the limit @xmath823 is a solution of @xmath42 .
[ jacobi-5 ] when @xmath125 satisfies also property [ pty_order - equal ] , one can take @xmath824 in the previous assertion . applying the previous result to the tropical extension @xmath254 ( in which @xmath738 is invertible if and only if @xmath739 or equivalently @xmath740 ) , we recover ( * ? ? ?
applying the same result to the case of the bi - valued tropical semiring @xmath717 , and using the same arguments as for corollary [ new - th ] of theorem [ th - jac ] , we obtain immediately the following result .
[ theo - jacobinew ] let @xmath825 have a dominant diagonal and let @xmath796 be a jacobi - decomposition .
then the sequence @xmath812 of thin vectors defined by @xmath826 , @xmath827 is stationary after at most @xmath5 iterations , meaning that @xmath828 .
moreover , @xmath829 is a thin solution of @xmath42 and @xmath830 before proving theorem [ theo - jacobi ] , we illustrate it by an example .
[ ex - jacobi ] we take @xmath831 , and apply the tropical jacobi algorithm to the linear system @xmath832 denoting by @xmath38 the matrix of the system and by @xmath40 its right - hand side , we get : @xmath833 let us choose the jacobi - decomposition @xmath796 with : @xmath834 applying the jacobi algorithm , we get the following sequence starting from @xmath826 : @xmath835 @xmath836 @xmath837 choosing @xmath838 yields another solution : @xmath839 ( where @xmath840 denotes the transpose of @xmath41 ) , whereas taking the opposite sign in @xmath841 and @xmath842 would have lead to the other possible solutions @xmath843 , and @xmath844 . for the proof of theorem [ theo - jacobi ] we shall need the following lemmas which are derived from the results of section [ sec - cramer ] and from theorem [ th - yoeli ] .
[ lem - dom ] let @xmath733 be a matrix with a dominant diagonal and let @xmath796 be a jacobi - decomposition .
then @xmath845 is an invertible diagonal matrix which coincides with the diagonal submatrix of @xmath805 : @xmath797 for all @xmath18 $ ] .
since @xmath796 is a jacobi - decomposition and @xmath228 is a morphism , we have @xmath846 .
this implies that @xmath847 is invertible in @xmath138 , since @xmath38 has a dominant diagonal .
@xmath845 is a diagonal matrix .
thus it is necessarily a monomial matrix with entries in @xmath138 . from @xmath848 and
the property that @xmath847 is invertible in @xmath138 , we get that all diagonal entries @xmath849 of @xmath845 are invertible in @xmath138 .
thus @xmath845 is invertible in @xmath850 .
from @xmath851 we deduce that @xmath852 for all @xmath18 $ ] .
let us show the reverse inequalities , which will imply the equalities .
since @xmath38 has a dominant diagonal , we have @xmath853 are invertible in @xmath138 .
we have for instance @xmath854 .
since @xmath855 are invertible , we deduce that @xmath856 .
the same argument shows that @xmath857 holds for all @xmath18 $ ] .
[ lem - sec - cir ] let @xmath733 be a matrix with a dominant diagonal and let @xmath796 be a jacobi - decomposition . then @xmath858 .
hence every circuit of @xmath859 has a weight less than or equal to @xmath65 . by lemma [ lem - dom ]
, @xmath845 is invertible .
let us show that @xmath860 . since @xmath845 is a monomial matrix of @xmath850 , we have @xmath861 . since @xmath862 , we obtain that @xmath860 .
let us denote @xmath863 .
since @xmath864 , we get that @xmath865 for the natural order of @xmath138 .
this implies that @xmath866 and that the weight of a circuit for @xmath867 is less than or equal to its weight for @xmath868 . from theorem
[ th - yoeli ] we get that the weight of a circuit of @xmath868 is less than or equal to @xmath65 , which implies the same for @xmath867 .
[ lem - meta - yoeli ] let @xmath733 be a matrix with a dominant diagonal and let @xmath796 be a jacobi - decomposition
. then @xmath869 .
let @xmath863 and @xmath870 . since @xmath228 is a morphism , we have easily @xmath871 . since @xmath796
, we also have @xmath872
. then @xmath873 and it remains to show that @xmath874 .
by lemma [ lem - sec - cir ] @xmath875 . since by lemma [ lem - dom ] @xmath845 is the diagonal of @xmath805 , the diagonal entries of @xmath876 are equal to @xmath65 .
thus @xmath876 satisfies the conditions of theorem [ th - yoeli ] .
hence @xmath877 .
since @xmath845 is a monomial matrix , lemmas [ llem - util07 ] and [ lhabitude ] imply that @xmath878 , thus @xmath879 , which finishes the proof .
we first prove item . since @xmath796 is a jacobi - decomposition
, @xmath373 is a diagonal matrix with entries in @xmath685 . by lemma [ lem - dom ]
, @xmath845 is an invertible diagonal matrix , which implies that all its diagonal entries are invertible in @xmath138 .
hence , one can construct from property [ pty_inv ] , the diagonal @xmath39 matrix @xmath880 such that @xmath881 .
then condition of theorem [ theo - jacobi ] is equivalent to : @xmath882 let us show by induction on @xmath413 that there exist thin @xmath5-dimensional vectors @xmath883 satisfying , together with @xmath884 when @xmath885 , the above conditions are equivalent to @xmath886 , @xmath887 and @xmath888 . thus @xmath889 can be constructed by applying property [ pty_order ] entrywise .
hence , the properties of the induction hold for @xmath885 .
assume now that these properties hold for some @xmath890 .
since the map @xmath891 is non - decreasing and @xmath892 , we have from that @xmath893 . then applying property [ pty_order ] entrywise
, we construct @xmath894 such that @xmath895 and @xmath896 .
this shows that the induction hypothesis holds for @xmath897 .
thus item of the theorem is proven .
we next prove item .
denote @xmath898 .
since @xmath228 is a morphism , condition of theorem [ theo - jacobi ] implies that @xmath899 .
since @xmath845 is invertible in @xmath850 , we obtain that @xmath900 for @xmath863 .
hence , for all @xmath890 we get that @xmath901 lemma [ lem - sec - cir ] allows us to apply the theorem of carr and gondran ( proposition - definition [ propdef ] ) to the matrix @xmath867 .
it follows that @xmath902 the last assertion of item follows from the previous equation and lemma [ lem - meta - yoeli ] .
we now prove item . since all @xmath903 are thin , @xmath816 for all @xmath817 , and @xmath904 .
applying property [ pty_order - finite ] entrywise implies that @xmath905 is stationary after some finite time @xmath906 .
then by condition of the theorem we get that @xmath907 , which is equivalent to @xmath908 .
we finally prove item .
when @xmath125 satisfies also property [ pty_order - equal ] , the above properties imply that @xmath909 .
thus one can take @xmath824 in the previous conclusions . applying theorem [ theo - jacobi ] after the transformation described in remark [ rem - dom ] and using proposition [ exist - jacobi - dec ] , we deduce theorem [ th - jac ] . we now introduce a gauss - seidel type algorithm .
it is a variant of the jacobi algorithm in which the information is propagated more quickly .
[ exist - gauss - seidel - dec ] let @xmath125 be a semiring satisfying property [ pty_order ] .
then any matrix @xmath780 with a dominant diagonal can be decomposed into the sum @xmath910 of matrices @xmath373 , @xmath911 and @xmath912 such that @xmath373 is a diagonal matrix with diagonal entries in @xmath685 , @xmath783 , all the entries on the main diagonal and above the main diagonal of @xmath911 equal to @xmath173 , and all the entries below the main diagonal of @xmath913 equal to @xmath173
. the latter decomposition will be called a _
gauss - seidel - decomposition _ of @xmath38 . by proposition
[ exist - jacobi - dec ] there exists a jacobi - decomposition @xmath796 . taking for @xmath911 the ( strict ) lower diagonal part of @xmath793 and for @xmath913 the upper diagonal and diagonal part of @xmath793
, we get the result .
[ theo - gauss - seidel ] let @xmath125 be a semiring allowing the construction of monotone algorithms , see definition [ monotone - def ] . let @xmath780 have a dominant diagonal and let @xmath914 be a gauss - seidel - decomposition
. then 1 .
one can construct a sequence @xmath812 of thin vectors satisfying : 1 .
[ gauss11 ] @xmath915 ; 2 .
[ gauss12]@xmath916 ; 3 .
[ gauss13]@xmath917 .
[ gauss2]the sequence @xmath815 is stationary after at most @xmath5 iterations , meaning that @xmath816 for all @xmath817 , and we have @xmath918 3 . [ gauss3 ] when @xmath125 allows the convergence of monotone algorithms ( definition [ monotone - def ] ) , the sequence @xmath819 is stationary , meaning that there exists @xmath820 such that @xmath919 .
moreover , the limit @xmath823 is a solution of @xmath42 .
[ gauss4 ] when @xmath125 satisfies also property [ pty_order - equal ] one can choose @xmath824 in the previous assertion .
we consider the system of example [ ex - jacobi ] , with the gauss - seidel decomposition @xmath914 such that @xmath920 with @xmath373 and @xmath793 as in .
then starting from @xmath826 we obtain the following sequence : @xmath921 @xmath922 we find the solution @xmath923 after 2 iterations only , whereas the jacobi algorithm required 3 iterations .
[ lem - star - u ] let @xmath733 be a matrix with a dominant diagonal and let @xmath914 be a gauss - seidel - decomposition .
let @xmath924 .
then @xmath925 has no circuit ( all circuits have a zero weight ) and we have @xmath926 . since the entries of @xmath911 are @xmath173 on and above the diagonal , the graph of @xmath925 has no circuit .
. then @xmath928 .
taking @xmath929 and @xmath930 , we get a jacobi - decomposition @xmath931 of @xmath807 . applying lemma [ lem - meta - yoeli ] to it ,
we get the last assertion of the lemma .
[ lem-2-nonpos ] let @xmath733 be a matrix with a dominant diagonal .
assume @xmath914 is a gauss - seidel - decomposition .
let @xmath924 .
then * every circuit of @xmath932 has a weight less than or equal to @xmath174 .
* @xmath933 .
we start by proving item ( b ) . indeed , adding a top element to @xmath138 allows one to define the kleene star @xmath934 of a matrix @xmath387 with entries in @xmath138 .
then every circuit of @xmath387 has a weight less than or equal to @xmath174 if and only if @xmath934 has all its entries in @xmath138 ( meaning that they are all different from the top element ) .
let @xmath935 .
it is easy to see that @xmath936 is a jacobi - decomposition of @xmath38 .
thus by lemma [ lem - sec - cir ] , every circuit of @xmath859 has a weight less than or equal to @xmath174 . by lemma [ lem - meta - yoeli ] , we have @xmath937 .
denote @xmath938 and @xmath939 . then by lemma [ lem - star - u ]
, we have @xmath940 , so @xmath941 , and @xmath942 .
this implies that the assertion of item ( b ) is equivalent to the equality @xmath943 which is indeed a well known
_ unambiguous rational identity _
( by expanding the kleene star and products in both expressions , we arrive at the sum of all words in the letters @xmath944 and @xmath945 ) .
\(a ) let us note that the identity also shows that @xmath946 and since every circuit of @xmath859 has a weight less than or equal to @xmath174 , the latter expression has all its entries in @xmath138 , so has @xmath947 .
this implies that every circuit of @xmath932 has a weight less than or equal to @xmath174 .
the proof follows the same lines as the one of theorem [ theo - jacobi ] .
in particular , constructing the same matrix @xmath880 as in this proof , we get that condition of theorem [ theo - gauss - seidel ] is equivalent to @xmath948 , which corresponds to the system : @xmath949 using property [ pty_order ] for each @xmath413 and each @xmath776 $ ] one chooses @xmath950 such that it satisfies the @xmath392th equation of this system , together with the conditions @xmath951 and @xmath952 then the sequence satisfies conditions and of the theorem . in particular
it satisfies @xmath953 . since @xmath925 has no circuit , the theorem of carr and gondran ( proposition - definition [ propdef ] ) implies that @xmath954 , which by lemma [ lem - star - u ] is equivalent to condition of the theorem .
again the latter condition implies that @xmath955 with @xmath867 as in lemma [ lem - star - u ] .
then by lemma [ lem-2-nonpos ] and the theorem of carr and gondran , we get that @xmath956 for all @xmath817 .
moreover from the second assertion of lemma [ lem-2-nonpos ] , we have @xmath957 which shows item of the theorem . items and of the theorem are obtained by the same arguments as for theorem [ theo - jacobi ] .
the particular case of theorem [ theo - gauss - seidel ] concerning the tropical extension @xmath254 was obtained in @xcite .
one can also apply the same result to the case of the bi - valued tropical semiring @xmath717 .
this leads to the same solution @xmath751 as in corollary [ theo - jacobinew ] , by using the gauss - seidel algorithm instead of the jacobi algorithm .
the following result was stated in @xcite .
[ tgma ] let @xmath958 .
then there exists @xmath959 such that @xmath960 if and only if @xmath961 . the special case in which @xmath962 is equivalent to the theorem of gondran and minoux quoted in the introduction ( theorem [ th-4 ] ) .
the `` only if '' part is obtained by taking @xmath963 in the first assertion of theorem [ th - cramer ] .
the `` if '' part was proved in @xcite .
an analogous result was proved by izhakian and rowen @xcite , when the symmetrized tropical semiring @xmath25 is replaced by the bi - valued tropical semiring @xmath336 .
we next provide a general result which includes theorem [ tgma ] as well as the result of @xcite as special cases .
let @xmath125 be a semiring with a symmetry , a thin set @xmath685 , and a modulus taking its values in a totally ordered semiring @xmath138 .
we shall need the following additional properties .
[ p - temp-0 ] @xmath138 is an idempotent semifield and the thin set @xmath685 is such that the set of invertible elements of @xmath125 is @xmath720 and that it coincides with @xmath964 .
this property is satisfied when @xmath686 , with the thin set @xmath687 , @xmath138 is an idempotent semifield and @xmath965 is the set of invertible elements .
[ p - temp-5 ] for all @xmath966 we have @xmath967 .
this property is satisfied when @xmath686 .
[ p - temp-2 ] for all @xmath968 such that @xmath969 , either there exists a single index @xmath970 $ ] such that @xmath971 and @xmath972 , or there exist two different indices @xmath973 $ ] such that @xmath974 and @xmath975 .
this property is satisfied when @xmath686 , with the thin set @xmath687 , and for all @xmath976 such that @xmath977 , either there exists a single index @xmath970 $ ] such that @xmath978 , or there exist two different indices @xmath973 $ ] such that @xmath979 .
[ p - temp-3 ] if @xmath980 and @xmath981 then @xmath982 .
this property is satisfied when property [ p - temp-0 ] and the result of proposition [ prop - equal ] hold .
we can then obtain the following assertion .
all the following semirings satisfy properties [ p - temp-0][p - temp-3 ] : the symmetrized max - plus semiring @xmath254 , the bi - valued tropical semiring @xmath336 , the tropical extension of the torus @xmath284 or that of any group with a non trivial symmetry @xmath279 ( example [ groupext ] ) , and any supertropical semifield ( see remark [ rem - supertropical ] ) .
however the phase extension of the tropical semiring @xmath301 ( example [ ex - viro ] ) does not satisfy properties [ p - temp-0 ] , [ p - temp-2 ] , nor [ p - temp-3 ] .
we note the following consequences , the first one being easy .
[ p - temp-1 ] let @xmath125 satisfy property [ p - temp-0 ] .
for all @xmath693 if @xmath983 then @xmath980 or @xmath984 . [ p - temp-4 ]
let @xmath125 satisfy properties [ p - temp-0 ] and [ p - temp-3 ] .
if @xmath985 where @xmath986 and @xmath987 , then @xmath988 or @xmath989 .
we have @xmath990 by property [ p - temp-3 ] . by proposition [ p - temp-1 ]
we must have @xmath991 or @xmath992 .
[ newtgma ] let @xmath125 be a semiring allowing weak balance elimination ( definition [ def - tropextension ] ) and the convergence of monotone algorithms ( definition [ monotone - def ] ) , and satisfying properties [ p - temp-0][p - temp-3 ] .
let @xmath993 .
then there exists @xmath994 such that @xmath960 if and only if @xmath961 .
the necessity of the condition @xmath961 follows by taking @xmath963 in the first assertion of theorem [ th - cramer ] .
indeed , @xmath37 and @xmath995 imply @xmath996 .
at least one entry of @xmath41 belongs to @xmath720 . by property [ p - temp-0 ] , this entry is invertible and so @xmath997 .
let us prove that @xmath997 implies the existence of @xmath994 such that @xmath960 .
_ case 1 _ : we first deal with the degenerate case in which @xmath998 .
then @xmath999 , so applying the frobenius - knig s theorem to @xmath805 , and using property [ p - temp-5 ] , there exists a reordering of rows and columns such that the matrix @xmath38 has the following form : @xmath1000 where @xmath1001 , @xmath1002 , and @xmath1003 denotes the @xmath1004 zero matrix .
it suffices to look for a solution @xmath41 such that @xmath1005 for all @xmath1006 . denoting by @xmath610 the vector with entries @xmath1007 , it remains to solve the system @xmath1008 which has @xmath1009 unknowns and @xmath1010 equations .
thus if this system has a non - zero @xmath1011-minor , an application of theorem [ th - jac - homogeneous ] provides a non - zero thin solution of the system @xmath1008 .
otherwise , we may assume by induction that the sufficiency in the theorem ( or at least its restriction to the case @xmath1012 ) is already proved for systems of lower dimension .
then we apply the induction to a square subsystem @xmath1013 obtained by setting to zero one coordinate of @xmath610 .
this completes the treatment of the degenerate case . _
case 2 _ : we now assume that @xmath1014 , so @xmath1015 . by corollary [ cor - but ]
applied to the matrix @xmath1016 , there exist two diagonal matrices @xmath373 and @xmath473 with invertible diagonal entries in @xmath138 and a permutation matrix @xmath474 , such that @xmath1017 satisfies @xmath1018 and @xmath1019 for all @xmath436 $ ] . then applying the injection @xmath233 to the matrices @xmath373 , @xmath473 and @xmath474 , and using the fact that @xmath233 is a multiplicative morphism , we obtain a matrix @xmath1020 such that @xmath1021 and so satisfies the above properties .
since a diagonal scaling of @xmath38 does not change the balanced character of the determinant , nor the existence of a thin solution of @xmath37 , we may always assume that @xmath1022 .
thus @xmath38 satisfies : latexmath:[\[\label{anorm } _ subcase 2.1 _ : we shall first consider the subcase in which there is a permutation @xmath224 such that @xmath1024 } a_{i\sigma(i ) } { { \,\nabla\ , } } { \mathbbold{0}}\text { and } { \operatorname{per}}|a|= |\det a| = |\bigodot_{i\in [ n ] } a_{i\sigma(i)}| = \prod_{i\in [ n ] } |a_{i\sigma(i)}| \enspace .\end{aligned}\ ] ] assume , possibly after permuting the rows of @xmath38 , that @xmath224 is the identity permutation ( this does not change property ) . since @xmath1025 }
a_{ii } { { \,\nabla\ , } } { \mathbbold{0}}$ ] , by proposition [ p - temp-1 ] , we must have @xmath1026 for some @xmath1027 $ ] and we may always assume that @xmath1028 .
then @xmath38 can be written in block form as @xmath1029 we set @xmath1030 and define @xmath1031 to be a thin solution of @xmath1032 provided by theorem [ th - jac ] .
thus @xmath1033 , and so , @xmath1034 for all @xmath393 .
since @xmath1035 and @xmath1036 , it follows from property [ p - temp-3 ] and that @xmath1037 . hence , @xmath1038 is a non - zero thin solution of @xmath37 .
_ subcase 2.2 _ : it remains to consider the subcase in which no permutation @xmath224 satisfies .
since @xmath997 , by property [ p - temp-2 ] , there must exist two distinct permutations @xmath224 and @xmath287 such that @xmath1039 } a_{i\sigma(i ) } | = |\bigodot_{i\in [ n ] } a_{i\pi(i ) } | \enspace,\ ] ] and @xmath1040 } a_{i\sigma(i ) } \oplus
\operatorname{{\operatorname{sgn}}}(\pi)\bigodot_{i\in [ n ] } a_{i\pi(i ) } { { \,\nabla\ , } } { \mathbbold{0}}.\ ] ] we may always assume , possibly after permuting the rows of @xmath38 , that @xmath224 is the identity permutation ( this does not change property ) . moreover , since the identity permutation @xmath224
does not satisfy , but satisfies @xmath1041 } a_{ii } |=|\det a|$ ] , we deduce that @xmath1042 } a_{ii } \not\in \tt^\circ$ ] .
hence by property [ p - temp-0 ] all @xmath1043 are invertible . multiplying the system by the inverse of the diagonal submatrix of @xmath38 , we get a new matrix @xmath38 with all the above properties and such that @xmath1044 for all @xmath18 $ ] .
thus @xmath1045 } a_{i\pi(i ) } { { \,\nabla\ , } } { \mathbbold{0}}.\ ] ] let us decompose @xmath287 as a product of disjoint cycles @xmath1046 , with supports @xmath1047 of cardinalities @xmath1048 , respectively
. then @xmath1049 } ( \ominus { \mathbbold{1}})^{p_m-1 } \bigodot_{i\in i_m } a_{i , c^m(i ) } { { \,\nabla\ , } } { \mathbbold{0}}.\ ] ] it follows from and proposition [ p - temp-4 ] that there exists a cycle @xmath1050 such that @xmath1051 we may assume , without loss of generality , that @xmath1052 , with @xmath1053
. then we define inductively the entries @xmath1054 of the vector @xmath1055 by @xmath1056 since the permutation @xmath287 does not satisfy , but satisfies @xmath1041 } a_{i\pi(i ) } |=|\det a|$ ] , the entries @xmath1057 , , @xmath1058 are all invertible in @xmath125 . hence , the former relations define the vector @xmath1059 uniquely . actually , @xmath1060 and
we observe that @xmath1061 for all @xmath1062 $ ] .
moreover , from and we deduce that @xmath1063 let @xmath264 denote the @xmath1064 top - left submatrix of @xmath38 .
it follows from formulas and and from property [ p - temp-3 ] that @xmath1065 holds for all @xmath1062 $ ] .
let us now write @xmath38 in the block form @xmath1066 where @xmath1067 and @xmath1068 are of sizes @xmath1069 and @xmath1070 , respectively , with @xmath1071 , and look for a solution of the form @xmath1072 .
then we may choose for @xmath1073 a thin vector solution of @xmath1074 given by theorem [ th - jac ] .
thus @xmath1075 .
hence , @xmath1034 for all @xmath1076 $ ] .
it follows from property [ p - temp-3 ] that @xmath1077 .
[ cor - dominant - unique ] let @xmath125 be a semiring allowing strong balance elimination ( definition [ def - tropextension ] ) and the convergence of monotone algorithms ( definition [ monotone - def ] ) , and satisfying properties [ p - temp-0][p - temp-3 ] .
let @xmath993 such that @xmath656 is invertible and let @xmath736 .
then the system @xmath42 has a unique thin solution if and only if @xmath655 is thin .
the sufficient condition follows from the second assertion of theorem [ th - cramer ] . for the necessary condition ,
let us assume that @xmath655 is not thin , and prove that there exist at least two different solutions . since @xmath655 is not thin , there exists @xmath18 $ ] such that @xmath1078 .
assume without loss of generality that @xmath1079 . from theorem [ th - jac ] , since @xmath745
, there exists a solution @xmath609 of @xmath42 such that @xmath1080 , thus @xmath1081 hence @xmath1082 .
now let us construct a solution @xmath1083 such that @xmath1084 .
let @xmath387 be the block matrix @xmath1085 .
we have @xmath1086 , so by theorem [ newtgma ] , there exists @xmath1087 such that @xmath1088 .
let us show that @xmath1089 .
indeed , if @xmath1090 then @xmath610 yields a thin solution of @xmath43 , and since @xmath656 is invertible , this implies that @xmath592 , a contradiction .
so @xmath1089 and since @xmath1091 , @xmath1092 is invertible .
so multiplying @xmath610 by @xmath1093 we obtain a solution of @xmath1088 such that @xmath1094 .
then the vector @xmath1095 is a solution of @xmath1096 such that @xmath1084 .
this shows that the system @xmath1096 has at least two non - zero solutions .
proposition 8.8 of @xcite gives a @xmath1097 matrix with entries in @xmath0 , such that there is no signed no - zero row vector @xmath41 such that @xmath1098 , but all maximal determinants taken from @xmath38 are balanced .
this shows that theorem [ newtgma ] can not be extended to the rectangular case .
we now give a geometrical interpretation of the previous results .
we first consider , as in section [ sec - elim ] , a semiring @xmath45 with symmetry and a thin set @xmath1099 .
we call _ hyperplane _ of @xmath1100 a set of the form @xmath1101 } { a}_i x_i { \,\nabla\ , } { \mathbbold{0}}\}\end{aligned}\ ] ] where @xmath1102 is a non - zero vector .
[ ex - htrop ] when @xmath45 is the bi - valued tropical semiring @xmath1103 , @xmath110 can be identified to @xmath0 , then @xmath1104 coincides with the tropical hyperplane .
[ ex - hsign ] assume now that @xmath45 is the symmetrized tropical semiring @xmath25 , so that @xmath1105 , with @xmath1106 .
identifying @xmath1107 with @xmath0 , and setting @xmath1108\mid { a}_i \in { \mathbb{s}_{\max}}^\oplus\}$ ] and @xmath1109\setminus i$ ] , it is readily seen that @xmath1110 is the signed tropical hyperplane defined in .
the following result is a simple consequence of theorem [ th - cramer ] .
we say that @xmath21 vectors @xmath1111 of @xmath1112 are in _ general position _ if every @xmath391 minor of the @xmath1113 matrix @xmath867 with columns @xmath1111 is thin and non - zero ( @xmath1114 ) .
similarly , we say that @xmath21 hyperplanes of @xmath69 are in _ general position _ if the vectors of parameters of these hyperplanes are in general position .
[ th - cramer - geom ] let @xmath45 be a semiring with a thin set @xmath1099 , allowing strong balance elimination ( definition [ def - tropextension ] ) .
assume that @xmath481 is the set of invertible elements of @xmath45 .
then * any @xmath21 vectors of @xmath1115 in general position are contained in a unique hyperplane . *
any @xmath21 hyperplanes of @xmath1115 in general position contain a non - zero vector which is unique up to an invertible scalar multiple .
we prove the primal statement ( the dual statement follows along the same lines ) .
assume that the vectors @xmath1111 are included in the hyperplane @xmath20 of .
then the vector @xmath16 of parameters of this hyperplane , thought of as a row vector , satisfies @xmath1116 where @xmath867 is as above . up to a transposition , and to the replacement of @xmath5 by @xmath44 ,
this system is of the type considered in corollary [ cor - cramer ] , and the conclusion follows from the latter corollary .
it follows from examples [ ex - htrop ] and [ ex - hsign ] that theorems [ th-1 ] and [ th-3 ] stated in the introduction can be re - obtained by specializing the primal form in theorem [ th - cramer - geom ] to @xmath1117 or @xmath1118 .
similarly , theorem [ newtgma ] admits the following geometric interpretation .
the derivation is straightforward .
[ th - hom - geom ]
let @xmath125 be a semiring allowing weak balance elimination ( definition [ def - tropextension ] ) and the convergence of monotone algorithms ( definition [ monotone - def ] ) , and satisfying properties [ p - temp-0][p - temp-3 ] .
then * a collection of @xmath5 vectors @xmath1119 of @xmath1120 is contained in a hyperplane if and only if the determinant of the matrix @xmath1121 is balanced ; * a collection of @xmath5 hyperplanes of @xmath1120 @xmath1122}a^{j}_i x_i { \,\nabla\ , } { \mathbbold{0}}\ } , \qquad j\in[n ] \enspace,\ ] ] contains a non - zero vector if and only if the determinant of the matrix @xmath1123}$ ] is balanced . when @xmath45 or @xmath125 are equal to @xmath25 , the dual statements in theorems [ th - cramer - geom ] and [ th - hom - geom ] turn out to have a geometric interpretation which can be stated elementarily , without introducing the symmetrized tropical semiring .
this interpretation relies on the notion of _ sign - transformation _ of a signed hyperplane .
such a transformation is specified by a sign pattern @xmath1124 , it corresponds , in loose terms , to putting variables @xmath517 such that @xmath1125 on the other side of the equality .
formally , the sign - transformation of pattern @xmath185 transforms the signed - hyperplane @xmath1126 to @xmath1127 figure [ fig - transform ] gives an illustration of this notion .
( 0,0 ) # 1#2#3#4#5 @font ( 9930,11946)(811,-11023 ) ( 5926,164)(0,0)[lb ] ( 10726,-7636)(0,0)[lb ] ( 1651,-10711)(0,0)[lb ] ( 826,-7411)(0,0)[lb ] ( 0,0 ) # 1#2#3#4#5 @font ( 9930,11946)(811,-11023 ) ( 5926,164)(0,0)[lb ] ( 10726,-7636)(0,0)[lb ] ( 1651,-10711)(0,0)[lb ] ( 826,-7411)(0,0)[lb ] ( 0,0 ) # 1#2#3#4#5 @font ( 9930,11946)(811,-11023 ) ( 5926,164)(0,0)[lb ] ( 10726,-7636)(0,0)[lb ] ( 1651,-10711)(0,0)[lb ] ( 826,-7411)(0,0)[lb ] the following theorem follows readily from the dual statement in theorem [ th - cramer - geom ] , it can also be derived from ( * ? ? ?
[ thnew - signt ] given @xmath21 signed tropical hyperplanes @xmath1128 , in general position , there is a unique sign pattern @xmath185 such that the transformed hyperplanes @xmath1129 , meet at a non - zero vector .
moreover , such a vector is unique up to an additive constant .
let @xmath20 be the hyperplane defined by , and let @xmath1130 as in example [ ex - hsign ] .
it is easily seen that a vector @xmath1131 belongs to @xmath20 if and only if the vector @xmath738 belongs to the transformed signed tropical hyperplane @xmath1132 where @xmath185 is the sign vector of @xmath41 .
then the theorem follows from the dual form of theorem [ th - cramer - geom ] .
( 0,0)),title="fig : " ] # 1#2#3#4#5 @font ( 11776,13521)(811,-12598 ) ( 5926,164)(0,0)[lb ] ( 10726,-7636)(0,0)[lb ] ( 1651,-10711)(0,0)[lb ] ( 4426,-12286)(0,0)[lb ] ( 826,-7411)(0,0)[lb ] ( 0,0)),title="fig : " ] # 1#2#3#4#5 @font ( 11776,13521)(811,-12598 ) ( 5926,164)(0,0)[lb ] ( 10726,-7636)(0,0)[lb ] ( 1651,-10711)(0,0)[lb ] ( 4426,-12286)(0,0)[lb ] ( 826,-7411)(0,0)[lb ] the interpretation of the dual form of theorem [ th - hom - geom ] , which could also be derived from the result of ( * ? ?
6.5 ) proved in ( * ? ? ?
* chap . 3 , th .
9.0.1 ) , can be stated as follows .
[ thnew-5 ] given @xmath5 signed hyperplanes @xmath1133 , there exists a sign pattern @xmath185 such that the transformed hyperplanes @xmath1134 contain a common non - zero vector if and only if the matrix having as rows the vectors of parameters of @xmath1133 has a balanced determinant .
argue as in the proof of theorem [ thnew - signt ] .
the present approach via the tropical jacobi algorithms leads to an algorithm to compute all the cramer permanents .
[ compute - cramer ] let @xmath1135 and @xmath1136 .
assume that @xmath1137 .
then the vector @xmath655 , the entries of which are the @xmath5 cramer permanents of the system with matrix @xmath38 and right - hand side @xmath40 , can be computed by solving a single optimal assignment problem , followed by a multiple origins - single destination shortest path problem .
let @xmath224 denote an optimal permutation for the matrix @xmath38 .
after permuting the rows of @xmath38 and @xmath40 , we may assume that this permutation is the identity .
dividing every row of @xmath38 and @xmath40 by @xmath1043 , we may assume that @xmath1044 for all @xmath18 $ ] .
then using yoeli s theorem we get @xmath1138 . computing
the latter vector is equivalent to solving a shortest path problem from all origins to a single destination .
the hungarian algorithm of kuhn runs in time @xmath1139 , where @xmath906 is the number of finite entries of the matrix @xmath38 .
the subsequent shortest path problem can be solved for instance by the ford - bellman algorithm , in time @xmath1140 .
hence , we arrive at the strongly polynomial bound @xmath1141 , for the time needed to compute all cramer permanents .
alternative ( non - strongly polynomial ) optimal assignment algorithms may be used @xcite , leading to incomparable bounds .
richter - gebert , sturmfels , and theobald developed in @xcite a different approach , based on earlier results of sturmfels and zelevinsky @xcite .
it provides an alternative method in which all the cramer determinants are obtained by solving a single transportation problem .
a beauty of this approach is that it directly works with homogeneous coordinates , preserving the symmetry which is broken by the jacobi approach , of an `` affine '' nature .
so , we next revisit the method of @xcite , in order to compare the results obtained in this way with the present ones . in passing
, we shall derive some refinements of results in @xcite concerning the case in which the data are not in general position .
first , we observe that one of the results of @xcite can be recovered as a corollary of the present elimination approach .
let @xmath1142 .
the _ tropical cramer permanent _
@xmath1143 is the permanent of the matrix obtained from @xmath38 by deleting the @xmath413th column .
assume that @xmath1142 is such that at least one of the tropical permanents @xmath1143 is finite .
then the vector @xmath1144 with @xmath1145 is such that in the expression @xmath1146 } a_{\cdot k } x_k\ ] ] the maximum is attained at least twice in every row .
moreover , if all the tropical cramer permanents @xmath1143 are non - singular , then the vector @xmath41 having the latter property is unique up to an additive constant .
this is a special case of theorem [ th - new - homogeneous ] in which the matrix @xmath38 is thin .
corollary 5.4 of @xcite gives an explicit construction of the solution @xmath1147 when the tropical cramer permanents are non - singular .
then it proceeds by showing that the solution of a certain transportation problem is unique . here
the uniqueness of the solution of the tropical equations is obtained by the elimination argument used in the proof of theorem [ th - cramer ] .
we now present the method of @xcite in some detail .
let @xmath1148 denote the transportation polytope consisting of all nonnegative @xmath1149 matrices @xmath1150 such that @xmath1151 } y_{ij } = n , \quad i\in[n-1],\qquad \sum_{i\in[n-1 ] } y_{ij } = n-1 , \quad j\in[n ] \enspace .\end{aligned}\ ] ] for @xmath1152 $ ] define @xmath1153 to be the birkhoff polytope obtained as the convex hull of the @xmath1149 matrices with @xmath1154-entries , representing matchings between @xmath1155 $ ] and @xmath9\setminus \{k\}$ ] .
define the minkowski sum : @xmath1156 } \pi^k_{n-1,n } \enspace .\ ] ] here ( partial ) matchings refer to the complete bipartite graph @xmath1157 with @xmath21 nodes in one class ( corresponding to the rows of @xmath610 ) and @xmath5 nodes in the other class ( corresponding to the columns of @xmath610 ) .
every vertex of @xmath1158 can be written as @xmath1159 where for all @xmath1152 $ ] , @xmath1160 represents a matching between @xmath1155 $ ] and @xmath9\setminus \{k\}$ ] , meaning that the set of edges @xmath1161 constitutes a matching between @xmath1155 $ ] and @xmath9\setminus \{k\}$ ] . more generally , sturmfels and zelevinsky @xcite considered the newton polytope @xmath1162 of the product of maximal minors of any @xmath66 matrix such that @xmath1163 .
the entries of this matrix are thought of as pairwise distinct indeterminates .
the former polytope is obtained by taking @xmath1164 .
they showed that @xmath1165 , see ( * ? ? ?
2.8 ) , but that @xmath1162 is no longer a transportation polytope when @xmath1166 .
they also showed that the vertices of @xmath1158 are in bijective correspondence with combinatorial objects called _ linkage trees_. a linkage tree is a tree with set of nodes @xmath9 $ ] , the edges of which are bijectively labeled by the integers @xmath1167 . given a linkage tree , we associate to every @xmath1152 $ ] the matching between @xmath1155 $ ] and @xmath9\setminus\{k\}$ ] , such that @xmath1168\setminus\{k\}$ ]
is matched to the unique @xmath1169 $ ] labeling the edge adjacent to @xmath393 in the path connecting @xmath393 to @xmath413 in this tree .
let @xmath1170 denote the matrix representing this matching .
then the vertex of @xmath1158 that corresponds to this linkage tree is @xmath1171 .
we now associate to every @xmath1149 matrix @xmath610 a subgraph @xmath1172 of @xmath1157 , consisting of the edges @xmath397 such that @xmath1173 .
if @xmath610 is a vertex of @xmath1158 , then the previous characterization in terms of linkage trees implies that @xmath1172 is a spanning tree of @xmath1157 . to simplify the exposition
, we shall assume first that @xmath1174 , @xmath1169 $ ] , @xmath1168 $ ] .
following the idea of richter - gebert , sturmfels , and theobald @xcite , we consider the _ transportation problem _ @xmath1175 @xmath1176,\ ; j\in[n ] } c_{ij}y_{ij } \ ; ; \qquad { y}=(y_{ij})\in t_{n-1,n } \tag{${\mathcal{p}}$}\end{aligned}\ ] ] and its dual @xmath1177 } u_i)+ ( n-1)(\sum_{j\in[n ] } v_j)\bigr ) \ ; ; \qquad { u}=(u_i)\in \re^{n-1 } , { v}=(v_j)\in \re^{n } , \\
c_{ij}\le u_i+v_j,\quad i\in[n-1],\ j\in[n ] \qquad \tag{${\mathcal{d}}$ } \enspace .\end{aligned}\ ] ] the values of these problems will be denoted by @xmath1178 and @xmath1179 , respectively .
recall the _ complementary slackness condition _ : a primal feasible solution @xmath610 and a dual feasible solution @xmath1180 are both optimal if and only if @xmath1181 holds for all @xmath1169 $ ] and @xmath1168 $ ] .
the case in which @xmath1182 holds for some @xmath471 can be dealt with by adopting the convention that @xmath1183 in the expression of the primal objective function .
equivalently , it can be considered as an ordinary linear programming problem by adding the constraints @xmath1184 for all @xmath397 such that @xmath1182 , and ignoring all @xmath397 such that @xmath1182 in the formulation of the dual problem and in the complementary slackness conditions .
we now give a slight refinement of a result of @xcite , allowing one to interpret the optimal dual variable in terms of tropical cramer determinants .
corollary 5.4 of @xcite only deals with the case where the matrix @xmath382 is generic , whereas theorem [ t_rgst ] shows that this assumption is not needed for the optimal solution of the dual problem @xmath1185 to be unique , up to a transformation by an additive constant .
genericity is only needed for the uniqueness of the optimal solution of the primal problem , which we do not use here .
[ t_rgst ] the primal problem @xmath1186 is feasible if and only if all the tropical cramer permanents of the matrix @xmath382 are finite .
when this is the case , the optimal solution @xmath1187 of the dual transportation problem @xmath1185 is unique up to a modification of the vector @xmath1188 by an additive constant , and @xmath1189 } u_i + \sum_{j\in [ n]\setminus \{k\ } } v_j \enspace , \qquad \text { for all } k\in[n ] \enspace .\end{aligned}\ ] ] we first consider the case in which all @xmath1190 are finite . as the initial step , we show the announced uniqueness result for the optimal solution @xmath1191 of the dual problem .
let @xmath610 denote an optimal solution of the primal problem , which we choose to be a vertex of @xmath1192 , so that @xmath1172 is a spanning tree of @xmath1157 . using the complementary slackness condition
, we have @xmath1193 using these relations and the fact that a spanning tree is connected , we see that all the values of the variables @xmath1194 and @xmath1195 , @xmath1196 $ ] and @xmath1027 $ ] , are uniquely determined by the value of a single variable @xmath1197 $ ] ( or dually , by the value of a single variable @xmath1198 $ ] )
. moreover , an increase of @xmath1199 by a constant increases every other variable @xmath1194 by the same constant and decreases every variable @xmath1195 by the same constant .
this establishes the announced uniqueness result .
since @xmath1200 holds for all @xmath1196 $ ] and @xmath1027 $ ] , we deduce that for all @xmath1152 $ ] and for all bijections @xmath224 from @xmath1155 $ ] to @xmath9\setminus \{k\}$ ] @xmath1201 } c_{i\sigma(i)}{\leqslant}\sum_{i\in[n-1 ] } u_i + \sum_{j\in [ n]\setminus \{k\ } } v_j \enspace .
\label{e - tight}\end{aligned}\ ] ] considering the maximum over all such bijections @xmath224 , we deduce that @xmath1202 is bounded from above by the right - hand side of the latter inequality .
now , using again the fact @xmath1192 , we can write @xmath1203 where every @xmath1170 represents a matching between @xmath1155 $ ] and @xmath9\setminus\{k\}$ ] , to which we associate a bijection @xmath1204 from @xmath1155 $ ] to @xmath9\setminus\{k\}$ ] . by the complementary slackness condition we have @xmath1205 for all
@xmath1196 $ ] .
it follows that @xmath1206 achieves the equality in , showing that @xmath1202 is given by the right - hand side of .
we now deal with the case in which some coefficients @xmath1190 can take the @xmath8 value .
the reward function @xmath1207,\,j\in[n]}c_{ij}y_{ij}$ ] over @xmath1148 , with the convention @xmath1183 , now takes its value in @xmath1 .
it is upper semicontinuous and concave , so it attains its maximum at a vertex of @xmath1192 .
it follows that the value of the primal problem is finite if and only if every tropical cramer permanent is finite .
then the remaining arguments of the proof above are easily checked to carry over , by working with the modified linear programming formulation , in which the constraints @xmath1184 for all @xmath397 such that @xmath1182 , are added .
to compute determinants over @xmath336 or @xmath25 we need to recall how tropical singularity can be checked .
let @xmath1135 .
we observed in proposition [ prop - hungarian ] that , as soon as @xmath1208 , the hungarian algorithm gives scalars @xmath1209 , for @xmath436 $ ] , such that @xmath1210 the optimal permutations @xmath224 are characterized by the condition that @xmath1211 , for all @xmath18 $ ] . after multiplying @xmath38 by a permutation matrix
, we may always assume that the identity is a solution of the optimal assignment problem .
then we define the digraph @xmath264 with nodes @xmath407 , and an arc from @xmath392 to @xmath393 whenever @xmath1212 .
butkovi proved two results which can be formulated equivalently as follows .
[ thbut ] let @xmath1135 , and assume that @xmath1208 .
then checking whether the optimal assignment problem has at least two optimal solutions reduces to finding a cycle in the digraph @xmath264 , whereas checking whether it has at least two optimal solutions of a different parity reduces to finding an even cycle in @xmath264 .
assume first that @xmath825 .
we set @xmath1214 , meaning that @xmath1215 for all @xmath471 , and compute @xmath1216 together with a permutation @xmath224 solving the optimal assignment problem for the matrix @xmath387 .
if @xmath1217 , then @xmath1218 , so we next assume that @xmath1219 . after multiplying @xmath38 by the inverse of the matrix of the optimal permutation @xmath224 , we may assume that this permutation is the identity .
if one of the diagonal coefficients belongs to @xmath1220 , then we conclude that @xmath1221 .
otherwise , we define the digraph @xmath264 as above , starting from the matrix @xmath387 instead of @xmath38 .
if @xmath264 has a cycle , then this cycle can be completed by loops ( cycles of length one ) to obtain an optimal permutation for @xmath387 distinct from the identity and so @xmath1222 .
otherwise , @xmath1223 . to compute @xmath656 when @xmath1213 , we consider firstly the case @xmath1135 , where @xmath0 is thought of as a subsemiring of @xmath25 .
we define again @xmath1214 and assume that the identity is an optimal permutation for the matrix @xmath387 and we define @xmath264 as above .
then @xmath1223 if @xmath264 has no even cycle .
otherwise @xmath1222 since any even cycle can be completed by loops to get an optimal permutation for @xmath387 of odd parity .
we assume finally that @xmath1213 and make the same assumptions and constructions .
if one diagonal entry of @xmath38 is in @xmath1224 , we conclude that @xmath1222 . otherwise ,
for every @xmath397 in @xmath264 such that @xmath1225 , we check whether @xmath397 belongs to a cycle of @xmath264 ( not necessarily an even one ) .
if this is the case , we conclude that @xmath1226 . otherwise , the elements @xmath1227 such that @xmath1225 do not contribute to the optimal permutations of the matrix @xmath387 , and we may replace them by @xmath173 without changing the value of @xmath656
. then we may write @xmath1228 where @xmath1229 .
recall that for all matrices @xmath1230 with entries in a commutative ring the following block formula for the determinant holds @xmath1231 where @xmath415 is the identity matrix .
moreover , when viewing the entries of @xmath382 and @xmath373 as independent indeterminates and expanding the expressions at left and at right of the latter equality , we see that the same monomials appear ( each with multiplicity one ) on each side of the equality . therefore , the equality is valid over an arbitrary symmetrized semiring . in particular ,
@xmath1232 which reduces the computation of @xmath656 to the computation of the determinant of a matrix with entries in @xmath0 , which is already solved thanks to theorem [ thbut ] .
m. akian , s. gaubert , and a. guterman .
linear independence over tropical semirings and beyond . in g.l .
litvinov and s.n .
sergeev , editors , _ proceedings of the international conference on tropical and idempotent mathematics _ ,
volume 495 of _ contemporary mathematics _ , pages 138 .
american mathematical society , 2009 .
m. develin , f. santos , and b. sturmfels . on the rank of a tropical matrix . in _ combinatorial and computational geometry _ ,
volume 52 of _ math .
_ , pages 213242 .
cambridge univ . press , cambridge , 2005 .
j. richter - gebert , b. sturmfels , and t. theobald .
first steps in tropical geometry . in g.l .
litvinov and v.p .
maslov , editors , _ idempotent mathematics and mathematical physics _ , volume 377 of _ contemp . math .
_ , pages 289317 .
soc . , providence , ri , 2005 .
o. viro .
dequantization of real algebraic geometry on logarithmic paper . in _
european congress of mathematics , vol .
i ( barcelona , 2000 ) _ , volume 201 of _ progr . math .
_ , pages 135146 .
birkhuser , basel , 2001 . | we prove general cramer type theorems for linear systems over various extensions of the tropical semiring , in which tropical numbers are enriched with an information of multiplicity , sign , or argument .
we obtain existence or uniqueness results , which extend or refine earlier results of gondran and minoux ( 1978 ) , plus ( 1990 ) , gaubert ( 1992 ) , richter - gebert , sturmfels and theobald ( 2005 ) and izhakian and rowen ( 2009 ) .
computational issues are also discussed ; in particular , some of our proofs lead to jacobi and gauss - seidel type algorithms to solve linear systems in suitably extended tropical semirings . |
we are grateful for conversations with m. albrow and c. chiu .
this work was supported in part by contracts de - fg03 - 93er40757 and de - ac02 - 76ch00016 with the u.s .
department of energy
. 99 m. berger and m. chanowitz , phys .
lett . * 68 * , 757 ( 1992 ) .
h. haber , g. kane , and t. sterling , nucl .
* b161 * , 493 ( 1979 ) . for a review , see j. gunion , h. haber , g. kane , and s. dawson , _ the higgs hunter s guide _ ( addison - wesley , new york , 1990 ) .
h. georgi , s. glashow , m. machacek , and d. nanopoulos , phys .
lett . * 40 * , 692 ( 1978 ) .
k. gaemers and f. hoogeveen , phys .
* 146b * , 347 ( 1984 ) .
l. jones and h. w. wyld , phys .
d * 17 * , 1782 ( 1978 ) ; b. combridge , nucl . phys . * b151 * , 429 ( 1979 ) ; r. k. ellis and j. sexton , nucl . phys . *
b282 * , 642 ( 1987 ) .
aleph , delphi , l3 , opal , and the lep electroweak working group , cern - ppe-93 - 157 ( 1993 ) .
f. berends , j. tausk , and w. giele , phys .
d * 47 * , 2746 ( 1993 ) .
l. resnick , phys .
d * 2 * , 1975 ( 1970 ) ; j. pumplin , phys .
d * 2 * , 1859 ( 1970 ) ; t. bauer , phys .
lett . * 25 * , 485 ( 1970 ) .
basdevant and e. berger , phys .
d * 16 * , 657 ( 1977 ) ; d * 19 * , 239 ( 1979 ) ; d * 19 * , 246 ( 1979 )
. d. morgan and m. pennington , z. phys .
* c37 * , 431 ( 1988 ) .
basdevant , e. berger , d. dicus , c. kao , and s. willenbrock , phys .
* 313b * , 402 ( 1993 ) .
t. kesson _ et al .
_ , nucl .
* b264 * , 154 ( 1986 ) . c. n. yang , phys .
* 77 * , 242 ( 1950 ) .
height0pt fig .
1 - feynman diagrams for @xmath65 : ( a ) @xmath66-channel higgs scalar ( @xmath7 ) or pseudoscalar ( @xmath25 ) exchange via a top - quark loop ( crossed diagram not shown ) , ( b ) leading - order qcd ( @xmath67- and @xmath66-channel diagrams not shown ) . fig .
2 - total cross section for @xmath10 , for @xmath23 170 gev , as a function of the @xmath2 invariant mass .
the calculation includes the effects of the heavy higgs scalar ( @xmath7 ) of the minimal supersymmetric model ( with the ratio of the vacuum - expectation values of the two higgs doublets close to unity ) , as well as the continuum qcd production of @xmath2 , for @xmath24 400 , 500 , 600 , 700 , and 800 gev .
4 - luminosity function for gluon - gluon collisions at hadron colliders of various energy .
multiplication of the @xmath0 cross section in figs . 2 and 3 at a given invariant mass , @xmath52 , by the luminosity function at the same invariant mass yields the inclusive hadronic differential cross section , @xmath53 , for @xmath54 .
the gluon distribution functions have been evaluated at @xmath68 . | higgs bosons which decay principally to top quarks , such as in the minimal supersymmetric model , produce a peak - dip structure in the @xmath0 invariant - mass spectrum .
this structure is potentially observable at the cern large hadron collider .
-.5 in 6.5 in * higgs decay to top quarks at hadron colliders * + + center for particle physics + university of texas + austin , tx 78712 + + physics department + brookhaven national laboratory + upton , ny 11973 + + physics department + university of illinois + 1110 w. green street + urbana , il 61801 + [ [ section ] ] in the standard model of the electroweak interaction , the electroweak symmetry is broken by a scalar @xmath1 doublet higgs field which acquires a vacuum - expectation value .
this mechanism provides masses for the weak vector bosons , as well as for the quarks and leptons .
however , the mechanism which is responsible for the quark and lepton masses need not be as closely linked to the mechanism which generates the weak - vector - boson masses as it is in the standard model . since fermion masses violate the @xmath1 symmetry , whatever provides their masses must break this symmetry , and contribute to the weak - vector - boson masses
. however , this contribution may be only a small component of the weak - vector - boson masses , which may arise dominantly via another mechanism @xcite .
an explicit realization of this scenario is provided by a two - higgs - doublet model .
one doublet may acquire a much larger vacuum - expectation value than the other doublet , and thereby provide the dominant contribution to the weak - vector - boson masses .
however , this doublet need not couple to fermions , in which case the fermion masses arise solely from the higgs doublet with a small vacuum - expectation value @xcite .
a two - higgs - doublet model produces a spectrum of physical particles which consists of two neutral higgs scalars , one neutral pseudoscalar , and a charged higgs boson @xcite .
the higgs scalar most closely associated with the doublet with small vacuum - expectation value couples very weakly to the weak vector bosons , but couples with enhanced strength to the fermions ( with respect to the standard - model coupling )
. the pseudoscalar higgs particle does not couple to the weak vector bosons at all , but couples to the fermions with enhanced strength . in this letter
we consider searching for a higgs boson , with zero or suppressed coupling to the weak vector bosons , via its decay to @xmath2 .
this particle is copiously produced at a hadron collider through gluon - gluon collisions , via a virtual top - quark loop @xcite , as shown in fig .
1(a ) .
however , at a hadron collider there is a large irreducible background from the qcd production of top quarks , shown in fig .
1(b ) .
we will show that the signal , @xmath3 , and the background , @xmath4 , interfere , generically resulting in a peak - dip structure at the higgs mass .
this phenomenon was first recognized in ref .
@xcite . in some cases the dip dominates , such that the signal for the presence of the higgs boson is a small _ deficit _ in the production of @xmath2 pairs of invariant mass near the higgs mass .
another model which yields higgs bosons with suppressed coupling to weak vector bosons is the higgs sector of the minimal supersymmetric model @xcite .
this model necessarily has two higgs doublets , one which couples to fermions with @xmath5 and one which couples to fermions with @xmath6 .
as usual , the pseudoscalar higgs boson does not couple to the weak vector bosons .
if this pseudoscalar is relatively heavy ( greater than about 100 - 150 gev , depending on the ratio of the vacuum - expectation values of the two higgs doublets ) , the heavier higgs scalar ( whose mass is then close to that of the pseudoscalar ) couples only weakly to the weak vector bosons .
this is true regardless of the ratio of vacuum - expectation values of the two higgs doublets . for definiteness , we consider the higgs sector of the minimal supersymmetric model , with the ratio of the vacuum - expectation values of the two higgs doublets close to unity .
the heavier higgs scalar , @xmath7 , nearly decouples from the weak vector bosons if its mass is greater than about 150 gev , and its coupling to top quarks is close to standard - model strength .
our qualitative results are generic to the scenario where the dominant decay of the higgs boson is to top quarks .
if the ratio of vacuum - expectation values is much larger than unity , the coupling of @xmath7 to the top quark is suppressed and its coupling to the bottom quark enhanced , such that the production of the higgs via a bottom - quark loop , and its decay to @xmath8 , become non - negligible . in the case of a non - supersymmetric
scalar higgs boson from a two - doublet model , the decay of the higgs to weak vector bosons may also be non - negligible .
both of these scenarios decrease the branching ratio of @xmath9 , and hence decrease the size of the effect of the higgs boson on top - quark production . the differential cross section for @xmath10 , including the squares of the scalar - higgs amplitude and the continuum qcd amplitude , as well as the interference of the two amplitudes , is , the cosine of the scattering angle between a gluon and the top quark .
the resulting expression agrees with that of eq . 7 of ref .
@xcite , except the interference term , which is a factor of 2 too large in that paper .
the resulting cross section , fig . 2 in that paper , should display a large peak , as shown , followed by a small dip , rather than an equally - large peak and dip . ]
@xmath11 \nonumber \\ & + & \frac{d\sigma_{qcd}}{dz}\end{aligned}\ ] ] where @xmath12 are the momenta of the incoming gluons , @xmath13 are the outgoing top - quark and top - antiquark momenta , @xmath14 is the cosine of the scattering angle between an incoming gluon and the top quark , @xmath15 is the top - quark mass , and @xmath16 is the velocity of the top quark and top antiquark in the center - of - momentum frame .
the dot product of the four momenta are @xmath17 the function associated with the virtual top - quark loop is @xmath18\ ] ] where @xmath19 ^ 2 \quad(s>4m^2)\;.\ ] ] the energy - dependent higgs width is @xmath20 the cross section for the continuum qcd production of @xmath0 is @xcite @xmath21\;.\ ] ] we show in fig .
2 the cross section for @xmath10 as a function of the @xmath2 invariant mass , @xmath22 , for @xmath23 170 gev ( the approximate central value from precision electroweak experiments @xcite ) , so the qualitative results for other top - quark masses can be obtained by scaling the higgs mass , modulo the effect of the change in the higgs width . ] and @xmath24 400 , 500 , 600 , 700 , and 800 gev . for @xmath24 400
gev , the higgs boson produces a narrow peak with a width of about 3.3 gev . for @xmath24 500 and 600 gev ,
the presence of the higgs boson produces a peak , followed by a dip , near the higgs mass . for
larger higgs masses the peak is absent , and the higgs reveals itself as a dip in the @xmath2 invariant - mass spectrum . for @xmath24 500 gev , the _ total _ top - quark cross section differs little from the cross section with no higgs present , due to the cancellation between the peak and the dip . for larger higgs masses , the presence of the higgs results in a small decrease in the total top - quark cross section .
we also consider the effect of the pseudoscalar higgs boson , @xmath25 , on the @xmath2 invariant - mass spectrum .
this particle does not couple to the weak vector bosons , and couples to the top quark with standard - model strength if the ratio of the vacuum - expectation values of the two higgs doublets is close to unity . for larger values of this ratio , the pseudoscalar higgs coupling to top quarks
is suppressed , so its width is narrower .
the production of the pseudoscalar higgs via a bottom - quark loop and the decay to @xmath8 may also become significant , reducing the branching ratio to @xmath2 and the effect of the higgs on top - quark production .
however , the pseudoscalar higgs does not couple to weak vector bosons in a generic two - higgs - doublet model , so there is no competition from the decay to weak vector bosons . the differential cross section for @xmath10 , including the squares of the pseudoscalar - higgs amplitude and the continuum qcd amplitude , as well as
the interference of the two amplitudes , is @xmath26 \nonumber \\ &
+ & \frac{d\sigma_{qcd}}{dz}\;.\end{aligned}\ ] ] there is no interference between the scalar - higgs and the pseudoscalar - higgs amplitudes .
the function associated with the virtual top - quark loop is @xmath27 where @xmath28 is given above , and the energy - dependent higgs width is @xmath29 the resulting cross sections , given in fig .
3 , are qualitatively similar to the cross sections in fig . 2 for the scalar higgs .
the pseudoscalar - higgs width is suppressed by @xmath30 , rather than @xmath31 as is the scalar - higgs width , so the pseudoscalar higgs boson is noticeably wider than the scalar higgs boson for @xmath32 400 and 500 gev .
the peak - dip structure in the @xmath2 invariant - mass spectrum due to the scalar or pseudoscalar higgs bosons can be viewed as due to a final - state interaction of the @xmath2 pair @xcite .
if one cuts vertically through the top - quark loop in fig .
1(a ) , putting the cut propagators on shell , the amplitude factorizes into the product of the amplitudes for the qcd production of @xmath2 ( fig .
1(b ) ) and the elastic scattering of the @xmath2 pair via the higgs resonance , times a phase - space factor @xmath30
. the elastic @xmath2 partial - wave scattering amplitude may be represented by a unitary phase - shift expression , @xmath33 where @xmath34 is the phase shift , which passes through @xmath35 at @xmath36 . using the cutkosky rules , the absorptive part of the color - singlet , zeroth - partial - wave higgs amplitude is thus @xmath37 , where @xmath38 is the color - singlet , zeroth partial wave of the qcd amplitude .
the total color - singlet , zeroth - paritial - wave amplitude is the sum of the qcd amplitude and the higgs amplitude , @xmath39 \nonumber \\ & = & a_0^{qcd}\frac{s - m_h^2}{s - m_h^2+im_h\gamma_h(s ) } + a_0^{disp}\frac{-m_h\gamma_h(s)}{s - m_h^2+im_h\gamma_h(s)}\end{aligned}\ ] ] where @xmath40 is the one - loop - induced coupling of the higgs to the initial gluons due to the dispersive part of the top - quark loop .
we see that the interference of the absorptive part of the loop diagram with the tree diagram yields an exact zero at the resonance energy , @xmath36 ( @xmath41 ) .
the one - loop - induced coupling due to the dispersive part of the loop produces the usual resonant peak , so generically a peak - dip structure results . if the dispersive part of the loop is small compared with the absorptive part , the dip will be the dominant feature of the structure in the invariant - mass spectrum .
this is what occurs for @xmath32 700 and 800 gev in figs . 2 and 3 .
peak - dip structures due to final - state interactions are well - known in hadronic physics @xcite .
an example of a dip is the effect of the @xmath42 on the @xmath43 invariant - mass spectrum in double - diffractive pion production @xcite .
the argument in the preceding paragraph shows that a peak - dip structure in the @xmath2 invariant - mass spectrum results whenever there is a resonant final - state interaction of the @xmath2 pair , regardless of the physics which produces the resonant final - state interaction .
invariant - mass distribution at leading order , because the incoming gluons are forbidden to form a state of unit total angular momentum by yang s theorem @xcite . ]
it also shows that the peak - dip structure is not washed out by qcd corrections , since it depends only on the final - state interaction of the @xmath2 pair , whose production may in principle be calculated to any order in qcd . at the fermilab tevatron ,
a top quark of mass greater than 150 gev is predominantly produced via quark - antiquark annihilation .
for example , for @xmath23 170 gev , the gluon - fusion contribution to the total top - quark cross section is about @xmath44 @xcite .
thus any structure in the @xmath45 cross section due to a higgs boson would be buried beneath the @xmath46 continuum , and require very large statistics to uncover .
in contrast , at a higher - energy hadron collider , such as the cern large hadron collider ( lhc ) , the gluon - fusion process is the dominant source of top quarks . at a 4 tev @xmath47 collider ,
the gluon - fusion process accounts for about half of the total top - quark cross section ( for @xmath48 170 gev ) .
the @xmath2 invariant mass can be reconstructed at hadron colliders via @xmath49 , followed by hadronic decay of one @xmath50 boson and leptonic decay of the other .
the presence of two @xmath50 bosons and two top quarks constrains the kinematics sufficiently that the event can be fully reconstructed , in principle .
we assume in the following that the @xmath2 invariant mass can be reconstructed with reasonable resolution .
the effect of the scalar and pseudoscalar higgs bosons on the @xmath2 invariant - mass distribution is quite small , and will require large statistics to observe .
for example , the peak and dip in the invariant - mass distribution in fig .
2 due to a scalar higgs of mass 500 gev are each about a @xmath51 effect , each contained in a bin of width about 25 gev .
the statistical significance of the peak and dip depend on the number of @xmath2 events in the mass bin containing the signal .
we present in fig .
4 the luminosity function of gluon - gluon collisions at hadron colliders of various energy .
multiplication of the @xmath0 cross section in figs .
2 and 3 at a given invariant mass , @xmath52 , by the luminosity function at the same invariant mass yields the inclusive hadronic differential cross section , @xmath53 , for @xmath54 . at the lhc , with 100 @xmath55 of integrated luminosity , the raw number of events in an invariant - mass bin of width 25 gev at @xmath56 500 gev is about @xmath57 . including a branching ratio for the one hadronic - one leptonic decay of the @xmath50 bosons of about 0.3 , and a factor of 1/3 to account for acceptance and efficiencies ,
leaves approximately @xmath58 events .
this is far more than necessary to detect a @xmath51 effect at the @xmath59 level . for a scalar higgs of mass 800 gev
, the dip corresponds to a @xmath60 decrease in the number of events in a bin of width 50 gev .
the number of events , including branching ratios , acceptance , and efficiencies , is about @xmath61 , yielding a significance of about 3.5@xmath62 for the signal , marginally significant .
the results for the pseudoscalar higgs are comparable .
lower - energy machines also have the potential to observe a signal , depending on the luminosity and the higgs mass .
a 4 tev @xmath47 collider requires about 100 @xmath55 of integrated luminosity to detect a scalar higgs of mass 500 gev with a 5@xmath62 significance .
however , the fermilab tevatron would require an integrated luminosity far in excess of 100 @xmath55 to detect such a higgs boson .
in contrast to the class of models being considered here , in the standard higgs model the branching ratio of the higgs boson to @xmath2 is much less than unity . for @xmath23 170
gev , the branching ratio is at most @xmath63 ( this value occurs for @xmath64 500 gev ) .
the higgs boson produces only a tiny wiggle in the @xmath2 invariant - mass spectrum at the higgs mass , too small to appear in figs . 2 or 3 .
we conclude that the peak - dip structure in the @xmath0 invariant mass distribution , due to a scalar or pseudoscalar higgs boson decaying principally to top quarks , is potentially observable at the lhc , if the @xmath2 invariant - mass resolution is less than the width of the higgs boson .
although the standard - model higgs boson is not of this type , higgs bosons which decay predominantly to top quarks are natural in two - higgs - doublet models , in particular the heavy scalar higgs , @xmath7 , and the pseudoscalar higgs , @xmath25 , of the minimal supersymmetric model .
the peak - dip structure is unfortunately not observable at the fermilab tevatron with 1 @xmath55 of integrated luminosity , due to the limited statistics and the relatively large rate of production of top quarks via quark - antiquark annihilation . |
debris disks around main - sequence stars are dusty , optically thin , and gas - poor .
radiation pressure ( rp ) and poynting - robertson ( pr ) drag eliminate dust grains on timescales shorter than the stellar age , so the observed dust must be continuously replenished by collisions among , or evaporation of , planetesimals @xcite . the _ infrared astronomical satellite ( iras ) _ and the _ infrared space observatory ( iso ) _ revealed over 100 stars with far - infrared excesses indicating the presence of debris disks .
however , spatially resolved images of these disks are relatively rare : only about a dozen debris disks have been resolved since the early 1980s ( see @xcite and references therein ) .
the resolved disks are not featureless .
they frequently display warps , spiral structures , and other azimuthal and radial asymmetries .
different mechanisms have been suggested to explain these features .
they may be caused by interactions between the dust and the gas in the disk ( @xcite , hereafter ta01 ) , the formation of small planets ( @xcite ) , the dynamical forces of embedded planets ( @xcite ) , or stellar - mass companions ( @xcite ; @xcite ; @xcite ) .
whatever their cause(s ) , the features observed in the dust disks provide insight into the characteristics of the unseen planetesimal population and illuminate the dynamical processes of young planetary systems .
hd 141569a ( a0 v ; age 5@xmath63 myr , @xcite ) has a resolved circumstellar debris disk .
its _ hipparcos _ distance is 99 pc from the sun and has co - moving m2 v and m4 v companions separated by 1.4 and located at distances of 755 and 893 , respectively ( @xcite ; @xcite ) .
the disk was first resolved in scattered light by @xcite and @xcite using the _ hubble space telescope s ( hst s ) _ near - infrared camera multi - object spectrometer ( nicmos ) .
@xcite imaged the disk with _
hst s _ space telescope imaging spectrograph ( stis ) and noticed strong brightness asymmetries .
these observations have recently been complemented by images from _
hst s _ advanced camera for surveys ( acs ) ( @xcite , hereafter c03 ) .
together , these scattered - light images reveal very complex structure , including an inner clearing within 175 au of the star , a bright spiral `` ring '' with a sharp inner edge from 175 to 215 au , a faint zone from 215 to 300 au , and a broad spiral `` ring '' from 300 to 400 au ( c03 ) .
these distances are measured along the projected disk s southern semimajor axis .
( for convenience , we will hereafter refer to the two tightly - wound spiral rings simply as `` rings . '' ) c03 also observed two low - intensity , large - scale spiral arms in the outermost part of the disk , that they attributed to tidal interaction with the m dwarf companions .
the influence of the m dwarf companions on the disk s morphology was first discussed by @xcite .
they argued that if all three stars were coplanar with the disk , the system would not be stable .
given its young age , however , the system may be bound . with this assumption
, @xcite concluded that resonant interactions between the companions and the disk do not account for the disk structure . assuming that the companions revolve around the primary star in a highly eccentric orbit , @xcite constructed dynamical models of the optical disk comprising dust grains that respond to the stellar gravitational field .
they assumed that the grains are large enough that rp , pr drag , and gas drag are unimportant .
they also ignored the presence of the large - scale spiral arms reported by c03 .
their model reproduces the general appearance of the brightness asymmetries observed in the disk .
@xcite constructed hydrodynamic models of hd 141569a s gas disk without dust , and concluded also that the companions lie in an eccentric orbit .
their models reproduced the large - scale spiral arms ( even after several periastron passages ) and some other disk asymmetries .
ta01 modeled the interaction between disks of gas and intermediately - sized ( @xmath7 m ) dust grains , and @xcite considered disks with very small particles .
it is not clear that the conditions explored by any of these models is the correct one for the scattered light images of the disk around hd 141569a . in [ obscons ] , we argue that submicron - sized grains account for most of the scattering opacity at optical wavelengths .
such small grains are subject to rp which quickly expels them from the disk .
the gas detected in the disk @xcite tempers somewhat the short `` blow out '' timescales , but its mass is not enough to dominate the dynamics of the dust ( [ conf ] ) . in this paper , we explore the effects of strong rp , gas drag , and gravity on small dust grains , using hd 141569a s disk as a benchmark .
we simultaneously track the behavior of three different populations
the unseen planetesimals controlled by gravity , the gas controlled by gravity and volumetric fluid forces , and the dust controlled by gravity , rp , and gas drag . the insights obtained from this analysis should be applicable to other systems where rp is important . in [ obscons ] , we review the observational constraints that every model of the system should satisfy .
we consider evidence that indicates the dominance of submicron - sized grains and we set limits on the amount of gas in the disk . in [ conf ]
, we explore the interaction between gas and dust by means of a one - dimensional model .
we show that the gas slows the outward motion of the dust and can produce broad ring - like structures . in [ model ] , we present two - dimensional dynamical simulations of the interactions between planetesimals , dust grains , gas , and the three stars . assuming that the two m dwarf companions are co - moving but unbound , we find that a recent parabolic fly - by causes some of the observed disk structure , including the large - scale arms . however , our models produce a more disorganized disk than is observed .
we argue that the central hole in the disk can be produced by a planet with a mass a few times that of jupiter ( or , alternatively , some number of smaller planets ) in a highly eccentric orbit about the primary star .
figure [ profiles ] shows the optical depth profile of the disk , taken from figure 5 of c03 .
this profile represents the median values of concentric annuli centered on the primary star .
drawn from a composite of f435w ( acs @xmath8 band ) and f606w ( acs broad-@xmath9 band ) images , it is roughly indicative of the optical characteristics of the disk at @xmath10 m .
power - law fits to the inner and outer edges of the profile ( using as errors the standard deviation at every radius ) give @xmath11 for @xmath12 au and @xmath13 for @xmath14 au . beyond 500 au
, the average azimuthal brightness seems to be dominated by the light from hd 141569bc .
the optical depth profile indicates that the amount of mass in @xmath15m - sized grains is @xmath16m@xmath17 ( assuming a constant dust opacity of @xmath18 @xmath19/gr and albedo of @xmath20 .
see @xcite ) .
the disk is occulted by acs s coronagraphic mask within @xmath21 au . @xcite and @xcite detected mid - infrared thermal emission within this radius and @xcite concluded that the optical depth at @xmath22 m decreases by a factor of @xmath23 within this region .
* hereafter ll03 ) modeled the thermal spectrum of the disk by assuming a density profile similar to the one shown in figure [ profiles ] .
their model suggests that the reduced emission within @xmath24150 au is due to reduced dust density and not a change in the scattering properties of the grains .
c03 reported that the disk is redder than the star , with color excesses of @xmath25 and @xmath26 .
they also reported no color variation as a function of distance from the star ( which supports our use of a constant opacity in the mass calculation above ) .
c03 inferred that the disk s colors are consistent with astronomical - silicate grains having a size distribution of @xmath27 and a minimum radius of @xmath28 m .
@xcite used the same color information to derive lower size limits between @xmath29 m and @xmath30 m .
the existence of such small grains is also implied by the small scattering asymmetry factor ( @xmath31 for the hd 141569a disk is quoted as @xmath32 in c03 ] ) of the henyey - greenstein function derived by c03 .
the multicomponent models by ll03 suggest that the minimum grain size is between @xmath33 m and @xmath34 m : the smaller limit produces too little emission in the _ iras _ 60 @xmath35 m band and the larger limit produces too much emission in the _ iras _ 60 @xmath35 m and 100 @xmath35 m bands . taken together ,
these arguments imply that there is a population of grains whose radii may extend down to @xmath33 m .
the combined influences of the reduced scattering efficiency ( @xmath36 ) of small grains and the small number of large grains ( assuming a size distribution going as @xmath27 ) produce sharply peaked scattering opacity , which indicates that the grain sizes responsible for the optical images of the disk are of the order of the wavelength of observation .
figure [ draine ] shows the scattering opacity at @xmath10 m as a function of grain radius for astronomical silicate grains @xcite larger than @xmath37 m .
the scattering opacity is expressed as the scattering cross section per unit mass ( @xmath38 , where @xmath39 ) , weighted by mass and number of grains at each radius @xmath40 @xcite .
the function is centered at @xmath41 m with a characteristic width of @xmath42 m . the sharp decrease in the scattering opacity of grains larger than @xmath41 m
is determined by the behavior of the first scattering peak in @xmath36 . in the limit of large grains ,
the scattering opacity decreases more slowly ( @xmath43 , for intervals of constant @xmath44 ) . for this dust model ,
@xmath45 m grains scatter @xmath46 times less than @xmath41 m grains , and @xmath47 times more than @xmath34 m grains .
the location and width of the peak in the scattering opacity depends on the surface properties of the grains and the exact mixture of silicates , ices , and empty space .
for example , increasing porosity moves the peak to smaller sizes and less - reflecting grains broaden the peak and move it to larger sizes @xcite . if as suggested by the disk colors , the phase of the scattering function and the spectral energy distribution
submicron - sized grains exist in the disk , their dynamics are strongly influenced by rp . according to ta01 ,
the ratio ( @xmath2 ) of the forces from rp and gravity is @xmath23 for a @xmath45 m dust grain , and @xmath48 for @xmath49 m . for the very porous grains assumed by ll03 , @xmath50 for @xmath51 m . formally , and in the absence of gas , grains with @xmath52 ( `` @xmath2-meteoroids '' ) are expelled from the disk .
grains with large @xmath2 will acquire a radial velocity on the order of their keplerian velocity in @xmath53 , where @xmath54 is the orbital period at a given position ( ta01 ) . at the edge of a 500 au disk around a 2.3 m@xmath55 star , @xmath56 yr for @xmath57 .
such short expulsion timescales are somewhat mitigated by the effects of another force : the gas drag .
@xcite measured the gas content of hd 141569a s disk using the detected @xmath58 transition for @xmath59co and the upper limit on the same line for @xmath60co .
they assumed that the star s distance was 200 pc , its disk had a radius of 130 au , and that h2/co @xmath24 10000 .
they calculated gas masses between 20 and @xmath61 for optically thin and optically thick limits , respectively .
assuming that the gas and dust disks are coincident , we recompute the gas content using the observed extent of the optical disk and the measured _ hipparcos _ distance to hd 141569a of 99 pc .
@xcite have generously provided us with their unpublished digitized co spectra , which are reproduced in figure [ zuckerman ] . the spectra show a double peak characteristic of sharp disks , although the depths of the central line - emission in both measurements are a little over @xmath62 per resolution element .
assuming an disk inclination of @xmath63 ( c03 ) , the velocity difference between the peaks is @xmath64 km s@xmath65 , which implies emission at @xmath66 au .
( the upper limit is dictated by the size of the optical disk . ) assuming h2/co @xmath24 4000 ( @xcite ) , the optical depth of the central line is @xmath67 , where @xmath68 is the surface density and @xmath69 is the excitation temperature of the gas @xcite .
nondetection of the @xmath60co line yields an upper limit of @xmath70 mk , so the opacity of the @xmath59co line is uncertain .
if the @xmath59co line is optically thick , then we obtain a gas mass of @xmath71 .
if the line is optically thin and fills a disk of radius 380 au , we obtain a gas mass of @xmath72 . in each case
, the gas is assumed to be in local thermal equilibrium ( lte ) .
we now explore the interaction between small dust grains , gas , and the radiation and gravitational fields of hd 141569a , while ignoring the presence of the binary companions .
most of the results of this axisymmetric treatment should be applicable to other debris and gas disks in which rp is important .
the regime that we explore here is that of modest amounts of gas ( up to 50 m@xmath17 ) and not very large values of @xmath2 ( @xmath73 ) .
our calculations are complementary to those of @xcite , who consider @xmath74 and ta01 , who consider @xmath75 .
notice that @xcite has shown that if the observed dust particles all have @xmath76 , a single ring of planetesimals in a gasless disk will produce a dust disk with surface density given by @xmath77 , as observed here .
we solve the equations of motion for a dust grain , assuming that it is subject to rp , pr and gas drag . the gas is assumed to be in a circular orbit and we assume the same dust model as ta01 .
the force of the gas on the dust is given by @xmath78 where @xmath79 is the gas density , @xmath40 is a dust grain radius , @xmath80 is 4/3 times the thermal velocity of the gas , and @xmath81 is the relative velocity of the dust with respect to the gas . to determine the trajectory of a dust grain , we assume that the surface density of the gas ( as well as that of planetesimals ) decreases radially as @xmath82 . this function is characteristic of the dust surface density of optically thick protoplanetary disks @xcite . in other words
, we assume that the gas and planetesimal density profiles of the optically thin disks are the same as the dust density profile of the optically thick disks .
we further assume ( unless otherwise stated ) that the gas is in lte , with temperature @xmath83 where @xmath84 is the stellar luminosity @xcite .
this implies that @xmath85 , the vertical scale of the gas disk , is given by @xmath86 without gas , the radial velocity of a dust grain is given by @xmath87^{1/2},\ ] ] where @xmath88 is the radius at which the dust grain is created and @xmath89 is the keplerian velocity at @xmath88 . in general , there is not an analytic expression for the radial velocity of a dust grain in the presence of gas . for the range of parameters we are considering here ,
the dust grain speeds are supersonic and almost radial at the outer edge of the disk ( figure [ rad_vel ] ) . the dimensionless stopping time
( the stopping time in units of the inverse local keplerian frequency ) at the edge of the disk is given by ( ta01 ) : @xmath90 which for the parameters considered here is @xmath91 .
notice that , unlike in the case considered by @xcite , the radial velocity shown here does not have a deceleration region , because in our case the gas surface density decreases with radius : the radiation pressure always dominates the dynamics of the dust .
the effect of gas on the grains depends on where the grains were created .
figure [ time_it_takes ] shows that for a fixed value of @xmath2 and without gas , particles that start closer to the star soon overtake particles farther out .
this changes with the presence of gas , because the drag force decreases outward , and for @xmath92m@xmath17 masses of gas , the particles created closer in will not overtake those created farther out
. figure [ time_it_takes ] also shows that the gas mass , not the grain size , determines the escape timescale : there is not much difference between @xmath57 and @xmath93 for @xmath94 of gas , because both the gas drag and the rp are proportional to @xmath2 .
if one assumes a rate of dust generation , the results of these calculations can be used to predict the steady - state dust surface density in an isolated disk . in such a state ,
grains are continuously generated by collisions and then blown away by rp . the dust surface density
( assuming conservation in the number of particles ) is given by @xmath95 where @xmath96 is the inner radius of dust generation , @xmath88 is the position at which the grains are created , @xmath97 is the number of grains created at @xmath88 , and @xmath98 is the radial velocity at @xmath99 of grains created at @xmath88 .
the dust creation rate per unit volume is taken to be v@xmath100 , where @xmath101 is the number of planetesimals per unit volume , @xmath102 is the collision cross section and v@xmath103 is the relative velocity between the planetesimals , which is proportional to the dimensionless planetesimal disk thickness ( @xmath104 ) times the keplerian velocity @xcite .
this implies @xmath105 where @xmath106 is the surface density of planetesimals at the creation point , which we assume to be proportional to @xmath82 . in practice , this simple prescription ( the `` quadratic '' dust generation prescription ) may be affected by uncertainties in the planetesimal size distribution and in the relative velocities among the the planetesimals . to explore the sensitivity of the results to the exact generation mechanism we also consider a `` linear '' dust generation prescription in which @xmath107 . for these calculations
we assume that the planetesimals and the gas are well mixed and therefore @xmath108 .
the resulting steady - state surface densities are shown in figures [ 1d_model ] and [ 1d_model2 ] . in the figures ,
the unit length is 1000 au .
figure [ 1d_model ] shows that even modest amounts of gas have a significant effect on the surface - density profiles .
the amounts of gas considered ( 10 and @xmath94 ) are not enough to confine the dust grains created at distances larger than 0.01 , but gas drag sharpens the profiles , as it slows down the dust particles .
notice that if dust creation starts very close to the central star , the resultant surface - density profiles are effectively featureless .
figures [ 1d_model ] and [ 1d_model2 ] also show that a broad ring of dust is produced by truncating the dust creation within a certain radius . just outside this limit ,
the number and surface density of planetesimals , and hence dust grains , are large .
far from the creation limit , the surface density of the dust decreases because of decreasing surface density of planetesimals and the geometric dilution of `` blown out '' dust grains created at smaller radii .
thus , if a mechanism exists for preventing the creation of dust at small radii ( for example , a planet that clears out the parent planetesimals ) , the result is , very naturally , a broad ring of dust at larger radii . the rate at which dust is lost from the disk depends on the rate of planetesimal erosion . in principle
, one can use use the former to estimate the latter .
ll03 calculated the amount of mass lost from hd 141569a s disk , assuming no gas , a grain - size distribution @xmath109 , @xmath2=1 for grains of all sizes , and a constant rate of mass loss throughout the age of the system .
they also assumed a surface - density profile for the dust that is slightly different than the one shown in figure [ profiles ] .
they concluded that @xmath110 of solid material have been lost , in particles with @xmath111 m , from the disk due to rp and pr drag . to examine the effect of the gas on the rate of mass loss
, we consider a disk of radius 500 au about hd 141569a and no binary companions .
figure [ timescale ] shows the time required for small dust grains to move from a given radius to the edge of the disk for different values of gas mass and temperature .
the presence of just @xmath112 of gas triples the escape time from 200 au , and a colder disk confines the dust more than a hotter one .
( equation [ fg ] shows that , neglecting @xmath80 , a colder denser gas disk , produces more drag ) . assuming a grain - size distribution @xmath113 , with @xmath94 of gas in lte , varying escape times , the ta01 dust model and a constant rate of mass - loss , we determine that @xmath114 of solid material have been lost over the @xmath115 myr age of the disk , in particles with sizes @xmath116 m . without gas , @xmath117 of solid material would have been lost .
the difference with ll03 is due to the fact that in our calculation different size particles have different blow - out timescales .
however , neither ll03 s estimate nor ours can be correct . because small grains ( @xmath118 m ) are quickly expelled from the disk , they must be the dominant component of the lost mass .
if the grain - size distribution is @xmath113 for grains with radii @xmath119 mm , then grains up to 10 @xmath120 m compose only ten percent of the disk mass .
the bulk of the dust mass comprises large grains that are eliminated in much longer timescales .
if the collisional processes responsible for the replenishment of small grains also produce large grains , then submillimeter observations should detect @xmath121 of dust , which greatly exceeds the amount of @xmath122 measured by @xcite .
one possible solution to this discrepancy is that the dust is not generated with a power - law size distribution , but with a distribution weighted toward small particles .
the collisional - evolution models of @xcite suggest that , although large variations in the dust distribution are possible , most of the mass produced by the collisions is in large particles .
these models assume that the collisional timescale is smaller than any other timescale in the disk .
another possible solution is that the current rate of dust creation is larger than it has been in the past .
this possibility is consistent with the idea that the disk has been recently stirred by a close encounter with a companion or unbound star . in section [ model ]
we show that a parabolic encounter would occur over timescales on the order of @xmath123 yrs . over this time the disk would lose ( according to the model above ) @xmath124 of solids in small particles , or @xmath125 in particles up to 1 mm in size , close to the measured value .
observationally , the presence of @xmath126 of solids in @xmath127m - sized particles ( section [ obscons ] ) suggests that there is currently on the order of @xmath128 in particles up to 8 @xmath35 m in size , with the exact number depending on the assumed value of the opacity and dust size distribution . furthermore
, these estimates support the idea of a single exciting event , like an encounter with an unbound companion , as opposed to repeated exciting events . in the models by @xcite , which assume a bound companion , repeated encounters are required to explain the brightness asymmetries in the debris disk .
the presence of large - scale spiral arms suggests that the dynamical influence of the binary companions , hd 141569b and c , is important .
if they are coplanar with the disk , their center of mass is @xmath129 au away from the primary star and their separation is @xmath130 au .
@xcite assumed that the companions are bound to hd 141569a in a very eccentric orbit .
their dynamical models show that close encounters between the companions and the disk produced well - developed spiral rings within the disk after only a few periastron passages
. however , the models do not maintain large - scale spiral arms after a few orbits .
thus , within their model , the interaction geometry needed to create the spiral rings in the disk is not consistent with the presence of the large - scale spiral arms .
the hydrodynamic model by @xcite does produce spiral arms after repeated close encounters , because the viscosity and gas pressure help generate a spiral at every periastron passage .
however , the amount of gas actually present in the disk ( 2.2 ) discourages the notion that the dynamics of the dust conforms to that of the gas .
figure [ time_it_takes ] shows that , for a reasonable range of gas mass , @xmath2-meteoroid grains responsible for the scattered - light disk are impeded by gas drag , but they are not bound by the gas .
so the assumption by @xcite that gas dynamics control the observed structures in the dust disk is not appropriate .
we assume that the encounter between the disk and the binary companions is a parabolic fly - by , a situation known to produce spiral arms in dusty disks @xcite .
such an encounter is consistent with the proper motion of the system . within an error box of 01
, the position of the companions relative to hd 141569a has not changed over the 60 yr baseline of observations noted by @xcite .
the largest expected relative motion is obtained when the system is currently at periastron and the encounter occurs in the plane of the sky . in this scenario , the relative motion of the companions over 60
yr would be 0012 , which is well inside the error box of the proper motion measurements .
dust grains are created by collisions of planetesimals .
we assume that the planetesimals themselves experience gravity , but not rp or gas drag .
we also ignore self - gravity and the dynamical effects of the collisions .
our simulation is performed in two phases .
first , we obtain the distribution of planetesimals and gas as a function of time .
then , for each planetesimal configuration , we generate a dust distribution and follow it as a function of time .
we repeat this sequence for all planetesimal and gas configurations . at any given time , the dust in the disk consists of dust created at that time and dust remaining from earlier times .
the model is two - dimensional .
we assign fixed masses of 2.3 , 0.5 , and @xmath131 to hd 141569a , b , and c , respectively .
we assume that the encounter is prograde , as a retrograde encounter fails to produce large - scale spiral arms .
the phase of the encounter is fixed so that , at periastron , the three stars are aligned , with hd 141569b between a and c. ( the results are mostly insensitive to the choice of phase . )
most of our simulations start with @xmath132 planetesimals .
the inner dust destruction limit is set at @xmath133 , where the distance @xmath99 is measured relative to the separation of hd 141569a and hd 141569bc s barycenter at periastron .
we assume that the initial surface density of planetesimals declines as @xmath82 .
the outer radius of the planetesimal disk is a free parameter .
figure [ panel ] shows the evolution of the planetesimals in a disk of unit radius .
we define @xmath134 as the time of periastron .
the time unit is such that the period at unit distance is 2@xmath135 time units , where @xmath136 and @xmath137 ( the later is the sum of the companion masses ) .
notice that the particular fly - by configuration shown in figure [ panel ] shows that some planetesimals can be captured by the companions and therefore that some of the ir excess measured by _
iras _ may be associated with them .
the total mass of gas is a free parameter .
we consider gas disks with @xmath138 , an initial surface - density profile @xmath139 and masses of 0 , 10 , 50 , and 100 m@xmath17 , before truncation by the encounter .
we model the gas evolution using the smoothed particle hydrodynamics ( sph ) formalism @xcite , which allows one to track the motion of gas pseudo - particles using lagrangian equations .
viscosity is parametrized according to equation 4.2 of @xcite .
sph permits the exploration of parameter space without a large investment of computer time , but with limited resolution at small scales . we explore different limits for the gas sound speed , which is parametrized by the disk opening angle , @xmath140 , where @xmath85 is the scale height of the gas
, @xmath141 is the speed of sound , and @xmath142 is the keplerian angular frequency at radius @xmath99 .
we consider gas with @xmath143 , 0.1 , 0.2 , and in lte ( equation [ lte ] ) .
@xcite has shown that the empirical opening angle of the @xmath2 pictoris gas disk is @xmath144 , which is close to the expected lte value at 1000 au .
figure [ panel_gas ] shows the evolution of gas particles for the case of @xmath145 . to simulate the dust grains , we consider the linear ( particles created with @xmath146 ) and quadratic ( particles created with @xmath147 ) dust - creation prescriptions described in [ conf ] .
particles are created every 0.01 time units .
the ratio @xmath2 , which couples the dust and the radiation , is also a free parameter .
we explore two values : ( 1 ) @xmath57 , which corresponds to @xmath51 m ( ta01 ) or @xmath148 m ( ll03 ) , and ( 2 ) @xmath93 , which corresponds to @xmath149 m ( ta01 ) or @xmath51 m ( ll03 ) .
we assume a single value of @xmath2 per simulation , which is equivalent to assuming that the scattering opacity ( figure [ draine ] ) is a delta function . to obtain the dust distribution at a given time , we consider all the dust particles created at previous times .
figure [ multi_prof ] shows the behavior of dust particles created at various time intervals for @xmath150 , @xmath151 , @xmath145 , and linear dust creation .
the generations of particles were tracked to @xmath152 , and their respective surface densities were summed and azimuthally averaged at each time interval .
this simulation tracks the entire history of dust generation , but in practice , dust blowout eliminates from the `` current '' surface density profile all the contributions of dust created before the last half time unit ( i.e. , @xmath153 ) .
we now present our models for the fly - by encounter with a composite disk of planetesimals , gas , and dust .
representative simulations are shown in figures [ evolution][results4 ] .
although most of the models are shown at @xmath152 , the images and profiles are representative of @xmath154 .
to give our distance and time scales physical meaning , we set the distance between hd 141569a and the barycenter of the companions at @xmath152 equal to that observed in the acs images ( c03 ) .
thus , one distance unit corresponds to 718 au and one time unit corresponds to 1800 yr . the companions are out of the frames . for each of the figures [ evolution][results4 ] we show the predicted logarithm of the density profile with linear dust generation ( top row ) or quadratic dust generation ( bottom row ) .
the use of the logarithmic stretch allow us to highlight faint features .
the images are all normalized to the same arbitrary constant . in general
, the color table of the bottom row has a lower limit value that the color table at the top row .
this means that `` white '' in the bottom panels corresponds to somewhat less material than it does in the top panels . in this way
, we can show faint features in the bottom rows that would not be visible were we to use the same color table as that from the top rows .
this modification has little effect on the color the dark features , which implies that all panels within a given figure can be compared to each other . as mentioned before ,
an artificial inner hole has been set for the linear ( at @xmath155 ) and quadratic simulations ( at @xmath156 ) . in each panel
, the plot inset compares the measured density profile with the model density profile .
the latter is normalized to the value measured at 500 au .
figure [ evolution ] shows the evolution of the dust structures for @xmath4 , @xmath151 , @xmath145 , an initial disk size of one unit distance , and both linear and quadratic dust generation .
figures [ results1 ] and [ results2 ] show the effect of the amount of gas on @xmath57 and @xmath93 dust particles , respectively .
the sharp , concentric rings in the gasless models ( panels 1 and 4 ) are artifacts of discrete dust generation .
figure [ results3 ] illustrates the role of the gas temperature , and figure [ results4 ] shows the roles of disk sizes and planets .
a side - by - side comparison of the observed disk and our favorite model is presented in figure [ comparison ] .
all the models look similar . despite the dynamical importance of rp , the overall appearance of the dust disk
resembles the underlying distribution of planetesimals .
figures [ results1 ] and [ results2 ] show that models with gas reproduce the observed profiles better than models without gas , but it is not clear that @xmath112 of @xmath157 gas ( panel 2 of each figure ) is better than @xmath94 of @xmath157 gas ( panel 3 of each figure ) .
there are no significant qualitative differences between the @xmath57 and @xmath93 conditions shown in figures [ results1 ] and [ results2 ] , respectively , as expected from the results in [ conf ] .
there is also no clearly preferred prescription for dust creation .
the quadratic prescription does produce a more disorganized structure , as the dust is mostly generated in the denser spiral arms .
overdensities in the planetesimal distribution produce streams of matter in these models .
nevertheless , the combination of large @xmath2 and quadratic dust generation with @xmath157 gas ( figure [ results2 ] , panels 5 and 6 ) reproduces very well the decay of the surface - density profile . notice that because more gas confines the dust better , the spirals in panels 3 and 6 of each figure look darker than the others . figure [ results3 ] shows that the simulations are sensitive to the gas temperature .
cold gas ( panels 1 and 4 ) tends to produce sharper spiral structures than hot gas , because cold gas is denser and imparts greater drag on the dust .
this effect is less pronounced for quadratic dust generation , as the large density variations obscure the temperature - dependent effects of the gas .
all the models produce a two - armed spiral .
the arms are less pronounced for quadratic dust generation than for the linear dust generation because the former emphasizes regions of large planetesimal density .
the density and extent of the arm diametrically opposite the companions can be reduced ( and made more compatible with the observations ) by diminishing the size of the initial disk ( figure [ results4 ] ) .
the appearance of this arm can also be changed with a different encounter geometry .
@xcite show that , in a encounter between a circumstellar disk and a passing star , the spiral arm opposite the projectile is affected by the indirect tidal component of the potential , which diminishes as the collision becomes more hyperbolic . in general , the models with quadratic dust generation produce more disorganized images than models with linear dust generation .
planetesimal condensates are strong sources of dust that produce `` noise '' in the models .
however , they also produce more peaked surface - density profiles , which match the observed profiles better .
flatter initial density profiles ( not shown here ) produce large amounts of dust at large distances . disks with initial radii larger than @xmath241000 au produce remnant structures that are not observed , including very long spiral arms .
conversely , disks with an initial radius of 400 au yield dust distribution that reproduce well the observed large - scale arms and surface - density profiles ( figure [ results4 ] ) .
we surmise from our models that hd 141569a s initial planetesimal and gas disks had radii @xmath158 au and 1050 @xmath159 of @xmath157 gas ( figure [ comparison ] ) . from figure [ multi_prof ]
it seems clear that different dust configurations ( i.e. large holes ) can be created by manipulating the dust generation history of the system .
the size of the hole can be controlled by varying @xmath2 or the time between dust generations .
though theoretically possible , such contrived scenarios are physically unreasonable . a more reasonable agent for creating a hole in the dust distribution is a planet ( or planets ) within @xmath21 au .
if large enough , planets would clear out the planetesimals and the gas , and create a region without dust .
each planet would sweep up a lane along its orbit as wide as a few times its hill radius , @xmath160 . from @xcite , a @xmath5 planet with a semimajor axis of 100 au and eccentricity
@xmath161 would clear a region with a half - width of 90 au .
alternatively , multiple planets in less eccentric orbits would have similar effect . here
we present exploratory simulations that include the effect of a giant planet .
panels 2 and 4 of figure [ results4 ] show models that include a @xmath5 planet in an eccentric orbit ( @xmath162 ) around hd 141569a with a semi - major axis of 100 au .
the planet clears the central region enough to greatly reduce the dust generation close to the star and produce an inner hole in the dust disk .
weak signatures of the eccentricity ( the elongated shape and slight dust overdensities at the top and the bottom of the hole ) are observed .
the inner ring in the observed density profile is not reproduced in the observations , at least with these orbital parameters .
considering the uncertainties in the exact dust configuration close to the coronagraphic mask , this is a satisfactory result .
a @xmath5 planet of age 5 myr would be @xmath163 times less luminous , and over 12 magnitudes fainter in @xmath164 , than hd 141569a @xcite .
such an planet would be below the non - coronagraphic nicmos detection limits @xcite .
the formation of small planets may produce sharp rings in the disk by stirring nearby planetesimals and enhancing dust generation .
this process has been modeled in detail by @xcite . both this effect and the presence of a giant planet modeled above produce inner ring variability on timescales of @xmath165 yr .
our favored model of the scattered - light disk is shown in figure [ comparison ] .
this model assumes an initial disk size of 400 au , @xmath150 , and gas with with mass @xmath94 and an opening angle of @xmath157 .
the model reproduces the large - scale spiral arms and , despite a large value of @xmath2 , produces localized structures .
it does not reproduce the sharp and coherent rings or disk asymmetries of the disk around hd 141569a , as observed by c03 , which suggests that a different encounter geometry should be considered .
the model includes a @xmath5 planet in an eccentric orbit ( @xmath161 ) with apastron 160 au for the purpose of clearing out a region of planetesimals close to the central star .
the characteristics of this planet are not tightly constrained by the model , and other configurations ( including multiple planets ) are possible .
the simulations clearly indicate that a mechanism capable of clearing the inside of the disk ( like a planet or planets ) is necessary to explain the observed density profile . there may be additional observational evidence supporting the existence of a planet .
@xcite have detected h@xmath166 within 7 au of hd 141569a , which they suggest may come from the extended atmosphere from a giant protoplanet .
although there is favorable evidence for the existence of planet(s ) , we should also consider the effect of ice condensation on the surface density of the dust . an abrupt change in the surface density is expected upon crossing the `` snow line , '' which for hd 141569a s disk is @xmath21 au ( ll03 ) .
the location of the snow line depends strongly on grain size , so it likely varies with observed wavelength .
current coronagraphic images lack the spectral resolution needed to probe these variations . while crossing the snow line can not explain the paucity of dust close to the star , it likely affects the shape of the observed surface density .
most of the gas simulations do a similar job reproducing the optical depth images and profiles .
it would seem that the only hard constraint that they provide is that the presence of gas is necessary .
the quadratic dust generation method does a somewhat better job than the linear one in matching the density profiles .
interestingly , figure [ conclusion ] shows that the planetesimal distribution itself can also match the outer surface density profile .
the surface density of planetesimals starts as @xmath82 , but the truncation by the encounter makes it steeper by the time the simulations are examined .
the surface density of dust generated quadratically also matches the distribution of planetesimals : while the dust is generated as the square of the planetesimal distribution , it is blown outward by radiation pressure .
this process tends to flatten the simulated profile .
the applicability of our models to the debris disk around hd 141569a depends on the premise that the observed dust grains are small enough so that @xmath167 .
this condition is borne out by the disk colors , the mean - scattering phase , and thermal models .
could larger particles ( @xmath168 ) be contributing to the observed scattered light image ? in deriving the scattering - opacity function ( figure [ draine ] ) , we assumed that the collisional equilibrium timescale is smaller than any other systemic timescale , which implies that the number of grains of size @xmath40 decreases as @xmath27 .
if small grains were more quickly removed from their points of creation , then an excess of larger grains would result because they would no longer suffer erosion from collisions with the smaller grains @xcite . in the absence of gas ,
@xmath168 grains are eliminated by pr drag on timescales orders of magnitude larger than the rp timescales @xcite , so the proportion of larger grains would further increase . however , in the presence of modest amounts of gas , @xmath168 grains migrate to stationary orbits ( ta01 ) , and small particles settle down at the outer edge of the disk .
the timescale over which this migration occurs is quite small for small grains , but larger than the purely dynamical timescales associated with @xmath52 motions .
all these processes may alter the shape of the dust size distribution . if , for example , the distribution is flatter than @xmath27 , then the peak of the scattering - opacity function is broadened . while the optical image of the disk may still be dominated by @xmath2-meteoroids , the contributions of larger
, bound grains may be significant .
the entangled roles of rp , gas drag , and the dynamical role of the companions greatly complicate any effort to model all components of the disk .
to do so successfully requires multi - wavelength images of the disk .
observations by the _ spitzer space telescope _ should soon help this situation , as _ spitzer _ will observe larger grains that are less affected by rp .
future observations with the atacama large millimeter array ( alma ) will also contribute valuable information about the composition and distribution of the dust .
we have developed dynamical models to investigate the structures observed in the debris disk around hd 141569a .
the models include , for the first time , the effects of radiation pressure and gas drag .
the disk s colors and scattering phase indicate that the scattered - light images are produced by submicron - sized dust grains . to understand these images
, we must consider the behavior of grains with values of @xmath2 ( the ratio of radiation and gravitational forces ) larger than one .
these grains are expelled from the disk in shorter - than - dynamical timescales .
we show that up to @xmath0 of gas may be present in the disk .
gas drag slows the blow - out of grains and is fundamental to understanding the structure of the disk .
if the dust is continuously created , its steady - state surface density is featureless and has a radial profile dependent upon the amount of gas and the prescription for dust creation . to produce a broad ring of dust
, its creation must be prevented within some inner radial limit ( figure [ 1d_model2 ] ) . in principle , one can use the observed surface - density profile to estimate the amount of mass lost from by the disk throughout its history .
however , because large grains remain in the disk longer than smaller grains , this estimation implies an implausible amount of solid material in the disk .
consequently , either the observed rate of dust generation is limited to times near the encounter of the disk with hd 141569a s binary companions , or small grains are created highly preferentially over larger ones .
inspired by disk s large - scale spiral arms , we explored the effect on the disk from a parabolic fly - by of the binary companions .
such an encounter is consistent with the measured proper motion of the system .
our models consider the dynamical evolution of three different populations : planetesimals ( subject only to gravity ) , gas ( subject to gravity and hydrodynamic forces ) , and dust ( subject to gravity , radiation pressure and gas drag ) .
the poynting - robertson effect , self - gravity , and the dynamical effects of collisions are ignored .
we vary @xmath2 , the initial gas mass , the gas disk opening angle , and the prescription for dust generation .
we find that the dust distribution resembles that of the contemporary planetesimal distribution .
the models are most sensitive to the gas temperature , with hotter gas producing more disorganized structures than colder gas .
five - fold differences in the gas mass and two - fold differences in @xmath2 do not produce significant differences in the simulated disks .
quadratic dust generation produces more disorganized and less distinct structures than linear dust generation .
smaller initial disks produce smaller and less pronounced spiral arms .
our favored model ( figure [ comparison ] ) has an initial disk size of 400 au , @xmath150 , and gas with with mass @xmath94 and an opening angle of @xmath157 .
a planet is introduced to reduce the surface density of the planetesimals close to the central star , but the characteristics of this planet are not tightly constrained .
the model successfully reproduces the large - scale structures seen in the scattered - light images of the disk .
however , the model does not reproduce the tightly - wound spiral rings or other observed asymmetries , which suggests that another population of grains ( perhaps with smaller @xmath2 ) or a non - coplanar encounter geometry should be investigated .
the authors thank b. zuckerman , t. forveille , and j. kastner for generously providing the data from their 1995 paper .
we are grateful to k. anderson , w. j. mccann , s. busching , a. framarini , and t. allen for their invaluable contributions to the acs project at jhu .
acs was developed under nasa contract nas 5 - 32865 , and this research has been supported by nasa grant nag5 - 7697 .
we are grateful for an equipment grant from sun microsystems , inc .
the space telescope science institute is operated by aura , inc .
, under nasa contract nas5 - 26555 .
furthermore , we gratefully acknowledge support from nasa origins of solar systems grants nag5 - 10732 and nng04gg50 g . finally we wish to thank the anonymous referee whose comments greatly improved the paper . | we study the dynamical origin of the structures observed in the scattered - light images of the resolved debris disk around hd 141569a .
the disk has two conspicuous spiral rings and two large - scale spiral arms .
we explore the roles of radiation pressure from the central star , gas drag from the gas disk , and the tidal forces from two nearby stars in creating and maintaining these structures .
the disk s color , scattering function , and infrared emission suggest that submicron - sized grains dominate the dust population observed in scattered light .
co observations indicate the presence of up to @xmath0 of gas .
the dust grains are subject to the competing effects of expulsive radiation pressure ( @xmath1 , where @xmath2 is the ratio of the radiation and gravitational forces ) and retentive gas drag .
we use a simple one - dimensional axisymmetric model to show that the presence of the gas helps confine the dust and that a broad ring of dust is produced if a central hole exists in the disk .
this model also suggests that the disk is in a transient , excited dynamical state , as the observed dust creation rate applied over the age of the star is inconsistent with submillimeter mass measurements .
we model in two dimensions the effects of a fly - by encounter between the disk and a binary star in a prograde , parabolic , coplanar orbit .
we track the spatial distribution of the disk s gas , planetesimals , and dust .
we conclude that the surface density distribution reflects the planetesimal distribution for a wide range of parameters .
our most viable model features a disk of initial radius 400 au , a gas mass of @xmath3 , and @xmath4 and suggests that the system is being observed within 4000 yr of the fly - by periastron .
the model reproduces some features of hd 141569a s disk , such as a broad single ring and large spiral arms , but it does not reproduce the observed multiple spiral rings or disk asymmetries nor the observed clearing in the inner disk . for the latter , we consider the effect of a @xmath5 planet in an eccentric orbit on the planetesimal distribution of hd 141569a . |
modern society features an increasing degree of interaction between cultures ( `` cultural contact '' ) owing to , e.g. , communication technologies , immigration and other socio - political forces @xcite . in many countries cultural contact
is perceived as both an opportunity and a threat and is related to important public issues such as immigration management and the need to `` protect '' national culture @xcite .
our understanding of these phenomena is , however , limited : we can not predict the outcome of cultural contact , nor make plausible conjectures that can be used in policy making . within this context ,
the aim of this paper is twofold : we first describe a general mathematical framework for modeling social interactions , then we make specific assumptions relevant to studying immigration , i.e. , social contact between two groups that , typically , differ both in culture and relative size .
for simplicity , we focus on a single _ cultural trait _ , which may represent an idea , an opinion or a behavior that has two mutually exclusive forms .
a useful example to keep in mind is being in favor or against an issue such as the death penalty , or any other issue that might be the subject of a yes / no referendum vote .
our framework allows to consider multiple traits without conceptual differences , although model analysis may in general be much more difficult .
we consider a population of @xmath0 individuals , labeled by an index @xmath1 ranging from @xmath2 to @xmath0 .
we associate to the @xmath1-th individual the variable @xmath3 , which may take values @xmath4 or @xmath5 representing the two possible trait values .
for instance , @xmath4 might represent a yes vote in a referendum , and @xmath5 a no vote .
the state of the whole population is thus encoded in an array @xmath6 of @xmath0 numbers , such as @xmath7 .
the hallmark of social interactions is that individuals may change their opinions or behavior owing to interactions with others .
a given couple @xmath8 can in principle be in one of the four states @xmath9 , @xmath10 , @xmath11 and @xmath12 , but these outcomes , in general , do not have the same probability .
which one is more likely will depend on the characteristics of individuals such as their culture and personality .
our starting assumption is that individuals have no particular bias towards @xmath4 or @xmath5 opinions : what matters most is whether , by adopting one or the other value , an individual is in agreement or disagreement with others .
there are two reasons for this assumptions .
first , social psychologists have shown that , in most cultures , agreement or disagreement with others is a powerful determinant of individual opinions and behavior , often more important than holding a particular opinion @xcite ; we will expand on this point in our model of immigration below .
second , our framework allows to introduce biases that favor a particular trait value , if needed .
indeed , any model in which individuals are biased can be recast as a model with unbiased individuals , plus an additional `` force '' that orients individual opinions .
thus our starting assumption of unbiased individuals does not reduce the generality of the framework .
again , we will make a specific example for the case of immigration below . to formalize these notions
, we assume that individuals take on the trait that minimizes a _
cost function_. we define the cost @xmath13 for individual @xmath1 to agree or disagree with individual @xmath14 as @xmath15 where @xmath16 is a number that summarizes the nature of the interaction between @xmath1 and @xmath14 , as follows .
when @xmath1 and @xmath14 agree ( @xmath17 ) we have a cost @xmath18 , while when @xmath1 and @xmath14 disagree ( @xmath19 ) we have @xmath20 .
thus whether agreement or disagreement carries the lesser cost depends on the sign of @xmath16 : @xmath21 favors agreement while @xmath22 favors disagreement .
the magnitude of @xmath16 gives how important it is for @xmath1 to agree or disagree with @xmath23 .
if , for instance , @xmath24 then it is more important for @xmath1 to agree with @xmath14 than with @xmath25 , while @xmath26 means that agreement with @xmath14 is not relevant to @xmath1 .
the signs and magnitudes of the @xmath16 s become important when we consider that an individual interacts with many others . in this case , we assume that the costs @xmath13 relative to each interaction sum up to a total cost @xmath27 @xmath28 as anticipated above , we can take into account additional factors that may influence individuals modifying equation ( [ eq : hi ] ) as follows : @xmath29 meaning that individual @xmath1 is subject to an additional `` force '' that favors @xmath30 if @xmath31 and @xmath32 if @xmath33 .
the quantity @xmath34 may represent any factor that is not explicitly taken into account by the direct interaction with other individuals .
for instance , it may summarize the influence of media , government campaigns or existing culture ( see below ) .
we can now write a population - level cost function as the sum of individual cost functions : @xmath35 we stress that the cost function is a theoretical computational aid to track which trait values are favored by the interactions @xmath16 and the external forces @xmath34 .
we do not assume that individuals explicitly compute or are aware of such costs .
rather , @xmath36 should be designed so that its minima correspond to those trait values that are favored by the mechanisms with which individuals interact .
once a cost function has been specified , it is possible to calculate population level quantities such as the average trait value using the methods of statistical mechanics , a branch of theoretical physics .
statistical mechanics was originally concerned with deriving the laws of thermodynamics from the behaviour of atoms and molecules @xcite , but can actually be applied to understand the macroscopic ( population level ) properties of any system composed of many parts that interact according to given microscopic ( individual level ) rules .
more recently its methods have found application in fields as diverse as biology @xcite , neuroscience @xcite , economy and finance @xcite and also social science @xcite .
the starting point is to assign to each system configuration @xmath6 a probability @xmath37 according to the boltzmann - gibbs formula @xcite @xmath38 where the sum runs over all possible configurations of the system . by means of ( [ eq : bg ] ) a given configuration
is considered more or less likely according to whether it is more or less costly : a low value of @xmath36 results in a high probability of @xmath6 , and vice - versa .
assigning probabilities based on a given cost function is the heart of statistical mechanics and is inspired by the principles of thermodynamics ( see the appendix for a short discussion , and , for a fuller treatment ) . once a probability is assigned to system configurations , it is possible to compute the expected values of quantities of interest and to relate them to the parameters that describe the system .
for instance the average cultural trait defined by @xmath39 would have an expected value given by @xmath40 note that , while @xmath41 is the average trait value in a given configuration , @xmath42 is the average trait value over all possible system configurations , each one weighted according to its probability of occurrence .
these probabilities depend on the cost function @xmath43 and thus on the parameters that appear in its expression , i.e. , the @xmath16 s and @xmath34 s . rather than directly attempting to compute
expected values such as ( [ eq : av ] ) , statistical mechanics aims to compute the so - called _ free energy _ of a system , defined as @xmath44 the rationale for this strategy is that important quantities such as ( [ eq : av ] ) can be easily computed from knowledge of the free energy function , typically by taking suitable derivatives with respect to system parameters ( see appendix ) .
the basic task of statistical mechanics is thus , after the cost function @xmath43 has been specified , to calculate the free energy function for a given system .
we stress that _ the form of the cost function is not given by statistical mechanics ; rather , it is the outcome of a modeling effort relative to a specific problem .
_ we now make an example of how one may proceed .
we illustrate here the potentials of our framework considering the impact of immigration on culture .
we consider a large and a small population , which will be referred to , respectively , as _ residents _ ( @xmath45 ) and _ immigrants _ ( @xmath46 ) .
we let @xmath47 be the number of residents , and @xmath48 of immigrants , with @xmath49 and @xmath50 the total number of individuals .
we are interested in how cultural contact changes the average trait values in the two populations , with the aim of understanding the effect of one culture upon the other .
our main assumption regarding how residents and immigrants interact is that people , generally speaking , tend to agree with those who are perceived as similar to oneself and to disagree with those perceived as different . in social psychology
this is known as the _ similarity - attraction _ hypothesis .
it has received ample support , although the details of how we interact with others often depend on social context @xcite .
we consider this assumption a general guideline , and in modeling a specific case it can be modified without difficulty .
we formalize the similarity - attraction hypothesis by assuming that high perceived similarity corresponds to positive values of @xmath16 , and low perceived similarity to negative values .
since residents and immigrants have generally different cultures , we may assume the following structure for the interaction coefficients @xmath16 .
we let the interaction between any two residents be @xmath51 ; the similarity - attraction hypothesis suggests that this be a positive number , whose magnitude reflects how strongly residents prefer to agree among themselves .
likewise , we let the interactions between immigrants be @xmath52 .
the mutual interactions @xmath53 and @xmath54 should model whether residents prefer to agree or disagree with immigrants , and vice - versa , and how strongly so .
if resident and immigrant cultures are very different , the similarity - attraction hypothesis suggests to take both @xmath53 and @xmath54 as negative , but the best choice of values depends on the specific case one intends to model .
note that we are assuming that @xmath16 depends only on population membership and not on the particular individuals @xmath1 and @xmath14 ( the so - called _ mean field _ assumption in statistical mechanics ) .
this assumption greatly simplifies mathematical analysis but is not wholly realistic .
it can capture the average variation in interactions across population but not the variation that exists within each population .
for instance , a more realistic assumption would be to take the @xmath16 s as random variables whose mean and variance depend on population membership .
we plan to return on that model ( which would represent the two - population generalization of the sherrington - kirkpatrick model in statistical mechanics , ) in future studies .
when modeling interactions , a technical requirement is that the value of the cost function be proportional to total population size @xmath0 .
this guarantees that the free - energy function and important quantities such as average trait value , equation ( [ eq : av ] ) , scale appropriately with @xmath0 . in our case
the appropriate scaling is @xmath55 , hence the interactions are : @xmath56 before the two populations start to interact , residents and immigrants are each characterized by a given average trait value , say @xmath57 and @xmath58 , respectively .
we consider @xmath59 and @xmath58 as experimental data about the beliefs or behavior of each population , which could be obtained from , say , a referendum vote on a particular issue ( e.g. , the death penalty ) or from statistical sampling of the population . that a population is characterized by a given average value @xmath60 means that the two forms of the trait are not equally common .
specifically , the individuals with the @xmath4 form are @xmath61 , while @xmath62 individuals have the @xmath5 form .
pre - existing culture , in other words , is like a bias or force that favors one trait value over the other .
for modeling purposes , it is convenient to describe pre - existing culture as a `` cultural force '' that acts to orient the opinion of otherwise unbiased individuals
. this is possible including a force term in the cost function , as shown in ( [ eq : hh ] ) . by standard methods of statistical mechanics ( see appendix ) it is possible to show that the force term corresponding to a particular average opinion @xmath63 is @xmath64 where @xmath65 is the inverse hyperbolic tangent function . to summarize , a model in which individuals are biased so that the average opinion is @xmath63 is equivalent to a model with unbiased individuals subject to a force given by ( [ eq : h ] ) .
so far we have specified interaction terms @xmath16 to model cultural contact between two populations and we have introduced equation ( [ eq : h ] ) to represent the pre - existing culture in the two populations .
the next step is to compute the average trait values @xmath66 and @xmath67 in the two populations after immigration has taken place . the same method that allows to derive equation ( [ eq : h ] ) enables us to derive the following equations for @xmath66 and @xmath67 ( see appendix ) : [ eq : m1m2 ] @xmath68 @xmath69 where @xmath70 is the fraction of immigrants in the total population and @xmath71 is the hyperbolic tangent function .
the values of @xmath66 and @xmath67 predicted by ( [ eq : m1m2 ] ) depend of course on values of the @xmath72 and @xmath73 parameters , and on @xmath74 .
we give here a qualitative description of the different regimes that one can observe varying these parameters .
we refer to for a proof of the following statements , in the context of an analogous model from condensed matter physics .
the two key parameters are @xmath74 , the fraction of immigrants , and the overall scale of the interactions @xmath16 , which we label @xmath72 .
if @xmath72 is below a critical value @xmath75 , equation ( [ eq : m1m2 ] ) has always one pair of solutions , for all values of @xmath74 . in this case
the two populations are essentially merging into a homogeneous one , with average cultural trait in between the two initial ones more toward one or the other according to the value of @xmath76 .
this regime is not surprising and corresponds to the nave prediction that one could have made a priori without applying statistical mechanics .
if the interaction scale is large ( @xmath77 ) , however , model predictions are highly non - trivial , suggesting that the outcome of cultural contact can be surprising . depending on @xmath72 there are two critical values for @xmath76 : @xmath78 and @xmath79 that delimit qualitatively different behavior . for @xmath80
the resident culture dominates dominant and the immigrant culture disappears , i.e. , @xmath67 is close to @xmath66 irrespective of the initial value @xmath81 .
the converse happens when @xmath82 , i.e. , immigrant culture dominates .
the most interesting case occurs when @xmath83 . in this regime ( [ eq : m1m2 ] ) has two distinct solutions in which either of the two cultures dominates .
that is , both cultures may survive the immigration process , generally with a different probability determined by system parameters .
the parameter values that favor the resident or immigrant culture , have still to be worked out and will be the topic of future work . here
we analyze the case in which the intensity of the interactions is uniform both within and between groups , @xmath84 .
this is interpreted as two groups that do not really discriminate between themselves , so that disagreement with any particular individual carries the same cost independent of which group the individual belongs to .
we assume , however , that the two groups have initially a very different average trait value : @xmath85 and @xmath86 . in figure
[ fig : jalpha ] we explore this system by plotting the average trait value after the interaction , @xmath87 , for @xmath74 between 0 and 10% and for @xmath88 . for @xmath89 ( no interaction )
@xmath25 is simply the weighted sum of pre - existing trait values , @xmath90 , where each group contributes according to its size . as a function of @xmath74
, this is a straight line .
as the interaction increases the line slowly bends and for higher values of @xmath76 we see a slight exaggeration of the pre - existing opinion @xmath91 ( the surface in figure [ fig : jalpha ] rises slightly over the level @xmath92 ) .
when @xmath72 crosses a critical value @xmath93 , however , a dramatic phenomenon occurs : the population undergoes a sudden change of opinion , switching to a value of @xmath25 that is closer to , and indeed exaggerates , the initial value in the immigrant population , @xmath94 .
note that this sudden transition occurs for all values of @xmath76 , i.e. , irrespective of the proportion of immigrants .
the solution with @xmath25 closer to @xmath66 is still available to the system ( not plotted in figure[fig : jalpha ] ) , but as @xmath72 grows past @xmath75 it is less and less likely that the system remains in such state ( technically , for @xmath95 the solution with @xmath96 has a higher free - energy than the solution with @xmath97 and thus becomes metastable , allowing fluctuations to cause a transition between the two solutions ) .
thus , according to this model , to prevent dramatic changes in resident culture , it would do little to restrict immigration ( the effect of @xmath74 is small in the graph ) .
rather , one should concentrate in reducing the scale of the interaction @xmath72 , i.e. , the strength of attitudes within and between groups .
attempts to apply mathematical - physics methods to social sciences have appeared in the litterature since the pioneering work of . in this paper
we have focused on statistical mechanics as a tool to bridge the gap from individual - level psychology and behavior to population - level outcomes .
our framework prescribes that researchers build a cost function that embodies knowledge of what trait values ( opinions , behaviors , etc . )
are favored by individual interactions under given social conditions . the cost function , equation ( [ eq : h ] ) ,
is defined by a choice for the interactions @xmath16 and the fields @xmath34 that represent social forces influencing individual opinions and behavior .
this modeling effort , of course , requires specific knowledge of the social issue to be modeled .
after a cost function has been specified , the machinery of statistical mechanics can be used to compute population - level quantities and study how they depend on system parameters .
we have demonstrated our framework with an attempt to understand the possible outcomes of contact between two cultures .
even the simple case we studied in some detail the model suggests that cultural contact may have dramatic outcomes ( figure [ fig : jalpha ] ) . how to tailor our framework to specific cases , and what scenarios such models predict , is open to future research .
we thank f. guerra for very important suggestions .
i. gallo , c. giardina , s. graffi and g. menconi are acknowledged for useful discussion .
amit , d. : 1989 , _ modeling brain function _ ,
vol . 1 .
cambridge : cambridge university press .
arbib , m. a. : 2003 , _ the handbook of brain theory and neural networks_. mit press , 2 edition .
bond , r. , smith , p. b. : 1996 , ` culture and conformity : a meta - analysis of studies using asch s ( 1952b,1956 ) line judgment task ' .
, 111137 .
bouchaud , p. , potters , m. : 2000 , _ theory of financial risks_. cambridge university press .
byrne , d. : 1997 , ` an overview ( and underview ) of research and theory within the attraction paradigm ' .
, 417431 .
cornelius , w. a. , p. l. martin , and j. f. hollifield ( eds . ) : 1994 , _ controlling immigration : a global perspective_. stanford , ca : stanford university press .
durlauf , s. n. : 1999 , ` how can statistical mechanics contribute to social science ? ' . , 1058210584 .
galam , s. , gefen , y. , shapir , y. : 1982 , ` sociophysics : a mean field model for the process of strike ' .
, 113 .
givens , t. and a. luedtke : 2005 , ` european immigration policies in comparative persepctive : issue salience , partisanship and immigrant rights ' .
, 122 .
cohen , e.g.d . : 1973 , _ tricritical points in metamagnets and helium mixtures_. in _ fundamental problems in statistical mechanics _ , proceedings of the 1974 wageningen summer school .
north - holland / american elsevier .
lull , j. : 2000 , _ media , communication , culture_. cambridge , uk : polity press .
mezard , m. , g. parisi , and m. a. virasoro : 1987 , _ spin glass theory and beyond_. singapore : world scientific .
michinov , e. and j .- m .
monteil : 2002 , ` the similarity - attraction relationship revisited : divergence between affective and behavioral facets of attraction ' .
, 485500 .
thompson , c. : 1979 , _ mathematical statistical mechanics_. princeton , nj : princeton university press .
it is a standard result of statistical mechanics @xcite that the free energy function of a system defined by a cost function of the form @xmath98 is obtained for the value of @xmath25 that minimizes the function @xmath99 the minimization of this function with respect to @xmath25 yields the condition ( [ eq : h ] ) which relates @xmath25 and @xmath73 and the hamiltonian parameters .
the structure of the free energy ( [ eq : fcw ] ) admits the standard statistical mechanics interpretation as a sum of two contributions : the internal energy ( the average of the cost function ) @xmath100 minus the entropy @xmath101 one can indeed show that the distribution function ( [ eq : bg ] ) may be deduced from the second principle of thermodynamics i.e. as the distribution for which the entropy is minimum at an assigned value of the cost function @xcite . equation ( [ eq : m1m2 ] )
is obtained similarly from the representation of the free energy of the two population system as the minimum of the function @xmath102\\ & + { \alpha}[+\frac{1+m_2}{2}\log\frac{1+m_2}{2}+\frac{1-m_2}{2}\log\frac{1-m_2}{2 } ] \end{array}\ ] ] the minimum condition yields ( [ eq : m1m2 ] ) . | we introduce a general modeling framework to predict the outcomes , at the population level , of individual psychology and behavior .
the framework prescribes that researchers build a cost function that embodies knowledge of what trait values ( opinions , behaviors , etc . )
are favored by individual interactions under given social conditions .
predictions at the population level are then drawn using methods from statistical mechanics , a branch of theoretical physics born to link the microscopic and macroscopic behavior of physical systems .
we demonstrate our approach building a model of cultural contact between two cultures ( e.g. , immigration ) , showing that it is possible to make predictions about how contact changes the two cultures . |
the aharonov - bohm effect leads to a number of remarkable interference phenomena in mesoscopic systems , especially in rings @xcite .
based on the discovery of the geometric phases @xcite , including the adiabatic berry phase @xcite and the nonadiabatic aharonov - anandan ( aa ) phase @xcite , it has been predicted that analogous interference phenomena can be induced by the geometric phases which originate from the interplay between electrons orbital and spin degrees of freedom .
such interplay can be produced by external electric and magnetic fields , which lead to zeeman coupling and spin - orbit ( so ) interaction respectively .
loss _ et al . _ first studied the textured ring embedded in inhomogeneous magnetic field @xcite .
they found the inhomogeneity of the field results in a berry phase , which can produce the persistent currents .
the effects of this berry phase on conductivity were then discussed @xcite .
it was further pointed out that the adiabatic condition is not necessary for the geometric phase to exist , and the aa phase in textured rings can produce the persistent currents as well @xcite . on the other hand ,
the aharonov - casher ( ac ) effect @xcite in mesoscopic systems has attracted much attention .
et al . _ showed for the first time that so interaction in one - dimensional ( 1d ) rings results in an effective magnetic flux @xcite .
mathur and stone then pointed out that observable phenomena induced by so interaction are the manifestations of the ac effect in electronic systems @xcite .
these authors investigated the effects of so interaction on the persistent - current paramagnetism and the quantum transport in disordered systems , and obtained specific reduction factors for harmonics in ab oscillations @xcite . in case
the ac flux is not random , it can lead to interference phenomena as ab flux .
mathur and stone proposed an observation of the ac oscillation of the conductance on semiconductor samples @xcite .
balatsky and altshuler @xcite and choi @xcite studied the persistent currents produced by the ac effect .
inspired by the study on textured rings , the ac effect has also been analyzed in connection with the spin geometric phase .
aronov and lyanda - geller considered the spin evolution in conducting rings , and found that so interaction results in a spin - orbit berry phase which plays an interesting role in the transmission probability of the rings @xcite . in their models , there is a zeeman coupling from uniform magnetic field , but the so berry phase can be caused by so interaction alone .
so they has indeed shown the existence of the berry phase in the ac effect . since so interaction is usually not strong enough to guarantee the validity of adiabatic approximation , a nonadiabatic treatment of the problem is necessary . in ref .
@xcite , we demonstrated the existence of a nonadiabatic aa phase in the ac effect in 1d rings .
we found the ac flux and local spin orientations of the electronic eigenstates are determined by a spin cyclic evolution .
in particular , we showed the ac phase comprises both the aa and the dynamical phases which are acquired in the cyclic evolution , and the adiabatic limit of the aa phase is just the so berry phase . based on this geometric phase approach for the ac effect , oh and ryu studied the persistent currents produced by the cylindrically symmetric so interaction in 1d rings @xcite . as is well known , so interaction is time - reversal invariant while zeeman coupling breaks the time - reversal symmetry ( trs ) .
many prior works have shown the significance of the trs and its breaking with regard to various interference phenomena caused by ab flux and so interaction .
it is therefore worthwhile to investigate if the coexistence of so interaction and zeeman coupling can produce any new observable effect with the spin degree of freedom being fully taken into account .
however , most of the previous studies have focused on the rings in the presence of zeeman coupling or so interaction only . in ref .
@xcite , we have demonstrated that the competition between zeeman coupling and so interaction can produce persistent currents through the trs breaking in a many - electron ring with a complete set of current - carrying single - particle states .
for the transport properties , aronov and lyanda - geller @xcite have derived a transmission probability for a conducting ring in the presence of both the so interaction and the zeeman coupling , by making use of the concept of berry phase .
unfortunately , they failed to take into account correctly the different properties of so interaction and zeeman coupling under the time - reversal transformation .
as a result , they have not realized that their picture of effective flux for the interference of spin - polarized electrons is actually invalidated by the trs - breaking zeeman coupling .
furthermore , even if the zeeman coupling is absent , their expression for the effective flux induced by the so interaction is still not complete .
so the transport properties of a ring in the presence of both so interaction and zeeman coupling have not been solved yet and the roles of trs and its breaking therein need to be clarified . in this paper , we will discuss the transport properties of a ring in the presence of both the so interaction and the zeeman coupling .
we will explore the roles of the trs and its breaking in the transport phenomena when the spin degree of freedom is taken into account explicitly .
we will also show the connection between the observable trs - breaking effects manifested by the persistent current and by the transmission probability . throughout the discussion
, we will emphasize the trs by investigating how the trs - breaking zeeman coupling affects the thermodynamic and transport properties of the system .
the paper is organized as follows . in sec .
ii , we first solve the spin cyclic evolution and find the corresponding geometric and dynamical phases for the system .
the electronic eigenstates of the closed ring are then derived . the geometric phase and the exact solution for the ring with only zeeman coupling @xcite or so interaction @xcite are shown to be the limit of zero electric field or the limit of zero magnetic field in our results .
on the other hand , the so berry phase , first introduced by aronov and lyanda - geller @xcite , is simply the adiabatic limit of the geometric aa phase here . in sec .
iii , we first derive an exact expression for the transfer matrices of the two ring branches ( arms ) by introducing four auxiliary spin states , which exhibit the orbital quantum number dependence of the spin orientations in electronic eigenstates .
then we calculate the transmission probability of the ring .
we show that the presence of zeeman coupling makes the spin orientations in energy eigenstates depend on the spin and the orbital quantum numbers simultaneously . as a consequence ,
the effective spin - dependent flux description , which has been established for so interaction only , becomes inadequate .
that explains why the results in ref .
@xcite are not correct
. when the zeeman coupling is absent , the derived transmission probability agrees with the relations obtained by meir _
et al . _ for general spin - independent thermodynamic and transport properties .
we finally carry out some numerical calculations to illustrate the effects of so interaction and zeeman coupling .
we find there is an interesting and observable correspondence between the trs - breaking effects manifested by the transmission probability and by the persistent current .
that correspondence , if experimentally verified or excluded in some specific ring , may serve to detect if the electron - electron interaction is of qualitative importance in determining electronic states . in sec .
iv , we conclude with a summary of our results .
the hamiltonian for an electron in the electric field @xmath0 and the magnetic field @xmath1 is @xmath2 we consider a ring that is effectively one - dimensional ( 1d ) and the fields which are cylindrically symmetric , i.e. , @xmath3 , @xmath4 in the cylindrical coordinate system . for the ring lying in the @xmath5 plane with its center at the origin ,
the hamiltonian is given by @xmath6^{2}+\displaystyle\frac{\hbar \omega_{b}}{2}(\sin\chi_{2}\sigma_{r}+\cos\chi_{2}\sigma_{z}),\ ] ] with @xmath7 , @xmath8 and @xmath9 , where @xmath10 is the ring radius , @xmath11 is the angular coordinate and @xmath12 is the enclosed magnetic flux in unit of flux quantum .
the eigenvalue equation of the system can be solved through a straightforward diagonalization , as presented in ref .
@xcite . here
we adopt the geometric phase approach @xcite , in order to identify the geometric and the dynamical phases in current - carrying eigenstates , which are responsible for transporting electrons when the ring is connected to current leads .
how the phases and the spin orientations jointly affect the transmission probability will be elaborated in the next section .
the cylindrical symmetry of the system leads to the conservation of total angular momentum @xmath13 , which means the eigenstates of hamiltonian ( 2 ) are of the form @xmath14 , in which @xmath15 , @xmath16 are arbitrary integers , and the spin states are given by @xmath17;\;\;\ ; \tilde{\psi}_{n,-}(\theta)=\left [ \begin{array}{c } \sin\displaystyle\frac{\beta_{n}}{2}\\ -e^{{\it i}\theta}\cos\displaystyle\frac{\beta_{n}}{2}\\ \end{array } \right ] , \ ] ] where @xmath18 is @xmath11-independent . from @xmath19 as a function of @xmath11 , it is readily seen that the local spin orientations at @xmath11 is in the direction of @xmath20 . the explicit expression for the spin tilt angle @xmath18
can be obtained by introducing a cyclic evolution of spin state for electrons encircling the ring , as presented in ref .
the geometric and the dynamical phases associated with the spin precession can thereby be identified for all of the energy eigenstates to determine the whole energy spectrum .
such an approach has the advantage of explicitly exhibiting the geometric phase and the adiabatic criterion , which acquire their original meanings in time - dependent problems .
the spin cyclic evolution is defined by a time - dependent schrdinger equation @xmath21 for a spin-@xmath22 particle in a time - varying magnetic field , where @xmath23 is given by @xmath24 \\ & + \displaystyle\frac{1}{2}\hbar\omega_{b}[\sin\chi_{2}\cos(\omega t)\sigma_{x}+ \sin\chi_{2}\sin(\omega t)\sigma_{y}+\cos\chi_{2}\sigma_{z } ] .
\end{array}\ ] ] > from the solution of the cyclic evolution governed by eq .
( 4 ) , we obtain for @xmath25 the spin tilt angle @xmath26 and the geometric and the dynamical phases @xmath27 and @xmath28 @xmath29 @xmath30,\ ] ] where @xmath31 is given by @xmath32 with @xmath33 . here
the geometric aa phase @xmath27 is the -1/2 of the solid angle subtended by a circuit traced on a sphere by the local spin orientation of @xmath25 .
it is readily seen that the zeeman coupling makes the spin orientations of electronic eigenstates depend on the orbital quantum number .
the consequence of such interplay between the spin and the orbital degrees of freedom will be explored when we discuss the transport properties of the ring . with use of @xmath18 , @xmath27 , and @xmath28 , the eigenvalues @xmath34 of @xmath25
is found to be @xmath35 the first term in the right side represents the energy from orbital motion , the second term the zero - point energy , while the third term comes from the spin precession originating from the interplay between the spin and the orbital degrees of freedom .
the exact solution derived above reduces to the various limits that have been obtained separately in literatures .
the cylindrically symmetric textured ring , first studied under the adiabatic approximation @xcite and then exactly solved in refs .
@xcite , corresponds to the @xmath36 limit .
the ac effect induced by cylindrically symmetric so interaction in the ring , first discussed for vertical field @xcite and then investigated for more general field configurations@xcite , corresponds to the @xmath37 limit . on the other hand , with both nonzero @xmath38 and @xmath39 , the so berry phase , first introduced for a conducting ring @xcite , is simply the adiabatic limit of the aa phase .
now we turn to the adiabatic limit of the exact solution .
since the original stationary schr " odinger equation is solved via the solution of the time - dependent problem , the adiabatic criterion can be easily deduced . comparing @xmath18 with the tilt angle of the effective magnetic field in @xmath23
, we find the adiabatic criterion is @xmath40 for the textured ring with @xmath41 , this condition states that the zeeman frequency @xmath42 must be much larger than the orbital frequency @xmath31 unless @xmath43 @xcite . for the ac effect with @xmath44
, this condition requires that the dimensionless so coefficient @xmath45 must be much larger than @xmath46 unless @xmath47 @xcite .
in this section , we discuss the transport properties of the ring described by hamiltonian ( 2 ) . now
the ring is connected to external current leads , schematicly illustrated in fig .
1 . we adopt the standard formulation developed in the study of quantum oscillations in 1d rings threaded by ab flux @xcite . in the upper and the lower branches ,
the wave amplitudes at one end are related to the wave amplitudes at the other end by the transfer matrices as @xmath48=\underline { t}_{i}\left[\begin{array}{c}\beta_{1 } ' \\ \beta_{1}\end{array}\right],\;\;\ ; \left[\begin{array}{c}\gamma_{1 } \\ \gamma_{1}'\end{array}\right]=\underline { t}_{ii}'\left[\begin{array}{c}\gamma_{2 } ' \\ \gamma_{2}\end{array}\right],\ ] ] where @xmath49 and @xmath50 denote the transfer matrices of the upper and lower branches respectively , and they depend on the energy @xmath38 of the incident wave . at the two junctions , the amplitudes of the three outgoing waves @xmath51
are related to the amplitudes of the incoming waves @xmath52 by @xmath53= \left[\begin{array}{ccc } -(a+b ) & \sqrt{\epsilon } & \sqrt{\epsilon } \\ \sqrt{\epsilon } & a & b \\ \sqrt{\epsilon } & b & a \end { array}\right ] \left[\begin{array}{c}\alpha \\ \beta \\ \gamma \end{array}\right],\ ] ] where @xmath54 and @xmath55 with @xmath56 .
when considering a wave incident from the right junction , we have @xmath57 and @xmath58 .
the amplitude of the transmitted wave is @xmath59\otimes\sigma_{0 } ) \underline{t}_{i } \pi^{-1}(\left[\begin{array}{c}b - a \\
-1 \end{array}\right]\otimes\sigma_{0})\alpha_{1},\ ] ] with @xmath60 given by @xmath61\otimes\sigma_{0})\underline{t}_{ii } ' ( \left[\begin{array}{cc } b^{2}-a^{2 } & a \\
-a & 1 \end{array}\right]\otimes \sigma_{0})\underline{t}_{i}-1\!\ ! 1,\ ] ] where @xmath62 is the @xmath63 unit matrix in spin space .
this formulation is in general applicable to the derivation of the transmission probability through any ring , provided the corresponding transfer matrices are known .
note that in the study of the ring only threaded by ab flux , electrons can be treated as spinless particles , so that all amplitudes are simply represented by complex numbers and matrix @xmath62 can be dropped . in this paper , @xmath64 , @xmath65 ,
@xmath66 have to be represented by two - component spinors and @xmath67 , @xmath68 are @xmath69 matrices .
to derive an explicit expression for the two transfer matrices , we first identify the electronic states in the ring by making use of its cylindrical symmetry . if we write @xmath67 , @xmath68 in @xmath70 matrix form ,
then each matrix element is a @xmath70 matrix in spin space .
we can easily conclude that the off - diagonal elements of @xmath67 , @xmath68 are zero because of the conservation of @xmath71 , which indicates that in each branch any propagating wave with fixed energy can posses a well - defined momentum and pass each branch without reflection , as a result of the cylindrical symmetry of the external fields and the absence of scattering potential .
so our task reduces to finding the four @xmath63 matrices which respectively relate @xmath72 with @xmath73 , @xmath74 with @xmath75 , for the upper branch , and relate @xmath76 with @xmath77 , @xmath78 with @xmath79 , for the lower branch .
these four @xmath63 matrices are the four nonzero diagonal elements of @xmath67 , @xmath68 directly .
the electrons tunneling through the ring is carried out by the energy eigenstate of the ring connected to two ideal conductors .
consider an incident wave with wavevector @xmath80 .
the corresponding eigenenergy of the steady transport state is @xmath81 . in the right conductor
the electronic state is a superposition of the incident plane wave @xmath64 and the reflected plane wave @xmath65 , while in the left conductor the propagating wave is just the transmitted plane wave @xmath82 .
the state inside the ring is a superposition of four wavefunctions of energy @xmath83 .
they actually determine the four non - zero matrix elements defined above for the two diagonal transfer matrices . to find the four components of the electronic wave inside the ring , we first use the energy expression @xmath84 to find four solutions of @xmath16 , which are positive @xmath85 and negative @xmath86 with @xmath87 , and positive @xmath88 and negative @xmath89 with @xmath90 . for arbitrary @xmath80 ,
these quantum numbers are not integers in general . for each @xmath91
, we can obtain a wavefunction @xmath92 which bears the same form as @xmath25 of the closed ring , but with @xmath16 being substituted by @xmath91 and accordingly the spin tilt angle @xmath18 being substituted by @xmath93 from eq .
these four @xmath94 are actually eigenstates of the hamiltonian ( 2 ) at energy @xmath83 but the periodic boundary condition @xmath95 is resolved due to the connection with external conductors . the electronic wave inside the ring is a superposition of the four @xmath94 by which the eight amplitudes @xmath96 can be represented .
this is a natural conclusion from the steadiness of the electronic state which transports electrons at fixed energy @xmath83 through the ring . with this understanding
, we can derive the transfer matrices in terms of @xmath94 . as shown in sec .
ii , the zeeman coupling brings the dependence on orbital quantum number to spin orientations . as a result , @xmath97 and @xmath98 , which carry the clockwise ( @xmath99 ) or the anticlockwise @xmath100 wave , are of nonorthogonal spin states @xmath101 and @xmath102 unless in the absence of zeeman coupling . to derive the transfer matrix associated with spin - polarized transport ,
it is crucial to distinguish the @xmath103 from the @xmath90 contribution for any wave propagating in fixed direction . for this purpose , we define four auxiliary spin states @xmath104 where @xmath105 .
it is easy to verify the relations of redefined orthogonality and completeness , @xmath106 and @xmath107 in the upper branch , the wave propagating anticlockwisely consists of the two components @xmath108 and @xmath109 .
@xmath72 and @xmath73 can thereby be expressed as @xmath110 where @xmath111 and @xmath112 are two specific constants .
using eqs .
( 3 ) and ( 15 ) , we obtain @xmath113 \beta_{1}^{'},\ ] ] and therefore find the @xmath63 matrix which is the first diagonal element of @xmath67 .
the other three matrix elements in diagonal @xmath67 and @xmath114 can be derived in the same way .
we finally obtain the two transfer matrices in the form of @xmath115;\ ] ] @xmath116.\ ] ] > from eq . ( 11 ) , the transmission probability for unpolarized incident electrons is @xmath117 in which @xmath118 denotes an averaging over @xmath64 with fixed @xmath119 .
explicitly , it is given by @xmath120\otimes\sigma_{0 } ) \underline{t}_{i } \pi^{-1}(\left[\begin{array}{c}b - a \\
-1 \end{array}\right]\otimes\sigma_{0})]_{ij}\right| ^2,\ ] ] in which @xmath67 , @xmath68 , and @xmath60 are all known . in the absence of zeeman coupling ,
the expression of the transmission probability can be greatly simplified and be explicitly related to the spin - independent transmission probability through the ring threaded by ab flux only .
> from eq .
( 6 ) , it is obvious that if @xmath44 , @xmath121 are independent of @xmath91 defined in eq . ( 13 ) and can be denoted by @xmath122 with @xmath123 . combining this fact with eqs .
( 15 ) , ( 16 ) , and ( 17 ) , we see that in the absence of zeeman coupling , the electronic wave in the ring actually consists of two orthogonal amplitudes , which propagate coherently and independently , with their local spin states being given by @xmath122 .
we then turn to the phase shift for spin - polarized electrons .
when @xmath124 is very large and quasiclassical approximation is therefore applicable , it is worthwhile to write @xmath91 in eq .
( 13 ) as @xmath125 where the last three terms in the right side are @xmath126 of the ab phase , the spin aa , and the dynamical phases contributed by the so interaction , respectively . in case
the zeeman coupling is absent , the last two terms give the @xmath127 of the ac phase , @xmath128 @xcite , and @xmath129 becomes an exact relation without quasiclassical approximation . since @xmath130 is @xmath16-independent , eq .
( 23 ) indicates that the effect of the so interaction can be regarded as an ab effect of the effective flux @xmath131 in unit of @xmath132 for the locally polarized electron gases with local spin states @xmath122 .
we can derive for @xmath44 the transmitted amplitude @xmath133 t(\phi-\displaystyle\frac{\phi_{ac}^{\mu}}{2\pi } ) \tilde{\psi}_{\mu}(\pi),\ ] ] where @xmath134 is the transmitted amplitude for the ring threaded by magnetic flux @xmath12 @xcite with vanishing so interaction . eq . (
24 ) indicates clearly that the real electronic wave in the ring is a superposition of the two locally polarized waves , which enclose different effective fluxes and propagate independently .
the transmission probability @xmath135 is given by @xmath136 : @xmath137 where @xmath138 is the transmission probability of the ring threaded by magnetic flux @xmath12 with vanishing so interaction .
to see what happens for unpolarized incident wave , we average @xmath135 over @xmath64 and obtain @xmath139 , which agrees with the relation predicted in ref .
@xcite for general spin - independent thermodynamic and transport quantities . in the competition with the so interaction
, the zeeman coupling brings the @xmath16-dependence to the spin orientations of energy eigenstates .
the @xmath16-dependent spin precession then results in the @xmath16-dependent spin phases .
it is seen that in the presence of the zeeman coupling , the last two terms in eq .
( 22 ) are @xmath16-dependent and the effect of the spin phases can no longer be regarded as that from the effective flux which must be independent of the specific orbital quantum numbers of the states .
it is thus quite clear that in the presence of zeeman coupling , we can not use 1 ) the identification of the two polarized wave amplitudes which are orthogonal to each other in spin space and propagate independently , and 2 ) the description that the phase effect from the spin degree of freedom is effectively some ab effect of a spin - dependent flux .
we want to point out that the above complexity due to zeeman coupling has not been recognized in ref .
consequently the transmission probability obtained therein was wrongly simplified by regarding the contribution from the spin degree of freedom as an ab effect of some @xmath140-dependent flux for polarized electrons .
we also want to point out that even when the zeeman coupling is absent and the picture of effective flux is applicable , only the geometric phase was included while the dynamical phase was ignored in the expression of the effective flux in ref .
@xcite .
numerical calculation has been carried out to illustrate some essential characteristics of the transmission probability derived here .
we find that the respective effects of zeeman coupling and so interaction can be reflected by the resonance of the transmission probability in the weak coupling limit at small @xmath141 .
in particular , we can see an interesting correspondence between the trs - breaking effects manifested by the transmission probability and by the persistent current .
we adopt the model of a inas ring @xcite .
the hamiltonian is of the form @xmath142_{z}-\displaystyle\frac{ge\hbar}{4mc } \sigma\!\!\!\!\sigma\cdot{\bf b},\ ] ] where @xmath143 is the effective mass , @xmath144 is the so coefficient and @xmath145 . here
the effective electric field is in the z - direction , hence @xmath146 . for the ring of radius @xmath147 , the dimensionless coefficient @xmath45 in eq .
( 2 ) is found to be @xmath148 which is large enough to result in an ac phase of order unity @xcite .
the fermi velocity @xmath149 is approximately @xmath150 , corresponding to @xmath151 .
the effective flux induced by so interaction and its effect on the transmission probability can be clearly seen in fig . 2 where @xmath152 and @xmath153 are plotted as functions of @xmath12 .
the magnitude of the ac phase can actually be approximately measured by a comparison between the @xmath12-coordinates of the transmission probabilities peaks in the absence and in the presence of the so interaction .
the energy splitting due to zeeman coupling is illustrated in fig .
3 . for @xmath154 and @xmath44 , since the kramers degeneracy makes each two eigenstates of the closed ring have the same energy , at certain @xmath83 the transmission probabilities in the two spin branches can reach their highest value 1 simultaneously , thereby making @xmath155 .
after the zeeman coupling is turned on , the resulted energy splitting destroys the simultaneous happenings of the resonances in the two spin branches and we see the maximum values of @xmath156 decrease with the strength of zeeman coupling appreciably . with @xmath141 being even smaller
, the energy dependence of the transmission probability manifests interesting trs - breaking effect , which also has its corresponding observability in persistent current . in ref .
@xcite , it has been demonstrated that in the presence of so interaction , the trs - breaking mechanism due to zeeman coupling is intrinsically different from that due to ab flux .
as the corresponding observable effect , it has been found that the direction of the persistent current induced by zeeman coupling changes periodically with the particle number @xmath157 with the periodicity @xmath158 while the direction of the persistent current induced by ab flux never changes with the particle number .
the dependence of the current direction on the particle number is actually the dependence on fermi energy .
such energy dependence of the current direction , an equilibrium phenomenon as it is , can actually be manifested in the resonant tunneling of electrons , a transport phenomenon as it is , in the weak coupling limit . for @xmath159 ,
the peaks of @xmath160 locate at the eigenenergies @xmath34 of the closed ring @xcite . in the presence of the so interaction and a weak zeeman coupling ,
the transmission probability is plotted as a function of the incident energy in fig .
every two peaks , which are closest to each other , locate at a pair of splitted energy levels , which come from the kramers doublet @xmath161 in the absence of zeeman coupling . with the ab flux being zero , the energy splittings in all the splitted energy levels are the same . here
we use the first - order perturbation which gives the energy correction but does nt change the eigenfunction . as shown in ref .
@xcite , those eigenstates of the closed ring , with increasing energy , have the spin orientations and current directions in a sequence of @xmath162,\;\ ; [ ( -,d),(+,u)],\;\;[(+,d),(-,u)],\;\;[(-,d),(+,u)],\;\;\cdot\cdot\cdot , \ ] ] where @xmath163 refers to a single quantum state with @xmath164 ( anticlockwise ) or @xmath165 ( clockwise ) denoting the current direction and @xmath166 ( up ) or @xmath167 ( down ) denoting the spin orientation , and @xmath168 $ ] refers to a pair of energy levels from the kramers doublet .
the eigenstate correspondence so identified for @xmath156 leads to interesting resonance behavior , as depicted in fig .
it is seen that when a small ab flux is added to distinguish the current directions , each two paired peaks are separated by a distance , which takes the larger or the smaller value alternatingly .
the reason is already clear in the sequence ( 27 ) .
in essence , since the current direction determines the sign of energy shift caused by a small ab flux , for @xmath169 $ ] the energy splitting due to the small ab flux enhances that first caused by the zeeman coupling , while for @xmath170 $ ] the energy splitting due to the small ab flux cancels part of that first caused by the zeeman coupling .
we want to point out that the essential character of the above correspondence between the equilibrium and the transport properties can be quantitatively , but not be qualitatively , affected by the disorder or scattering potential in the ring as long as the single particle picture holds for electronic states . in particular , such correspondence ,
if experimentally verified or excluded in some specific ring , may serve to detect if the electron - electron interaction qualitatively alters the electronic states .
we have studied the motion of electrons confined in the perfect ring in the presence of the cylindrically symmetric spin - orbit interaction and zeeman coupling , and the magnetic flux .
we have obtained the exact solution of the closed ring by using the aa phase approach in which the geometric and the dynamical phases can been explicitly identified for all energy eigenstates .
starting from the exact solution for the closed ring , we have investigated the transport properties of the ring connected to current leads , with emphasis on the roles of the trs and its breaking therein .
> from the derivation of the transmission probability , we have shown that in the presence of the trs - breaking zeeman coupling , it is physically impossible to adopt the picture that the spin precession of electrons encircling the ring results in some effective , spin - dependent aharonov - bohm flux in interference , thereby revealing the origin of the mistakes in some prior works .
we have provided the numerical results for illustrating the joint effects of spin - orbit interaction , zeeman coupling and magnetic flux .
> from the resonance behavior of the transmission probability in the weak coupling limit , we have found the observable correspondence between the trs - breaking effects manifested by the persistent current and by the transmission probability as long as the single particle picture of electronic states holds .
y. imry , in _ directions in condensed matter physics _ , edited by g. grinstein and g. mazenko ( world scientific , singapore , 1986 ) ; a. g. aronov and yu .
v. sharvin , rev .
59 * , 755 ( 1987 ) ; p. a. lee , a. d. stone , and h. fukuyama , phys . rev .
* b 35 * , 1039 ( 1987 ) ; m. b " uttiker , y. imry , and r. landauer , phys .
* 96a * , 365 ( 1983 ) ; y. gefen , y. imry , and m. ya .
azbel , phys .
lett . * 52 * , 129 ( 1984 ) . , edited by a. shapere and f. wilczek , ( world scientific , singapore , 1989 ) . m. v. berry , proc .
london * a 392 , * 45(1984 ) .
y. aharonov and j. anandan , phys .
lett . * 58 * , 1593 ( 1987 ) .
d. loss , p. goldbart , and a. v. balatsky , phys .
lett . * 65 * , 1655 ( 1990 ) .
d. loss and p. m. goldbart , phys .
b 45 * , 13544 ( 1992 ) .
a. stern , phys .
lett . * 68 * , 1022 ( 1992 ) ; d. loss , h. schoeller , and p. m. goldbart , phys . rev . *
b 48 * , 15218 ( 1993 ) .
x. c. gao and t. z. qian , phys . rev . *
b 47 * , 7128 ( 1993 ) .
y. aharonov and a. casher , phys .
lett . * 53 * , 319 ( 1984 ) .
y. meir , y. gefen , and o. entin - wohlman , phys .
lett . * 63 * , 798 ( 1989 ) ; o. entin - wohlman , y. gefen , y. meir , and y. oreg , phys . rev . *
b 45 * , 11890 ( 1992 ) . h. mathur and a. d. stone , phys .
68 * , 2964 ( 1992 ) .
h. mathur and a. d. stone , phys .
* b 44 * , 10957 ( 1991 ) . a. v. balatsky and b. l. altshuler , phys .
lett . * 70 * , 1678 ( 1993 ) .
m. y. choi , phys .
* 71 * , 2987 ( 1993 ) .
a. g. aronov and y. b. lyanda - geller , phys .
lett . * 70 * , 343 ( 1993 ) .
y. lyanda - geller , phys .
lett . * 71 * , 657 ( 1993 ) .
t. z. qian and z. b. su , phys . rev .
lett . * 72 * , 2311 ( 1994 ) .
s. oh and c.m .
ryu , phys .
b 51 * , 13441 ( 1995 ) .
tie - zheng qian , ya - sha yi , and zhao - bin su , phys .
* b * , in press . m. bttiker , y. imry , and m. ya .
azbel , phys . rev . * a 30 * , 1982 ( 1984 ) . | we consider the motion of electrons through a mesoscopic ring in the presence of spin - orbit interaction , zeeman coupling , and magnetic flux .
the coupling between the spin and the orbital degrees of freedom results in the geometric and the dynamical phases associated with a cyclic evolution of spin state . using a non - adiabatic aharonov - anandan phase approach
, we obtain the exact solution of the system and identify the geometric and the dynamical phases for the energy eigenstates .
spin precession of electrons encircling the ring can lead to various interference phenomena such as oscillating persistent current and conductance .
we investigate the transport properties of the ring connected to current leads to explore the roles of the time - reversal symmetry and its breaking therein with the spin degree of freedom being fully taken into account .
we derive an exact expression for the transmission probability through the ring .
we point out that the time - reversal symmetry breaking due to zeeman coupling can totally invalidate the picture that spin precession results in effective , spin - dependent aharonov - bohm flux for interfering electrons .
actually , such a picture is only valid in the aharonov - casher effect induced by spin - orbit interaction only .
unfortunately , this point has not been realized in prior works on the transmission probability in the presence of both so interaction and zeeman coupling .
we carry out numerical computation to illustrate the joint effects of spin - orbit interaction , zeeman coupling and magnetic flux .
by examining the resonant tunneling of electrons in the weak coupling limit , we establish a connection between the observable time - reversal symmetry breaking effects manifested by the persistent current and by the transmission probability . for a ring formed by two - dimensional electron gas
, we propose an experiment in which the direction of the persistent current can be determined by the flux - dependence of the transmission probability .
that experiment also serves to detect if the electron - electron interaction can qualitatively alter the electronic states .
spin precession and time - reversal symmetry breaking + in quantum transport of electrons through mesoscopic rings institute of theoretical physics , chinese academy of sciences , + p.o .
box 2735 , beijing 100080 , people s republic of china |
in this paper we explore an operator theoretic approach to self similarity and fractals which is common to problems involving both iterated function systems ( ifss ) , and an aspect of quantum communication .
each of the two areas involves a finite set of operations : in the case of ifss they are geometric , and in the quantum case , they involve channels of hilbert spaces and associated operator systems .
the particular aspects of ifss we have in mind are studied in @xcite ; and the relevant results from quantum communication in @xcite and @xcite .
we begin the introduction with some background on ifss , and we motivate our present operator theoretic approach . the operator theory and its applications are then taken up more systematically in section 2 below .
it was proved recently that a class of non - linear fractals is amenable to a computational harmonic analysis .
while this work is motivated by applications , it is significant in its own right ; for example these fractals do not carry the structure of groups , and so they do not come with a haar measure .
nonetheless , they have natural equilibrium measures @xmath3 which serve as substitutes .
specifically , these fractals are known as iterated function systems ( ifs ) indicating the recursive nature of their construction .
but the analysis so far has been limited to ifss without essential overlap ( see definition [ defcon.essove ] ) .
the ifs fractals @xmath10 are built by passing to a certain limit in a scaling iteration , and the overlap here refers to the parts of smaller scale - level .
we identify overlap for a given @xmath10 relative to the equilibrium measure @xmath3 for @xmath10 . while there is prior work on some isolated cases of overlap , the cases that are understood are those of negligible overlap . in this paper
we identify and study the harmonic analysis of ifss with substantial overlap .
an iterated function system ( ifs ) is a system of contractive mappings @xmath0 , @xmath1 ( finite ) where y is a complete metric space .
every such ifs has a unique ( up to scale ) equilibrium measure ( also called the hutchinson measure @xmath3 ) , and we study the hilbert space @xmath4 . in this paper
we extend previous work on ifss without overlap .
our method involves systems of operators generalizing the more familiar cuntz relations from operator algebra theory , and from subband filter operators in signal processing . before turning to the details , below we outline briefly the operator - theoretic approach to ifss . for each @xmath11
, there is a simple cuntz @xmath12-algebra on generators and relations , and its representations offer a useful harmonic analysis of general ifss , but there is a crucial difference between ifss without overlap and those with essential overlap ( this will be made precise in section [ mul ] ) .
the operators generating the appropriate cuntz relations @xcite are composition operators , e.g. , @xmath5 where @xmath6 is the given ifs .
if the particular ifs is essentially non - overlapping , it is relatively easy to compute the adjoint operators @xmath7 , and the @xmath8 operators will be isometries in @xmath4 with orthogonal ranges . in a way , for the more difficult case of essential overlap , we can use the extra terms entering into the computation of the adjoint operators @xmath9 as a measure of the essential overlap for the particular ifs we study . here
the adjoint operators @xmath9 refer to the hilbert space @xmath4 where @xmath3 is carefully chosen .
when the ifs is given , there are special adapted measures @xmath3 .
we will be using the equilibrium measure @xmath3 for the given ifs @xmath6 , and by @xcite , this @xmath3 contains much essential information about the ifs , even in the classical cases of ifss coming from number theory .
so far , earlier work in the general area was restricted to ifss without overlap .
for example , cuntz - like operator systems were used in recent papers on wavelets analysis by baggett , jorgensen , merrill and packer @xcite , where they serve as a first step to generating wavelet bases of parseval type ( alias normalized tight frames ) , i.e. , wavelet bases with redundancy .
similarly , it was shown in work by dutkay and jorgensen @xcite that the iterative operator approach works well for generating wavelets on fractals from ifss without overlap .
but so far the more general and more difficult case of essential overlap has resisted previous attempts at a harmonic analysis , and explicit basis constructions in particular . in this paper ( section [ ove ] ) we show that certain combinatorial cycles from a symbolic encoding @xcite of our ifss yield an attractive computational analysis of the `` ifs - overlap . ''
some results ( e.g. , corollary [ corovenov29.lowerbound ] ) are stated only for a model example , but as indicated the idea with cycle - counting works in general . while the modern study of iterated function systems ( ifss ) has roots in classical problems from number theory , infinite products @xcite , and more generally from probability and harmonic analysis ( see , e.g. , @xcite ) , the subject took a more systematic turn in 1981 with the influential paper @xcite by hutchinson . since then , ifss have found uses in geometry ( e.g. , @xcite ) , in infinite network problems ( e.g. , @xcite ) , in wavelets ( e.g. , @xcite ) , and in dynamics and operator theory @xcite .
but in operator algebras , the subject has its own independent start with the paper by cuntz @xcite . in operator - algebra theory , a class of representations of what have become known as the cuntz , or the cuntz
krieger , algebras is ideally suited for the study of iterative processes ( including ifss ) in mathematics and in mathematical physics .
perhaps the paper @xcite started this trend , but it was continued in various guises in the following papers : see @xcite and later related papers by popescu , @xcite , @xcite , @xcite , and @xcite , among others .
the cuntz and the related @xmath13-algebras have become ubiquitous tools in the study of iterative systems , or in the analysis of fractals or of graphs .
one reason for this is that there is an unexpected connection to signal and image processing ; see @xcite .
they are algebras on generators and relations ; and often , as is the case in the present paper , the generators may be identified as substitution operators @xcite . however , these substitution operators and their adjoints ( often called transfer operators ) are of independent interest ; see , e.g. , @xcite . since the cuntz and related @xmath13-algebras
typically are naturally generated by a finite set of concrete operators , it has become customary to study such finite sets as matrices with operator entries ; for example , a class of vectors with operator entries are studied in @xcite as row contractions . in fact , it is convenient to distinguish between row vectors with operator entries , and the corresponding columns of operators : one is the adjoint of the other . in this paper ,
our study of ifss with overlap leads naturally to a family of column isometries , where the operator entries in our column vectors are substitution operators built from the maps that define the ifs under consideration . while there is already a rich literature on ifss without overlap , the more difficult case of overlap
has received relatively less attention ; see , however , @xcite .
the point of view of the latter three papers is generalized number expansions in the form of random variables : real numbers are expanded in a basis which is a fraction , although the `` digits '' are bits ; with infinite strings of bits identified in a bernoulli probability space .
it turns out the distribution of the resulting random variables is governed by the measures which arise as a special cases of hutchinson s analysis @xcite of ifss with overlap .
see also @xcite .
however , concrete results about these measures have been elusive .
for example , it is proved in @xcite that the measures for expansions in a basis corresponding to ifss with overlap and given by a scaling parameter are known to be absolutely continuous for a.e .
value of the parameter .
in this paper we consider a rather general class of ifss with overlap .
we show that they can be understood in terms of the spectral theory of cuntz - like column isometries .
moreover we show that our column isometries yield exact representations of the cuntz relations precisely when the ifs has overlap of measure zero , where the measure is an equilibrium measure @xmath3 of the hutchinson type .
there is a separate development related to the study of fourier bases in the hilbert spaces @xmath4 for @xmath3 an ifs equilibrium measure ; see @xcite . we address some of these issues below , but it turns out that exact fourier bases for the case of ifs with overlap are harder to come by .
further , in @xcite , the coauthors noted that fourier bases in @xmath4 can only exist if the equilibrium measure @xmath3 has equal weights , and so in the present paper , we make this assumption .
recent references @xcite deal with wavelet constructions on non - linear attractors @xmath10 , such as arise from systems of branched mappings , e.g. , finite affine systems ( affine ifss ) , or julia sets @xcite generated by branches of the inverses of given rational mappings of one complex variable .
these @xmath10 come with associated equilibrium measures @xmath3 .
however , due to geometric obstructions , it is typically not possible for the corresponding @xmath14 to carry orthonormal bases ( onbs ) of complex exponentials , i.e. , fourier bases .
we showed instead in @xcite that wavelet bases may be constructed in ambient @xmath4 hilbert spaces via multiresolutions and representations of cuntz - like operator relations in @xmath14 as in definitions [ defmul.coliso ] and [ defmul.repcun ] in the next section .
in quantum communication ( the study of ( quantum ) error - correction codes ) , certain algebras of operators and completely positive mappings form the starting point ; see especially the papers @xcite and @xcite .
they take the form of a finite number of channels of hilbert space operators @xmath15 which are assumed to satisfy certain compatibility conditions .
the essential one is that the operators from a partition of unity , or rather a partition of the identity operator @xmath16 in the chosen hilbert space . here
( definition [ defmul.coliso ] ) we call such a system @xmath17 a column isometry .
an extreme case of this is when a certain cuntz relation ( definition [ defmul.repcun ] ) is satisfied by @xmath17 . referring back to our ifs application , the extreme case of the operator relations turn out to correspond to the limiting case of non - overlap , i.e. , to the case when our ifss have no essential overlap ( definition [ defcon.essove ] ) . in this section
we outline some uses of ideas from multivariable operator theory ( see , e.g. , @xcite ) in iterated function systems ( ifs ) with emphasis on _ ifss with overlap_. the tools we use are _ column isometries _ and systems of _ composition operators_. [ defmul.coliso]let @xmath18 be a complex hilbert space , and let @xmath19 , @xmath20 . a system @xmath21 of bounded operators in @xmath18 is said to be a _
column isometry _ if the mapping@xmath22 is isometric . here
we write the @xmath11-fold orthogonal sum of @xmath18 in column form , but we will also use the shorter notation @xmath23 . as a hilbert space , it is the same as @xmath24 , but for clarity it is convenient to identify the _ adjoint operator _ @xmath25 as a row@xmath26 the inner product in @xmath23 is @xmath27 , and relative to the respective inner products on @xmath23 and on @xmath18 , we have@xmath28 [ remmul.propproj]in the definition of a column isometry , it is stated that @xmath29 in is _ isometric _
@xmath30 , but it is not necessarily onto @xmath23 .
this means that in general the matrix operator @xmath31 is a proper _
projection_. by this we mean that the block matrix@xmath32 satisfies the following system of identities:@xmath33 note that @xmath34 if and
only if@xmath35 [ defmul.repcun]a column isometry @xmath29 satisfies @xmath36 if and only if it defines a representation of the cuntz algebra @xmath37 . in that case , the operators @xmath38 are isometries in @xmath18 with orthogonal ranges , and@xmath39 [ remmul.dimfin]the distinction between the operator relations in definitions and is much more than a technicality : definition is the more restrictive . because of the orthogonality axiom in definition , it is easy to see that if a hilbert space @xmath18 carries a nonzero representation of the cuntz relations definition see @xcite , then @xmath18 must be infinite - dimensional , reflecting the infinitely iterated and orthogonal subdivision of projections , a hallmark of fractals .
in contrast , the condition of definition , or equivalently @xmath40 , may easily be realized when the dimension of the hilbert space @xmath18 is finite .
in fact such representations are used in quantum computation ; see , e.g. , ( * ? ? ?
* theorem 2 ) and @xcite .
[ defmul.abscon]let @xmath41 be a finite measure space , i.e. , @xmath10 is a set , @xmath42 is a sigma - algebra of subsets in @xmath10 , and @xmath3 is a finite positive measure defined on @xmath42 .
we will assume that @xmath42 is complete , and the hilbert space @xmath43 will be denoted @xmath44 for short . if @xmath45 is a second measure , also defined on @xmath42 , we say that @xmath46 relative absolute continuity@xmath47 in that case , the corresponding radon
nikodym derivative will be denoted @xmath48 ; i.e. , we have @xmath49 , and @xmath50 an endomorphism @xmath51 is said to be _ measurable _
if@xmath52 where @xmath53 . in that case
, a measure @xmath54 is _ induced _ on @xmath42 , and given by @xmath55 [ remmul.comop]every measurable endomorphism @xmath56 , referring to @xmath41 , induces a _ _ composition operator__@xmath57 and it can be shown that @xmath58 is a _ bounded _ operator in the hilbert space @xmath59 if and only if @xmath60 with radon nikodym derivative in @xmath61 . moreover , if @xmath62 , set @xmath63 .
[ defmul.fintyp]a measurable endomorphism @xmath51 is said to be of _ finite type _ if there are a _ finite partition _
@xmath64 of @xmath65 and measurable mappings @xmath66 , @xmath67 such that @xmath68 [ lemmul.adjform]let @xmath41 be as above , and let @xmath51 be a measurable endomorphism of finite type .
suppose @xmath69 , and set @xmath70 .
[ lemmul.adjform(1)]then @xmath71 is supported @xmath72-@xmath73 on @xmath74 2 .
[ lemmul.adjform(2)]moreover , if @xmath64 is a non - overlapping partition as in , then the adjoint operator @xmath75 of the composition operator is given by the formula@xmath76 i.e. , for @xmath77 , @xmath78 .
we will begin by assuming @xmath69 . at the end of the resulting computation
, it will then be clear that the reasoning is in fact reversible .
the composition operator @xmath58 is well defined as in ( [ eqmul.12 ] ) , referring to the hilbert space @xmath59 with its usual inner product@xmath79 for the adjoint operator @xmath75 , we have@xmath80 for all @xmath81 . the desired formula ( [ eqmul.13 ] ) for the adjoint operator @xmath75 follows .
[ defmul.ifs]let @xmath41 be a finite measure space .
a system @xmath82 of measurable endomorphisms is said to be an _ iterated function system _ _ ifs_. let @xmath83 $ ] be given such that @xmath84 .
if@xmath85 we say that the measure @xmath3 is a @xmath86__-equilibrium _ _ measure .
motivated by earlier work on the harmonic analysis of equilibrium measures see , e.g. , @xcite and @xcite , we will here restrict attention to the case of equal weights , i.e. , assume that @xmath87 ; and we shall then refer to @xmath3 as an _ equilibrium measure _ , with the understanding that @xmath88 . in general , equilibrium measures might not exist ; and even if they do , they may not be unique .
[ promul.equmea]let @xmath41 be a finite measure space , and let @xmath82 be measurable endomorphisms .
then some @xmath3 is an equilibrium measure if and only if the associated linear operator@xmath89 is isometric , i.e. , if and only if the individual operators @xmath90 define a column isometry .
using polarization for the inner product @xmath91 in ( [ eqmul.14 ] ) , we first note that @xmath3 is an equilibrium measure if and only if@xmath92 holds for all @xmath93 . the terms on the left - hand side in ( [ eqmul.18 ] ) are @xmath94 , so ( [ eqmul.18 ] ) is equivalent to@xmath95 which in turn is the desired operator identity@xmath96 that defines @xmath29 as a column isometry . our main result in sections [ con ] and [ ove ] will yield a formula for the radon nikodym derivatives @xmath97 ; see remark [ remcon.equwei ] for details .
the study of contractive iterated function systems ( contractive ifss ) was initiated in a systematic form by hutchinson in @xcite , but was also used in harmonic analysis before 1981 .
suppose @xmath19 , @xmath20 , is given , and suppose some mappings @xmath98 , @xmath99 are contractive in some complete metric space @xmath100 : then we say that @xmath101 is a _
contractive ifs_. the contractivity entails constants @xmath102 , @xmath103 , such that@xmath104 hutchinson s first theorem @xcite states that there is then a unique compact subset @xmath105 such that@xmath106 this set @xmath10 is called the _ attractor _ for the system . if weights @xmath107 are given as in definition [ defmul.ifs ] , hutchinson s second theorem states that there is a unique ( up to scale ) positive finite ( nonzero ) borel measure @xmath3 on @xmath2 such that@xmath108 moreover , if @xmath109 for all @xmath110 , then the support of @xmath3 is the compact set @xmath10 ( fractal ) in ( [ eqcon.2 ] ) , the attractor .
for many purposes , it is convenient to normalize the measure @xmath3 in ( [ eqcon.3 ] ) such that @xmath111 . because of applications to harmonic analysis @xcite , we will also restrict the weights @xmath112 in ( [ eqcon.3 ] ) such that @xmath88 .
[ defcon.limmea]for borel measures @xmath45 on @xmath2 some given complete metric space , set @xmath113 set@xmath114 and@xmath115 then it is easy to see @xcite that @xmath116 is a metric and that the probability measures form a complete metric space with respect to @xmath116 .
moreover , regardless of the choice of the initial measure @xmath45 some probability measure on @xmath2 , the limit@xmath117 exists , where @xmath3 is the equilibrium measure from , @xmath88 , and where the convergence in is relative to the metric @xmath116 from .
[ exacon.intsin]let @xmath118 , and set @xmath119 , i.e. , the infinite cartesian product .
we equip @xmath120 with its usual tychonoff topology , and its usual metric .
points in @xmath120 are denoted @xmath121 , @xmath122 , @xmath123 . for @xmath124 , set @xmath125 .
further , we shall need the one - sided shifts@xmath126 if @xmath127 is a contractive ifs with attractor @xmath10 , it is easy to see that for each @xmath128 , the intersection@xmath129 is a singleton .
here we use the notation@xmath130 if @xmath128 is given , let @xmath131 be the unique point in the intersection .
the mapping @xmath132 is called the _
encoding_. [ lemcon.codcon]let @xmath127 and @xmath10 be as above , i.e. , assumed contractive .
then the coding mapping @xmath133 from is continuous , and we have@xmath134 the continuity is clear from the definitions .
we verify ( [ eqcon.11 ] ) : let @xmath135 .
then@xmath136 where we used contractivity of the mappings @xmath137 .
[ corcon.bernme]it follows from hutchinson s theorem @xcite that for each @xmath138 with @xmath139 , @xmath140 , there is a unique probability measure @xmath141 on @xmath120 such that@xmath142 when @xmath88 , this measure is called _ the bernoulli measure _ on @xmath120 .
the measures @xmath141 are also familiar infinite - product measures , considered first by kolmogorov @xcite .
[ corcon.equime]let @xmath127 be a contractive ifs , and let @xmath112 be given such that @xmath84 , @xmath140 .
let @xmath132 be the corresponding endoding mapping .
then the equilibrium measure @xmath143 for @xmath144 is@xmath145 i.e. , for borel subsets @xmath146 we have@xmath147 where @xmath148 let @xmath144 , @xmath149 , @xmath120 , and @xmath141 be as described . the following computation uses ( [ eqcon.11 ] ) in the form @xmath150 the conclusion will follow from the uniqueness part of hutchinson s theorem if we check that @xmath151 satisfies the identity ( [ eqmul.15 ] ) . let @xmath152 be a borel set @xmath153 .
then@xmath154 since @xmath151 is a probability measure , it follows that @xmath155 is the unique solution to ( [ eqmul.15 ] ) given by the normalization @xmath156 .
[ remcon.equwei]it is immediate from the definitions that all the measures @xmath3 solving for the case of positive weights @xmath109 will satisfy the relative absolute continuity condition@xmath157 but we shall be especially interested in the case of equal weights @xmath88 , and in this case we set @xmath158 [ exacon.xplus1]let @xmath159 , @xmath160 , @xmath161 , @xmath162 , and let @xmath163 be given . consider the two contractive maps @xmath164 for the attractor @xmath165 , we have the following three cases : case 1 : @xmath166 : @xmath167 case 2 : @xmath168 : @xmath169 \text { , and } \mu\text { \textup{(}}=\mu _ { 1/2}\text{\textup { ) } is the restriction of lebesgue measure . }
\label{eqcon.18}\ ] ] case 3 : @xmath170 : @xmath171 \text { ; overlap . } \label{eqcon.19}\ ] ] in the next section we will study these measures @xmath172 case @xmath173 in detail . in all three cases , the encoding mapping is @xmath174 the measure @xmath172 is determined by the following approximation : if @xmath152 is a borel subset of @xmath165 , then@xmath175 the only one of the stated conclusions which is not clear from corollary [ corcon.equime ] is the limit in ( [ eqcon.21 ] ) .
but ( [ eqcon.21 ] ) is an application of ( [ eqcon.6 ] ) to the case when the initial measure @xmath45 is @xmath176 ( the dirac measure ) , i.e. , @xmath177{ll}1 & \text{\quad if } 0\in e,\\ 0 & \text{\quad if } 0\notin e. \end{array } \right.\ ] ] note that@xmath178 the summation is over @xmath179 for each @xmath180 .
[ remcon.sumove]the three cases in are different in one important respect , _ overlap _ ; and we turn to this in the next section .
summary : overlap:@xmath181 \right ) \cap\tau_{1}\left ( \left [ \,0,1\,\right ] \right ) = \left\ { \frac{1}{2}\right\ } .\\ & \text{\textsc{case 3 : } } \lambda\in\left ( 1/2,1\right ) : & & \tau _ { 0}\left ( x_{\lambda}\right ) \cap\tau_{1}\left ( x_{\lambda}\right ) = \left [ \,\lambda,\frac{\lambda^{2}}{1-\lambda}\right ] .\end{aligned}\ ] ] [ defcon.essove]let @xmath41 be a finite measure space , and let @xmath127 be a finite system of measurable endomorphisms , @xmath182 , @xmath99 ; and suppose @xmath3 is some normalized equilibrium measure .
we then say that the system has _
essential overlap _
if@xmath183 in section [ gen ] , we will prove the following : [ thmcon.essove]let @xmath41 and @xmath101 be as in definition ; in particular we assume that @xmath3 is some @xmath184-equilibrium measure .
we assume further that each @xmath185 is of finite type .
let @xmath186 , @xmath187 and let @xmath188 be the corresponding column isometry then @xmath29 maps onto @xmath189 if and only if @xmath144 has zero @xmath3-essential overlap .
to better understand the geometric significance of this theorem , before giving the proof , we work out a particular family of examples ( sections [ ove ] and [ sie ] below ) in 1d and in 2d .
the intricate geometric features of ifss can be understood nicely by specializing the particular affine transformations making up the ifs to have a single scale number ( which we call @xmath190 ) . in the 1d examples , we will have two affine transformations ( [ eqove.1 ] ) , and in our 2d examples , three ( [ eqsie.3 ] ) .
( these special 1d examples also go under the name infinite bernoulli convolutions . here
we will use them primarily for illustrating the operator theory behind theorem [ thmcon.essove ] . ) as shown in section [ sie ] below , in passing from 1d to 2d , the possible geometries of the ifs - recursions increase ; for example , new fractions and new gaps may appear simultaneously at each iteration step . specifically , ( i ) fractal ( i.e. , repeated gaps ) and ( ii ) `` essential overlap '' co - exist in the 2d examples . here
`` essential overlap '' ( definition [ defcon.essove ] ) is of course defined in terms of the hutchinson measure @xmath3 ( @xmath191 ) .
our examples will be followed by a proof of theorem [ thmcon.essove ] in section [ gen ] . in this concluding section we further show ( theorem [ thmgen.isoext ] ) that every ifs with essential overlap ( definition [ defcon.essove ] ) has a canonical and minimal dilation to one with non - overlap .
[ remcon.repcun]note that the conclusion of the theorem states that the operators @xmath192 define a representation of the cuntz @xmath193-algebra @xmath37 if and only if the system has non - essential overlap , i.e. , if and only if @xmath194 for all @xmath195 .
the one - dimensional example in the previous section ( example [ exacon.xplus1 ] ) helps to clarify the notion of essential overlap . recall that in this example , @xmath190 is chosen such that @xmath196 , and the two endomorphisms are@xmath197 with attractor @xmath198 $ ] . but
the probability measure @xmath3 ( @xmath199 ) on this interval is not the restriction of lebesgue measure on @xmath200 .
in fact , @xmath172 is difficult to compute explicitly and there appears not to be a closed formula for @xmath201 \right ) $ ] , although the lebesgue measure of the overlap is @xmath202 .
it is known , for example , that @xmath172 is lebesgue - absolutely continuous for @xmath203 ; see @xcite .
from example [ exacon.xplus1 ] it also follows that @xmath172 is the distribution of the random variable @xmath204 , @xmath205 , or @xmath206 .
specifically , @xmath207 \right ) = : f_{\lambda}\left ( x\right ) .
\label{eqove.2}\ ] ] recall from section [ mul ] that @xmath208 .$ ] while explicit analytic properties ( like bounded variation ) satisfied by the cumulative distribution functions @xmath209 ( for specific values of @xmath190 ) are nt well understood , some geometric properties may be generated from the following proposition .
it gives an intrinsic scaling identity ( [ eqovenov26.pound ] ) for @xmath209 , and a recursive algorithm ( [ eqovenov26.poundpound ] ) for its computation .
first note that by ( [ eqove.2 ] ) , for every @xmath190 , the function @xmath210 is monotone nondecreasing .
moreover since the hutchinson measure @xmath172 is supported in the interval @xmath165 , @xmath209 is constantly @xmath211 on the negative real line , and it is @xmath212 on the infinite half line to the right of the bounded interval @xmath213 $ ] . of course ,
@xmath214 need not be strictly increasing in the interval .
but since it is monotone , by lebesgue s theorem , it is differentiable almost everywhere ( for a.e .
@xmath215 , taken with respect to lebesgue measure ) . however , a deep theorem @xcite states that , as a measure , @xmath172 has @xmath216 density for a.e .
@xmath190 in the open interval @xmath217 ; so in particular for these values of @xmath190 , the stieltjes measure @xmath218 is relatively absolutely continuous with respect to lebesgue measure . the next result ( needed later ! ) shows that with the choice of equal weights @xmath219@xmath219 in ( [ eqove.1 ] ) , the random variable @xmath220 $ ] from ( [ eqcon.20 ] ) is symmetric around the midpoint@xmath221 the symmetry property which is made precise in the next lemma reflects that the @xmath215-values are corners of a hypercube : repeated folding with shrinking intervals each with excess length .
[ symmetry][lemove.symmetry]let @xmath170 , and let @xmath222 and @xmath223 be as described in and . recall that @xmath224 is the closed interval @xmath225 $ ] where @xmath226 .
then for all @xmath227 the two tail - ends of the distribution are the same , i.e.,@xmath228 we will prove ( [ eqovenov29.new3 ] ) in the equivalent form@xmath229 the two events we specify on the left - hand side and the right - hand side in ( [ eqovenov29.new4 ] ) are described by the respective conditions@xmath230 and@xmath231 but since the fair - coin bernoulli measure @xmath223 on @xmath120 was chosen , the two sequences of independent random variables @xmath232 and@xmath233 are equi - distributed
. hence the numbers on the two sides in ( [ eqovenov29.new4 ] ) are the same .
the formula ( [ eqovenov29.new1 ] ) for the mean follows directly from ( [ eqove.1 ] ) as follows : set @xmath234 . then by ( [ eqove.1]),@xmath235 hence @xmath236 as claimed .
[ remove.cascademid ] actually every `` cascade approximant '' see proposition below is symmetric in a similar way around its midpoint @xmath237 .
[ remove.manymoments]the argument at the end of the proof of the lemma extends to yield a formula for all the moments @xmath238 beginning with@xmath239 and@xmath240 the formula for @xmath241 is recursive , and can be worked out from the binomial distribution .
there is a considerable amount of recent work see , e.g. , @xcite on moments in operator theory , and it would be interesting to explore the operator - theoretic significance of our present overlap - moments .
[ remove.antiatom]by the argument of the previous remark , the moment - generating function @xmath242 can be shown to have the following infinite - product expansion:@xmath243 using wiener s test on this , it follows that none of the measures @xmath172 have atoms , i.e. , that @xmath244 for all @xmath215 , and for all @xmath245 .
[ figure generation , hypercubes , and symmetry , by brian treadway][remove.hypercube]the @xmath215-values in the figures where steps occur in @xmath214 are generated using a recursive outer sum construction:@xmath246 this nested list of the @xmath247 sums has the structure of the coding map in fact , if it is expressed with tensor indices instead of nested braces , the indices are just our @xmath248 s .
the hypercube idea comes from thinking of each @xmath249 as a coordinate in an orthogonal direction .
opposite corners then have @xmath215-values that sum to @xmath250 , so the two tails may be matched point by point .
hence the @xmath215-values are corners of a hypercube , and hence symmetry . to make the plots of @xmath251 , the nested list above is `` flattened '' to a single list and sorted numerically .
the _ mathematica _ program that generates the sorted list is ` sort[flatten[nest[outer[plus , { 0 ,
l } , l # ] & , { 0 } , n ] ] , orderedq[{n[#1 ] , n[#2 ] } ] & ] ` , where `` ` l ` '' stands for @xmath190 and `` ` n ` '' is the level of iteration in the cascade algorithm . aside from erds s theorem @xcite about the case of @xmath190 equal to the golden ratio , where @xmath172 is singular , precious little is known about @xmath172 for other specific @xmath190 values , for example when @xmath190 is rational .
two interesting values of @xmath190 are @xmath252 ( fig .
[ figove.flambda ] ) and @xmath253 ( fig .
[ figfplotabcdefg ] ) .
the cases are qualitatively different both with regard to absolute continuity and overlap and with respect to fourier bases ; see @xcite . in the known results @xcite for which @xmath254 has a fourier orthonormal basis ( onb ) , @xmath190 is rational .
by a fourier basis , we mean an onb in @xmath254 consisting of complex exponentials .
because of @xcite , @xmath252 is likely to have its @xmath209 a little less `` differentiable '' than the @xmath209 for @xmath253 .
the following details will work as an iterative and cascading algorithm for @xmath255 when @xmath190 is fixed .
our algorithm is initialized so as to take advantage of ( [ eqcon.21 ] ) above .
it is illustrated in the figures ; see especially figure [ figove.flambdacascade ] .
[ proovenov26.1 ] let @xmath190 be given in the open interval @xmath256 , and set @xmath257 as a function on @xmath200 .
conclusions : 1 .
[ proovenov26.1(a ) ] then the function @xmath258 in satisfies @xmath259 or equivalently , for expansive scaling number @xmath260 , @xmath261 2 . [ proovenov26.1(b ) ] then the following iterative and cascading algorithm holds : initializing @xmath262 by setting @xmath262 to be the heaviside function @xmath263 we get the following recursion : @xmath262 , @xmath264 , @xmath265 , etc . , with @xmath266 or equivalently @xmath267 + moreover , @xmath268 holds for @xmath269 , and for all @xmath270 ; and the sequence @xmath271 is convergent pointwise in the closed interval @xmath272 = \left [ \,0,\frac{1}{s-1}\,\right ] .
\label{eqovenov26.interval}\ ] ] for each @xmath215 , this convergence is monotone , and @xmath273 the scaling identity ( [ eqovenov26.pound ] ) follows from the following facts ( see @xcite ) : the limit formula ( [ eqcon.6 ] ) ; the fixed - point property ( [ eqcon.3 ] ) , i.e. , @xmath274 ; and the fact that the equilibrium measure @xmath3 is supported in the closed interval @xmath272 .\ ] ] see also formula ( [ eqcon.21 ] ) .
_ ad _ ( [ proovenov26.1(a ) ] ) : specifically , for @xmath275 , we have @xmath276{$\scriptstyle\text{by ( \ref{eqove.2})}$}}{=}\int_{x_{\lambda}}\chi_{\left [ \,0,x\,\right ] } \left ( y\right ) \,d\mu\left ( y\right ) \\ & \underset { \makebox[\equallabelwidth]{$\scriptstyle\text{by ( \ref{eqove.1})}$}}{=}\frac{1}{2}\left ( \int_{0}^{\frac{\lambda}{1-\lambda}}\chi_{\left [ \,0,x\,\right ] } \left ( \tau_{0}\left ( y\right ) \right ) \,d\mu\left ( y\right ) + \int_{0}^{\frac{\lambda}{1-\lambda}}\chi_{\left [ \,0,x\,\right ] } \left ( \tau_{1}\left ( y\right ) \right ) \,d\mu\left ( y\right ) \right ) \\ & \underset{\,\operatorname*{supp}\left ( \mu\right ) \subset x_{\lambda}\,}{=}\frac{1}{2}\left ( \int_{0}^{\frac{x}{\lambda}}d\mu\left ( y\right ) + \int_{0}^{\frac{x-\lambda}{\lambda}}d\mu\left ( y\right ) \right ) \\ & \underset { \makebox[\equallabelwidth]{$\scriptstyle\text{by ( \ref{eqove.2})}$}}{=}\frac{1}{2}\left ( f\left ( \frac{x}{\lambda}\right ) + f\left ( \frac { x-\lambda}{\lambda}\right ) \right ) .\end{aligned}\ ] ] for the evaluation of the second integral , note : @xmath277 \iff-1\leq y\leq\frac{x-\lambda}{\lambda}$ ] .
but since @xmath278 , we get that@xmath279 _ ad _ ( [ proovenov26.1(b ) ] ) : figures [ figove.flambdacascade](a ) , ( b ) , ( c ) , @xmath265 are sketches of the successive functions @xmath262 , @xmath264 , @xmath280 , @xmath265 in the case of @xmath281 , i.e. , the reciprocal of the golden ratio @xmath282 .
as indicated , the sequence is monotone , i.e. , @xmath283 holds ; as will be proved .
( 324,117)(0,-45 ) ( 0,0 ) , @xmath284 , to the cumulative distribution function @xmath209 in the case when @xmath285 .
the marks on the horizontal axis in addition to @xmath190 and @xmath212 are @xmath286 , the endpoint of the support of the @xmath270th cascade iteration @xmath271 , and @xmath287 , the endpoint of the support of @xmath209 . in the case of @xmath288 for this particular `` golden '' value of @xmath190
, it may be observed that the set of node - points @xmath289 , and @xmath290 , not @xmath291 : one of the steps is doubled , due to the fact that @xmath292.,title="fig : " ] ( 18,70.4)(0,0)[r]@xmath293 ( 18,36)(0,0)[r]@xmath294 ( 18,0.5)(0,0)[tr]@xmath295 ( 63.5,0.5)(0,0)[t]@xmath296 ( 90,0.5)(0,0)[t]@xmath293 ( 132,0.5)(0,0)[t]@xmath297 ( 76,-13)(0,0)[t](a ) .
@xmath298 ( 171,0 ) , @xmath284 , to the cumulative distribution function @xmath209 in the case when @xmath285 .
the marks on the horizontal axis in addition to @xmath190 and @xmath212 are @xmath286 , the endpoint of the support of the @xmath270th cascade iteration @xmath271 , and @xmath287 , the endpoint of the support of @xmath209 . in the case of @xmath288 for this particular `` golden '' value of @xmath190
, it may be observed that the set of node - points @xmath289 , and @xmath290 , not @xmath291 : one of the steps is doubled , due to the fact that @xmath292.,title="fig : " ] ( 189,70.4)(0,0)[r]@xmath293 ( 189,36)(0,0)[r]@xmath294 ( 189,0.5)(0,0)[tr]@xmath295 ( 234.5,0.5)(0,0)[t]@xmath296 ( 261,0.5)(0,0)[t]@xmath293 ( 303,0.5)(0,0)[t]@xmath297 ( 247,-13)(0,0)[t](b ) .
@xmath299 ( 324,117)(0,-45 ) ( 0,0 ) , @xmath284 , to the cumulative distribution function @xmath209 in the case when @xmath285 .
the marks on the horizontal axis in addition to @xmath190 and @xmath212 are @xmath286 , the endpoint of the support of the @xmath270th cascade iteration @xmath271 , and @xmath287 , the endpoint of the support of @xmath209 .
in the case of @xmath288 for this particular `` golden '' value of @xmath190 , it may be observed that the set of node - points @xmath289 , and @xmath290 , not @xmath291 : one of the steps is doubled , due to the fact that @xmath292.,title="fig : " ] ( 18,70.4)(0,0)[r]@xmath293 ( 18,36)(0,0)[r]@xmath294 ( 18,0.5)(0,0)[tr]@xmath295 ( 63.5,0.5)(0,0)[t]@xmath296 ( 90,0.5)(0,0)[t]@xmath293 ( 132,0.5)(0,0)[t]@xmath297 ( 76,-13)(0,0)[t](c ) .
@xmath300 ( 171,0 ) , @xmath284 , to the cumulative distribution function @xmath209 in the case when @xmath285 .
the marks on the horizontal axis in addition to @xmath190 and @xmath212 are @xmath286 , the endpoint of the support of the @xmath270th cascade iteration @xmath271 , and @xmath287 , the endpoint of the support of @xmath209 . in the case of @xmath288 for this particular `` golden '' value of @xmath190
, it may be observed that the set of node - points @xmath289 , and @xmath290 , not @xmath291 : one of the steps is doubled , due to the fact that @xmath292.,title="fig : " ] ( 189,70.4)(0,0)[r]@xmath293 ( 189,36)(0,0)[r]@xmath294 ( 189,0.5)(0,0)[tr]@xmath295 ( 234.5,0.5)(0,0)[t]@xmath296 ( 261,0.5)(0,0)[t]@xmath293 ( 277.5,0.5)(0,0)[t]@xmath301 ( 303,0.5)(0,0)[t]@xmath297 ( 247,-13)(0,0)[t](d ) .
@xmath302 we now prove the assertion @xmath303 for all @xmath270 and all @xmath304 .
as before , @xmath170 is fixed , and @xmath305 is the right - hand endpoint in the interval @xmath165 .
the assertion follows by induction .
it holds for @xmath262 since @xmath262 is the heaviside function .
now suppose it holds for @xmath270 .
let @xmath306 be given .
then @xmath307 and@xmath308 so@xmath309 completing the induction . the separate assertion about pointwise convergence@xmath310 follows from the stronger fact : for each @xmath215
, the sequence @xmath311 is monotone , i.e.,@xmath312 induction : clearly ( [ eqovenov26.newpound2 ] ) holds if @xmath299 .
now suppose it holds up to @xmath270 .
then by the recursion , @xmath313 the conclusion ( [ eqovenov26.newpound2 ] ) now follows for @xmath314 , and the proof is complete .
[ removenov26.pound1]the three figures figs . taken by themselves offer an oversimplification in that the ordering of the node - points@xmath315 is unique only up to @xmath300 , i.e. , @xmath316 holds for all @xmath170 .
but in general for @xmath317 , the ordering of the @xmath247 points in @xmath318 is subtle .
for example , for @xmath302 , we have@xmath319 indicating that even in @xmath320 order - reversal may occur depending on the chosen value of @xmath190 in @xmath321 .
the list in further shows that the points in each set @xmath318 for @xmath317 typically occur with multiplicity . the assertion in ( [ eqovenov26.pound3 ] ) for the special case @xmath322 states that@xmath323 where @xmath324 is the encoding mapping @xmath325 from lemma [ lemcon.codcon ] and ( [ eqcon.20 ] ) .
while it is known that for all @xmath326 , @xmath327 is @xmath328@xmath212 , the infinite sets @xmath329 are not well understood .
[ proovenov28.onethird]if @xmath285 and @xmath222 is normalized , i.e. , @xmath330 , then @xmath331 [ proof of proposition ] from ( [ eqove.1 ] ) we see that @xmath332 $ ] and @xmath333 $ ] .
a symmetry consideration ( lemma [ lemove.symmetry ] ) further shows that @xmath334 .
we will compute @xmath335 using ( [ eqcon.21 ] ) .
in fact , we show that@xmath336 since @xmath337 , and @xmath338 , we get@xmath339 as claimed .
set @xmath340 . since @xmath341 , we then get @xmath342 , and therefore @xmath343 $ ] .
we now turn to ( [ eqovenov28.poundprime ] ) . by ( [ eqcon.21 ] ) , @xmath344 \right ) \right ) .\ ] ] recall that@xmath345 for all @xmath124 .
we claim that for @xmath317 , @xmath346 \right ) = \left\ { \,\omega\in\left\ { 0,1\right\ } ^{n}\mid\omega _ { 1}=1\,\right\ } \\ { } \cup\left\ { \,\omega\in\left\ { 0,1\right\ } ^{n}\mid\omega_{1}=0,\;\omega_{2}=\omega_{3}=1\,\right\ } \cup\cdots , \label{eqovenov28.poundprimeprime}\ ] ] where the union of the individual sets on the right - hand side is clearly disjoint .
the sets indicated by @xmath347 on the right - hand side in ( [ eqovenov28.poundprimeprime ] ) have the form@xmath348 the contribution to these sets is @xmath349 , where we used that @xmath350 .
it follows that@xmath351 \right ) \right ) = 2^{-1}+2^{-3}+2^{-5}+2^{-7}+\cdots=\frac{2}{3}\ ] ] as claimed .
it is clear that the right - hand side in ( [ eqovenov28.poundprimeprime ] ) is contained in @xmath352 \right ) $ ] for sufficiently large @xmath317 .
the assertion that they are equal follows from the fact that@xmath353 for all @xmath354 . but using @xmath355 , note that ( [ eqovenov28.poundprimeprimeprime ] ) is equivalent to @xmath356 , which is clearly true .
this proves the proposition .
recall that the overlap in @xmath172-measure is@xmath357 \right ) = p_{1/2}\left ( \left\ { \,\omega\in\omega\mid\pi _ { \lambda}\left ( \omega\right ) \in\left [ \,\lambda , b\left ( \lambda\right ) -\lambda\,\right ] \,\right\ } \right ) .\ ] ] this means that the contributions to @xmath358 \right ) $ ] with @xmath223-measure equal to zero may be omitted in the accounting ( [ eqovenov28.poundprimeprime ] ) above .
however , even for @xmath285 , even the infinite set @xmath359 has interesting dynamics .
but its contribution to the overlap in @xmath3-measure is @xmath360 see also remark [ remove.antiatom ]
. * notation . *
set@xmath361 then @xmath359 contains the following infinite lists:@xmath362 etc .
, and@xmath363 etc . [ removedec1.sive ]
the fact that for @xmath364 ( @xmath365 , @xmath366golden ratio ) the overlap measured in the hutchinson measure is @xmath367 appears also to follow from @xcite . in ( * ? ? ? * cor .
1.2 , p. 220 ) , entirely about the golden shift , sidorov and vershik do get @xmath367 by a completely different argument : they introduce a transition matrix on a combinatorial tree , and when translated into our @xmath368 example , their sidorov
vershik tree is then fibonacci .
that is key to their computations .
in contrast , our method is general and applies to general metric spaces : ifss with overlap .
even when specialized to 1d , for the special case of ( [ eqove.1 ] ) , our method has the advantage ( see corollary [ corovenov29.lowerbound ] and remark [ remove.goldenbound ] below ) of estimating the overlap in hutchinson measure also when @xmath190 is not @xmath369 , i.e. , is not `` golden . ''
[ remove.measureformeasure ] [ remove.measureformeasure(1)]measure of the intersection @xmath370 is @xmath371 , which for @xmath372 works out to@xmath373 [ remove.measureformeasure(2)]@xmath374 we conclude from the proposition that the hutchinson measure of the intersection is the smaller of the two .
[ remove.measureformeasure(3)]@xmath375 , @xmath376 , for all @xmath170 , and we conclude that there is essential overlap for all @xmath190 in @xmath321 , but an explicit formula for @xmath377 ( @xmath378 ) is not known .
the following is a consequence of the argument in the proof of proposition [ proovenov28.onethird ] .
[ corovenov29.lowerbound ] [ corovenov29.lowerbound(1 ) ] for all @xmath379 we have@xmath380 [ corovenov29.lowerbound(2 ) ] for all @xmath381 , there is some @xmath382 , depending on @xmath190 , such that@xmath383 and for such a choice of @xmath382 we have@xmath384 [ remove.goldenbound]if @xmath285 , the number @xmath382 in may be taken to be @xmath385 , and in that case is @xmath386 .
[ proof of corollary ] _ ad _ ( [ corovenov29.lowerbound(1 ) ] ) : an easy modification of the argument in the proof of proposition [ proovenov28.onethird ] shows that if @xmath387 , then ( [ eqovenov29b.new1 ] ) holds .
_ ad _ ( [ corovenov29.lowerbound(2 ) ] ) : let @xmath388 be given .
it follows from algebra that if @xmath382 is sufficiently large , then ( [ eqovenov29b.new2 ] ) must hold .
we pick @xmath382 to be the first number which gets the sum on the left - hand side in ( [ eqovenov29b.new2 ] ) @xmath389 .
consider the following specific finite word @xmath390 in @xmath391 given by@xmath392 and generate more words as follows:@xmath393 where free means unrestricted strings of bits .
as before , the resulting sequence of subsets in @xmath120 is disjoint .
then@xmath394 the argument from the proposition yields@xmath395 which is the assertion ( [ eqovenov29b.new3 ] ) .
_ cascade approximation_. each function @xmath396 from the approximation to the cumulative distribution in proposition [ proovenov26.1 ] has a finite set of node points @xmath397 , and @xmath398 but for fixed @xmath190 , the configuration of points in @xmath397 can be complicated for @xmath317 and large ; and each set @xmath397 also depends on the particular numerical choice of a value for @xmath190 .
this is borne out in the cascade of figures ( see figure [ figove.flambdacascade ] ) made for @xmath399 .
we have included pictures of @xmath400 , @xmath401 , @xmath265 , up to @xmath402 . as sketched in remark [ removenov26.pound1 ] , the reason is that the detailed configuration and the multiplicities in the sets @xmath397 of node points are reflected in the progression of graphs of the functions @xmath403 .
this fine structure only becomes visible for large @xmath317 . in figure [ figove.flambda ]
above , we summarize the distribution of @xmath404 , i.e. , the function @xmath405 in ( [ eqove.2 ] ) . by @xcite ,
@xmath405 is only known to have @xmath406-@xmath73 derivative , or @xmath406-density , for @xmath407 in @xmath408 .
the differences in cascade approximation when the value of @xmath190 varies is illustrated by figures [ figove.flambdacascade ] , [ figove.flambda ] , [ figfplotabcdefg ] , and [ figfplothijk ] .
we have already outlined that @xmath409 is special .
figure [ figove.flambdacascade ] illustrates the cascades from proposition [ proovenov26.1 ] , @xmath410 , with each step representing a @xmath190 scaled subdivision .
figure [ figove.flambda ] represents the `` idealized '' limit of the iteration . since there are only very few rigorous results in the literature for specific values of @xmath190 in the open interval @xmath256 ( see section [ int ] ) , we have included mathematica sketches of of @xmath209 for two chosen values of @xmath190 in figures [ figfplotabcdefg ] and [ figfplothijk ] .
( 2.118,1.5)(-0.25,-0.25 ) ( 0,0 ) , where @xmath409 . caution : a closed formula for @xmath411 is not known . but using the second formula in ( [ eqove.1 ] ) and ( [ eqove.2 ] ) , the reader may check that for fixed @xmath190 , the cumulative distribution @xmath214 ( @xmath412 ) satisfies the scaling identity @xmath413.,title="fig : " ] ( 0,1.2)(0,0)[bl]@xmath414 ( 0,1)(0,0)[r]@xmath212 ( 0,0)(0,0)[t]@xmath211 ( 1.618,0)(0,0)[t]@xmath415 ( 1.818,0)(0,0)[tl]@xmath215 ( 0.809,-0.1)(0,0)[t]@xmath416 our next result gives the two radon nikodym derivatives in the case of ( [ eqove.1 ] ) for @xmath190 fixed , i.e. , @xmath196 is chosen .
the measure @xmath222 is chosen such that @xmath417 . then set@xmath418 ( 252,117)(0,-45 ) ( 0,0),title="fig : " ] ( 21,70.4)(0,0)[r]@xmath293 ( 21,36)(0,0)[r]@xmath294 ( 21,0.5)(0,0)[tr]@xmath295 ( 75,0.5)(0,0)[t]@xmath296 ( 92.1111,0.5)(0,0)[t]@xmath293 ( 177.667,0.5)(0,0)[t]@xmath419 ( 227.971,0.5)(0,0)[t]@xmath420 [ proove.radnik]the two radon nikodym derivatives @xmath421 and @xmath422 from are given by the formulas@xmath423 the result follows from corollary [ corcon.equime ] and from the present discussion of example [ exacon.xplus1 ] . ( 360,117)(0,-45 ) ( 0,0),title="fig : " ] ( 10,70.4)(0,0)[r]@xmath293 ( 10,36)(0,0)[r]@xmath294 ( 10,0.5)(0,0)[tr]@xmath295 ( 26,0.5)(0,0)[t]@xmath424 ( 334,0.5)(0,0)[t]@xmath419 ( 177.939,0.5)(0,0)[t]@xmath301 ( 348,0.5)(0,0)[t]@xmath297 [ remove.sketch]for the convenience of the reader , we have sketched the two functions @xmath421 and @xmath422 in figures and above . _
details for , figure .
_ using , , corollary , and conditional probabilities , we get for all @xmath425 borel subsets of @xmath165 the formula@xmath426 where the function under the integral sign in coincides with the expression in .
[ remove.chomu]as above , let @xmath196 , and @xmath427 $ ] . then @xmath428 $ ] and @xmath429 $ ]
; see figures and .
we include this note to stress that the radon nikodym derivatives are sensitive to the choice of @xmath3 .
take for example @xmath430lebesgue measure on @xmath165 .
then an easy computation yields@xmath431 but the corresponding @xmath29 is then _ not _ a column isometry . [ proove.ranpro]continuing the example in proposition . if we choose @xmath3 to be the equilibrium measure , then the isometry @xmath432 yields the following formula for the range - projection in @xmath433:@xmath434 where the composite functions @xmath435 serve as multiplication operators in @xmath59 , while the two other operators making up the entries in are@xmath436 as before @xmath190 is fixed such that @xmath196 , and @xmath437 is the equilibrium measure .
the result follows from a computation , and an application of remark [ remmul.comop ] and lemma [ lemmul.adjform ] .
( 2.118,2.5)(-0.25,-0.3 ) ( 0,0).,title="fig : " ] ( -0.1,1.2)(0,0)[r]@xmath421 ( 0,2)(0,0)[r]@xmath438 ( 0.05,0)(0,0)[tr]@xmath439 ( 1.618,0)(0,0)[t]@xmath415 ( 1,0)(0,0)[t]@xmath440 ( 0.618,0)(0,0)[t]@xmath190 ( 0.809,-0.2)(0,0)[t]overlap ( 2.118,2.5)(-0.25,-0.3 ) ( 0,0).,title="fig : " ] ( -0.1,1.2)(0,0)[r]@xmath422 ( 0,2)(0,0)[r]@xmath438 ( 0.05,0)(0,0)[tr]@xmath439 ( 1.618,0)(0,0)[t]@xmath415 ( 1,0)(0,0)[t]@xmath440 ( 0.618,0)(0,0)[t]@xmath190 ( 0.809,-0.2)(0,0)[t]overlap [ corove.ortcom]continuing the example from propositions and , we note that the function @xmath441 is nonzero , and it is in the orthogonal complement of the range of the column isometry @xmath29 . using ( [ eqove.5 ] ) , a computation shows that @xmath442 in ( [ eqove.6 ] ) satisfies @xmath443 ; and moreover that @xmath444 is nonzero in @xmath445 .
this section includes a 2d variant of the 1d examples from section [ ove ] above .
the case of 2d is interesting and different in that the fractal features become apparent both for the associated hutchinson measures , as well as for their support .
contrast : in the best known gap fractal , the cantor set , the middle thirds are omitted , and as a result the familiar cascading gap - subintervals emerge . as we saw in example [ exacon.xplus1 ] , in 1d ( case 3 ) when @xmath190 is adjusted ,
@xmath447 so as to create @xmath172-essential overlap for the infinite convolution ifs , then the omitted middles disappear , and the resulting @xmath165 is simply an interval . not so for the analogous 2d construction ! as we see , in 2d overlap and gaps may coexist !
recall that the attractor for a contractive ifs in a complete metric space is equal to the support of the corresponding hutchinson measure . in the next result
, we show that when the same procedure from proposition [ proovenov26.1 ] ( and ( [ eqove.1])([eqove.2 ] ) ) is extended to 2d we get a one - parameter family of fractals with gaps , and essential overlap at the same time .
so in the context of ( [ eqcon.2 ] ) and ( [ eqcon.3 ] ) , the ambient space @xmath448 , the weights in ( [ eqcon.3 ] ) are @xmath449 , the number @xmath190 will be in the open interval @xmath450 , and the attractor @xmath451 will be a sierpinski fractal with essential overlap and gaps .
its hausdorff dimension @xmath452 will be @xmath453 this sierpinski fractal@xmath454 is sketched below in figures [ figsie.4 ] , [ figsie.1 ] , [ figsie.3 ] , and [ figsie.2 ] for the cases @xmath455 , @xmath456 , @xmath457 , and @xmath458 , respectively .
( 324,162)(0,-45 ) ( 0,0 ) for @xmath455 . ,
title="fig : " ] ( 76,-13)(0,0)[t](a ) .
@xmath299 ( 171,0 ) for @xmath455 .
, title="fig : " ] ( 247,-13)(0,0)[t](b ) .
@xmath300 let @xmath190 be fixed in the open interval @xmath459 .
the case @xmath460 is the familiar sierpinski gasket in 2d .
recursion : by analogy to the middle - third cantor - set construction , start with a triangle @xmath461 with vertices @xmath462 , @xmath463 , and @xmath464 .
the three @xmath190-scaled triangles @xmath465 are shaded in light grayscale , and they have pairwise overlaps as indicated in the first iteration in fig . [ figsie.1 ] .
these three pairwise overlaps are the first smaller dark - shaded triangles .
the recursion now continues .
omitted triangles : the first gap , i.e. , the first white interior triangle , is the set - theoretic difference @xmath466 .
the fractal @xmath165 now arises by iteration just as in the familiar case of recursion for the middle - third cantor set . in the first step of the recursion ,
there is just one interior and centered gap - triangle ; it is inverted from the position of the initial @xmath461 .
also note that the size of the overlaps in the iteration is monotone in @xmath190 , small when @xmath190 is close to @xmath467 , the usual sierpinski gasket .
in contrast , the omitted middles decrease gradually with @xmath190 and they collapse to points when @xmath468 .
the nature of the overlaps changes at the value @xmath469 .
we will discuss the nature of the overlaps in section [ natureoverlap ] .
initialize the recursion sequence at level @xmath298 with an inflated triangle @xmath461 , inflation factor@xmath470 , and then generate the fractal @xmath165 sequentially , @xmath180 , by the usual iteration limit , @xmath471 here @xmath472 denotes an @xmath270-fold composition of the individual @xmath185 maps with the indices @xmath16 making up the word @xmath390 , i.e. , @xmath390 giving the address of the respective small triangles @xmath473 for the @xmath270th level iteration : @xmath474 ( 324,162)(0,-45 ) ( 0,0 ) for @xmath409 .
see example [ exasie.1].,title="fig : " ] ( 76,-13)(0,0)[t](a ) .
@xmath299 ( 171,0 ) for @xmath409 .
see example [ exasie.1].,title="fig : " ] ( 247,-13)(0,0)[t](b ) .
@xmath300 [ exasie.1]*2d sierpinski with overlap and gaps , * @xmath475**. * * let the vectors @xmath476 , @xmath477 , @xmath478 in @xmath446 be given as follows:@xmath479@xmath480 with bernoulli measure @xmath481
corresponding to the infinite - product measure @xmath449 , @xmath482 , as in and corollary . we will denote by @xmath172 the measure in corollary , and we set @xmath483 .
the ifs is@xmath484 for @xmath485 . here
we make the restriction @xmath486 . as noted in section ,
@xmath487 and @xmath165 is the unique compact @xmath488 solution to .
let @xmath489 , and let @xmath481 be the usual bernoulli measure on @xmath120 with equal and independent probabilities @xmath490 .
the formula ( [ eqcon.20 ] ) for the random variable @xmath491 extends from 1d to 2d with the only modification that the coefficients @xmath492 from ( [ eqcon.20 ] ) now take values in the finite alphabet of vectors @xmath493 .
let @xmath494 , @xmath495 be the three vertices in @xmath165 ( figure [ figsie.1 ] ) ; i.e. @xmath496 , @xmath497 , and @xmath498 .
note that for each @xmath499 , @xmath165 is contained in the triangle @xmath461 with the vertices @xmath494 , @xmath495 . our first result concerns symmetry , and it is an immediate extension of lemma [ lemove.symmetry ] from the 1d case to the 2d case .
for each @xmath500 , let @xmath501 denote the equilateral triangle of side - length @xmath215 with vertex @xmath494 , which shares two sides with segments of sides in @xmath461 .
then the argument from lemma [ lemove.symmetry ] shows that for each @xmath502 , the three numbers @xmath503 coincide . since @xmath504 by corollary [ corcon.equime ]
, we conclude in particular that the three numbers @xmath505 agree for @xmath495 . for @xmath195 ,
set @xmath506 it follows that @xmath507 [ prosie.1 ] [ prosie.1(1 ) ] for @xmath508 , the fractal @xmath165 has hausdorff dimension@xmath509 it has essential overlap@xmath510 and the @xmath172-measure of the pairwise overlaps is independent of the pair @xmath511 with @xmath195 . [
prosie.1(2 ) ] setting@xmath512 @xmath513 , we get a vector - valued random variable , and@xmath514 holds , where @xmath515 is the encoding mapping .
[ prosie.1(3)]@xmath516 ( 324,162)(0,-45 ) ( 0,0 ) for @xmath517 .
note that @xmath518 , and the gaps are very small.,title="fig : " ] ( 76,-13)(0,0)[t](a ) .
@xmath299 ( 171,0 ) for @xmath517 .
note that @xmath518 , and the gaps are very small.,title="fig : " ] ( 247,-13)(0,0)[t](b ) .
@xmath300 since the proof is essentially contained in the previous sections , we shall be brief .
_ ad _ ( [ prosie.1(1 ) ] ) : the formula ( [ eqsie.5 ] ) in ( [ prosie.1(1 ) ] ) for the hausdorff dimension follows from the arguments in @xcite .
we begin with the case @xmath409 , and for the moment , we will drop @xmath190 from the notation .
our claim is that @xmath519 to see this , we introduce the bernoulli space @xmath120 and consider the cylinder sets in @xmath120 indexed by finite words in the alphabet @xmath520 , @xmath521 . if @xmath522 , set @xmath523 and note that @xmath524 .
set @xmath525 , and consider the following sequence of disjoint cylinder sets : @xmath526 the argument from section [ ove ] shows that @xmath527 using the symmetry argument from above , we also have @xmath528 the desired result @xmath529 follows
. starting at @xmath530 we see that the function @xmath531 is increasing .
hence , to prove ( [ eqsie.6 ] ) , we need only establish a lower bound for values of @xmath190 in the open interval @xmath532 .
but this can be done _
mutatis mutandis _ as in the proof of corollary [ corovenov29.lowerbound ] ; see ( [ eqovenov29b.new3 ] ) . if @xmath533 , determine @xmath534 as in ( [ eqovenov28.poundprimeprimeprime ] ) .
then it follows that @xmath535 _ ad _ ( [ prosie.1(2 ) ] ) : the proof of ( [ eqsie.6 ] ) in ( [ prosie.1(2 ) ] ) is based on symmetry considerations ( lemma [ lemove.symmetry ] ) extended from 1d to 2d , as well as the estimates in proposition [ proovenov28.onethird ] , remarks [ remove.measureformeasure ] , and corollary [ corovenov29.lowerbound ] .
_ ad _ ( [ prosie.1(3 ) ] ) : to see that the triple overlap is empty , calculate the distances between pairwise overlaps to the third sub - partition .
we conclude this section with the following open question : for what values of @xmath190 in the interval @xmath536 is @xmath172 absolutely continuous with respect to the @xmath438-dimensional lebesgue measure ? there are two pieces of partial evidence for absolute continuity of the two - dimensional @xmath172 for a.e .
@xmath537 : 1 .
since the interior gaps close at @xmath538 , so that for @xmath539 , @xmath540 ( the closed triangle ) , following the 1d analogy , it seems reasonable to expect the a.e .
conclusion in this range of @xmath190 .
one of the proofs @xcite in the literature for the 1d case introduces a clever fubini - tonelli argument with a function in several variables with @xmath190 as one of the integration variables . using a density argument
, one then gets finiteness of a corresponding key functions for a.e .
@xmath190 in the interval , and this in turn ( following the 1d analogy ) is likely to yield an expression in 2d for the radon - nikodym derivative for those values of @xmath190 . + in particular , for @xmath541 ,
let @xmath542 be the equilateral triangle centered at @xmath215 with side length @xmath543 and area @xmath544 .
define @xmath545 ( an analogue of the first formula on @xcite ) .
the argument from @xcite is likely to yield @xmath546 from that we could conclude that @xmath547 is finite for a.e .
@xmath548 , putting the radon - nikodym derivative of @xmath172 with respect to two - dimensional lebesgue measure in @xmath549 for a.e .
@xmath190 . as we noted before , the nature of the overlaps changes at the value @xmath409
here we refer to overlaps of monomials in the @xmath550 s of degree @xmath551 applied to the initial triangle @xmath552 as @xmath553 .
let @xmath554 denote overlaps at level @xmath270 for example , @xmath555 a. [ regularsierpinski]*the sierpinski gasket . * when @xmath556 , the resulting fractal @xmath165 is the sierpinski gasket , and the essential overlap is a set of lebesgue measure zero @xcite . b. [ essentialbutsimpleoverlap]*simple overlap .
* when @xmath533 , we have @xmath557 we call this type of overlap `` overlap of multiplicity one '' or `` simple overlap . '' when simple overlap occurs , the subset of @xmath165 which consists of the overlaps is itself an ifs .
so , an ifs with essential but simple overlap induces a new ifs with non - essential overlap .
see figure [ figsie.4 ] . c. [ essentialsimple]*the golden 2d fractal .
* let @xmath409 , as in figure [ figsie.1 ] .
we see @xmath558 as the dark shaded triangles in figure [ figsie.1 ] , picture ( a ) . as we move from ( a ) to ( b ) in figure [ figsie.1 ] , we see that successive overlap sets intersect at vertices : @xmath559 is non - empty .
in fact , @xmath558 consists of three triangles , and @xmath560 of @xmath561 .
each triangle in @xmath558 shares a vertex with one or two from @xmath560 , and the double - sharing happens for the interior triangles from @xmath560 .
this pattern continues , so that at each step of the iteration , each vertex of a triangle in @xmath554 coincides with some vertex in a triangle from @xmath562 .
we therefore have an induced system which forms a graph with edges and vertices , with each vertex from @xmath554 connecting to one or two vertices from @xmath562 .
+ in the following when we refer to disjoint pairs of triangles , we will mean disjointness of the respective interiors , " thus allowing the sharing of vertices .
the triangles in @xmath563 are disjoint from all the triangles in @xmath562 , but triangles from @xmath564 may be contained in triangles from @xmath563 .
in fact , a triangle from @xmath564 is either disjoint from triangles in the set @xmath563 , or that triangle is contained in a unique triangle from @xmath563 .
+ we can formalize the overlap between iterations by defining a set operation @xmath565 which takes a set @xmath566 to the set @xmath567we have already seen that @xmath568 and we also have @xmath569 suppose @xmath570that is , @xmath571 is a multi - index of length @xmath270 each of whose components is @xmath211 , @xmath212 , or @xmath438
. we can use @xmath571 to keep track of the monomials in the @xmath550 s which we mentioned above : @xmath572 then @xmath573 d. [ essentialmultiplicity]*the residual interval . *
when @xmath574 , the set @xmath575 no longer consists of discrete points for each @xmath270 , @xmath575 is uncountable .
see , for example , figure [ figsie.3 ] , where @xmath457 . to be more precise , if we rescale the initial triangle @xmath576 so that the length of each of its sides is @xmath577 , then for each @xmath270 , the set @xmath554 consists of @xmath578 disjoint triangles , each of side length @xmath579 .
e. [ closinggap ] * the closing of the gap .
* when @xmath538 , there is still overlap with multiplicity , but there are no longer gaps at each iteration .
this pattern continues for @xmath580 .
see figure [ figsie.3 ] , which shows a @xmath190 value slighly less than @xmath581 and figure [ figsie.2 ] , which illustrates no gaps ( @xmath582 ) .
summary of conclusions for 2d : we have sketched features that come out differently for `` ifs overlap - sierpinski fractals , '' stressing differences that arise when passing from 1d examples where @xmath583 to our analogous 2d attractors .
specifically , in the 2d case , for the range of values of @xmath190 , there are five separate cases for scaling numbers @xmath190 of interest , illustrating overlap features : a. @xmath584 : @xmath585 ( see figure [ figsie.4 ] ) b. @xmath409 : @xmath586 consists of vertices ( see figure [ figsie.1 ] ) ; simple overlap ; central gaps c. @xmath587 : @xmath586 consists of @xmath578 disjoint triangles ; overlap with multiplicity ; central gaps ( see figure [ figsie.3 ] for a sierpinski figure whose gaps are very small ) d. @xmath538 : central gaps close at @xmath538 ; overlap with multiplicity e. @xmath588 : no central gaps ; overlap with multiplicity ( see figure [ figsie.2 ] ) .
( 324,162)(0,-45 ) ( 86,0 ) for @xmath589 . in this case
, there are no gaps.,title="fig : " ]
we now return to the general case of ifss with essential overlap . in this case
, the size of the overlap can nicely be expressed in terms of the column isometry from definition [ defmul.coliso ] . to recall the setting
, we begin with the proof of theorem [ thmcon.essove ] that was postponed .
[ proof of theorem ] a system of measurable endomorphisms @xmath82 in a finite measure space @xmath590 is given , and it is assumed that @xmath591 , and that@xmath592 i.e. , that @xmath3 is a @xmath144-equilibrium measure . it then follows from proposition [ promul.equmea ] that the operators @xmath593 define a column isometry , i.e. , that @xmath594 satisfies @xmath595 .
[ lemove.mulope]let @xmath127 , @xmath3 , @xmath598 be as above , i.e. , @xmath29 is a column isometry @xmath599 .
let @xmath97 .
then@xmath600 where @xmath601 is the multiplication operator @xmath602 in @xmath59 .
the result follows essentially from the argument in lemma [ lemmul.adjform ] above . by that argument
we may pass to partitions of @xmath10 .
let @xmath110 be given , fixed ; and , following lemma [ lemmul.adjform ] , pass to a subset @xmath152 in @xmath10 such that there is a measurable mapping @xmath603 with@xmath604 see ( [ eqmul.12 ] ) .
setting @xmath97 and using lemma [ lemmul.adjform ] we show that there are measurable and invertible point transformations @xmath608 , @xmath609 , such that @xmath610 , @xmath611 , and@xmath612 but , for each @xmath110 , the function @xmath613 is supported on @xmath614 .
so if@xmath615 then@xmath616 the conclusion of the theorem is immediate from this ; and we get the following corollary . [ thmgen.isoext]let @xmath19 , @xmath20 , be given , and let @xmath621 be a contractive ifs in a complete metric space .
let @xmath622 be the hutchinson data ; see definition .
let @xmath623 @xmath624 be the bernoulli measure on @xmath625 ; see corollary .
let @xmath132 be the encoding mapping of lemma .
set@xmath626\intertext{and}s_{i}^{\ast}\psi & : = \frac{1}{\sqrt{n}}\psi\circ\sigma_{i}\text{\qquad for } \psi\in l^{2}\left ( \omega , p\right ) , \label{eqgen.10}\ ] ] where @xmath627 denotes the shift map of .
_ ad _ ( [ thmgen.isoext(1)])([thmgen.isoext(2 ) ] ) : let @xmath635 , and let @xmath636 and @xmath637 denote the respective @xmath216-norms in @xmath59 and @xmath638 .
\pi\right\vert ^{2}\,dp\\ & \underset{\makebox[\equallabelwidth]{{}}}{=}\int_{x}\left\vert f\right\vert ^{2}\,d\left ( p\circ\pi^{-1}\right ) \\ & \underset{\text{by ( \ref{eqcon.13})}}{=}\int_{x}\left\vert f\right\vert ^{2}\,d\mu=\left\vert f\right\vert _ { \mu}^{2}.\end{aligned}\ ] ] moreover,@xmath640{{}}}{=}\left ( f_{i}f\right ) \circ\pi\\ & \underset{\makebox[\equallabelwidth]{{}}}{=}\frac{1}{\sqrt{n}}f\circ \tau_{i}\circ\pi\\ & \underset{\text{by ( \ref{eqcon.11})}}{=}\frac{1}{\sqrt{n}}f\circ\pi \circ\sigma_{i}\\ & \underset{\makebox[\equallabelwidth]{{}}}{=}s_{i}^{\ast}vf,\end{aligned}\ ] ] which is assertion ( [ thmgen.isoext(2 ) ] ) .
_ ad _ ( [ thmgen.isoext(3 ) ] ) : let @xmath641 , and let @xmath642 and @xmath643 denote the respective hilbert inner products of @xmath59 and @xmath644 . to show that the space in ( [ eqgen.13 ] ) is dense in @xmath638 , suppose@xmath645 for all @xmath270 , all multi - indices @xmath646 , and all @xmath93 .
we will prove that then @xmath647 .
when @xmath646 is fixed , we denote the cylinder set in @xmath120 by@xmath648 using now ( [ eqove.10 ] ) in corollary [ corove.onto ] on @xmath120 , we get @xmath649 substitution into ( [ eqgen.14 ] ) yields @xmath650 we used the fact that ( [ eqgen.14 ] ) holds for all @xmath651 .
but the indicator functions @xmath652 span a dense subspace in @xmath633 when @xmath270 varies , and all finite words of length @xmath270 are used .
we conclude that @xmath647 , and therefore that the space in ( [ eqgen.13 ] ) is dense in @xmath633 .
baggett , p.e.t .
jorgensen , k.d .
merrill , and j.a . packer .
an analogue of bratteli - jorgensen loop group actions for gmra s . in c.
heil , p.e.t .
jorgensen , and d.r .
larson , editors , _ wavelets , frames , and operator theory ( focused research group workshop , college park , maryland , january 1521 , 2003 ) _ , volume 345 of _ contemp .
_ , pages 1125 .
american mathematical society , providence , 2004 .
dutkay and p.e.t .
methods from multiscale theory and wavelets applied to non - linear dynamics . in d. alpay , editor , _ wavelets , multiscale systems and hypercomplex analysis _ ,
volume 167 of _ oper .
theory adv .
_ , pages 87126 .
birkhuser , boston , 2006 . | an iterated function system ( ifs ) is a system of contractive mappings @xmath0 , @xmath1 ( finite ) where @xmath2 is a complete metric space .
every such ifs has a unique ( up to scale ) equilibrium measure ( also called the hutchinson measure @xmath3 ) , and we study the hilbert space @xmath4 . in this paper we extend previous work on ifss without overlap .
our method involves systems of operators generalizing the more familiar cuntz relations from operator algebra theory , and from subband filter operators in signal processing .
these cuntz - like operator systems were used in recent papers on wavelets analysis by baggett , jorgensen , merrill and packer , where they serve as a first step to generating wavelet bases of parseval type ( alias normalized tight frames ) , i.e. , wavelet bases with redundancy .
similarly , it was shown in work by dutkay and jorgensen that the iterative operator approach works well for generating wavelets on fractals from ifss without overlap .
but so far the more general and more difficult case of essential overlap has resisted previous attempts at a harmonic analysis , and explicit basis constructions in particular .
the operators generating the appropriate cuntz relations are composition operators , e.g. , @xmath5 where @xmath6 is the given ifs .
if the particular ifs is essentially non - overlapping , it is relatively easy to compute the adjoint operators @xmath7 , and the @xmath8 operators will be isometries in @xmath4 with orthogonal ranges . for the case of essential overlap , we can use the extra terms entering in the computation of the operators @xmath9 as a measure of the essential overlap for the particular ifs we study . here
the adjoint operators @xmath9 refers to the hilbert space @xmath4 where @xmath3 is the equilibrium measure @xmath3 for the given ifs @xmath6 . |
the first formulas for the multiplicities of singular points on schubert varieties in grassmannians appeared in abhyankar s results @xcite on the hilbert series of determinantal varieties ( recall that a determinantal variety gets identified with the opposite cell in a suitable schubert variety in a suitable grassmannian ) .
herzog - trung @xcite generalized these formulas to give determinantal formulas for the multiplicities at the identity of all schubert varieties in grassmannians . using standard monomial theory ,
lakshmibai - weyman @xcite obtained a recursive formula for the multiplicities of all points in schubert varieties in a minuscule @xmath3 ; rosenthal - zelevinsky @xcite used this result to obtain a closed - form determinantal formula for multiplicities of all points in grassmannians .
let @xmath4 be the base field , which we assume to be algebraically closed , of arbitrary characteristic .
let @xmath5 be @xmath6 , @xmath7 the subgroup of diagonal matrices in @xmath5 , and @xmath8 the subgroup of upper diagonal matrices in @xmath5 .
let @xmath9 be the root system of @xmath5 relative to @xmath7 , and @xmath10 the set of positive roots relative to @xmath8 .
let @xmath11 be the weyl group of @xmath5 .
note that @xmath12 , the group of permutations of the set of @xmath13 elements .
let @xmath14 be the maximal parabolic subgroup @xmath15 let @xmath16 , @xmath17 , and @xmath18 denote respectively the root system , set of positive roots , and weyl group of @xmath19 .
the quotient @xmath20 , with the bruhat order , is a distributive lattice .
the map @xmath21 taking a positive root to its corresponding reflection , embeds @xmath22 in @xmath20 .
we shall also denote the image by @xmath23 .
it is a sublattice of @xmath20 .
a _ multiset _ is similar to a set , but with repetitions of entries allowed .
define the _ cardinality _ of a multiset @xmath24 , denoted by @xmath25 , to be the number of elements in @xmath24 , including repetitions .
uniset _ to be a multiset which has no repetitions .
if @xmath26 is a set , define @xmath27 to be the collection of all multisets which are made up of elements of @xmath26 .
a _ chain of commuting reflections _ in @xmath28 is a nonempty set of pairwise - commuting reflections @xmath29 , @xmath30 , such that @xmath31 ; we refer to @xmath32 as the _ length _ of the chain . for a multiset @xmath33 ,
define the _ chainlength _ of @xmath24 to be the maximum length of a chain of commuting reflections in @xmath24 .
fix @xmath34 .
define @xmath35 to be the multisets @xmath24 of @xmath36 , such that the product of every chain of commuting reflections in @xmath24 is less than or equal to @xmath37 ; similarly , define @xmath38 to be the unisets of @xmath39 having the same property .
for @xmath40 a positive integer , define @xmath41 denote the schubert variety of @xmath42 corresponding to @xmath43 , the hilbert function of the tangent cone to @xmath44 at the identity is given by [ main1 ] @xmath45 .
second , letting @xmath46 denote the maximum cardinality of any element of @xmath38 , the multiplicity at the identity is given by [ main2 ] @xmath47 .
let @xmath8 be a graded , affine @xmath4-algebra such that @xmath48 generates @xmath8 ( as a @xmath4-algebra ) .
let @xmath49 .
the function @xmath50 ( or @xmath51 ) @xmath52 , @xmath53 is called the _ hilbert function _ of @xmath54 .
there exists a polynomial @xmath55 ( or @xmath56 ) @xmath57 $ ] , called the _
hilbert polynomial _ of @xmath58 , such that @xmath59 for @xmath60 .
let @xmath61 denote the degree of @xmath55 .
then @xmath62 , and the leading coefficient of @xmath55 is of the form @xmath63 , where @xmath64 .
the integer @xmath65 is called the _ degree of @xmath0 _ , and denoted @xmath66 . in the sequel we shall also denote @xmath66 by @xmath67 .
let @xmath0 be an algebraic variety , and let @xmath68 .
let @xmath69 be the stalk at @xmath70 and @xmath71 the unique maximal ideal of the local ring @xmath72 .
then the _ tangent cone _ to @xmath0 at @xmath70 , denoted @xmath73 , is defined to be @xmath74 , where @xmath75 .
multiplicity _ of @xmath0 at @xmath70 , denoted @xmath76 , is defined to be @xmath77 .
if @xmath78 is an affine closed subvariety , and @xmath79 $ ] is the maximal ideal corresponding to @xmath68 , then @xmath80 , m_p ) = \textrm{gr}(a,\mathfrak{m})$ ] .
let @xmath72 be the polynomial ring @xmath81 $ ] .
a _ monomial order _
@xmath82 on the set of monomials in @xmath72 is a total order such that given monomials @xmath40 , @xmath83 , @xmath84 , @xmath85 , @xmath86 , we have @xmath87 and @xmath88 . the largest monomial ( with respect to @xmath82 ) present in a polynomial @xmath89 is called the _ initial term _ of @xmath90 , and is denoted by @xmath91 . the _ lexicographic order _ is a total order defined in the following manner .
assume the variables @xmath92 are ordered by @xmath93 .
a monomial @xmath40 of degree @xmath61 in the polynomial ring @xmath72 will be written in the form @xmath94 , with @xmath95 .
then @xmath96 in the lexicographic order if and only if either @xmath97 , or @xmath98 and there exists an @xmath99 such that @xmath100 .
it is easy to check that the lexicographic order is a monomial order . given an ideal @xmath101 , denote by @xmath102 the ideal generated by the initial terms of the elements in @xmath103 .
a finite set @xmath104 is called a _ grbner basis _ of @xmath103 ( with respect to the monomial order @xmath82 ) , if @xmath102 is generated by the initial terms of the elements of @xmath105 .
given a monomial order and an ideal @xmath101 , there exists a flat family over @xmath106)$ ] whose special fiber @xmath107 is @xmath108 and whose generic fiber ( @xmath32 invertible ) is @xmath109)$ ] .
further , if @xmath103 is homogeneous , then the special fiber and generic fiber have the same hilbert function ( see @xcite for details ) .
let @xmath110 be such that @xmath111 .
the _ grassmannian _
@xmath112 is the set of all @xmath113dimensional subspaces @xmath114 .
let @xmath115 be an element of @xmath112 and @xmath116 a basis of @xmath115 , where each @xmath117 is a vector of the form @xmath118 thus , the basis @xmath119 gives rise to an @xmath120 matrix @xmath121 of rank @xmath110 , whose columns are the vectors @xmath122 .
we have a canonical embedding @xmath123\ ] ] called the _ pl
" ucker embedding_. let @xmath124 then the projective coordinates ( _ plcker coordinates _ ) of points in @xmath125 may be indexed by @xmath126 ; for @xmath127 , we shall denote the @xmath128-th component of @xmath129 by @xmath130 , or @xmath131 .
if a point @xmath115 in @xmath132 is represented by the @xmath133 matrix @xmath72 as above , then @xmath134 , where @xmath135 denotes the @xmath136 submatrix whose rows are the rows of @xmath72 with indices @xmath137 , in this order .
let @xmath5 , @xmath7 , @xmath8 , and @xmath14 be as in section [ summary ] .
let @xmath139 be the standard basis for @xmath140 .
for the natural action of @xmath5 on @xmath125 , the isotropy group at @xmath141 $ ] is @xmath14 , while the orbit through @xmath141 $ ] is @xmath112 .
thus we obtain an identification of @xmath142 with @xmath132 .
we also note that @xmath143 may be identified with @xmath126 . for the action of @xmath5 on @xmath132 , the @xmath7-fixed points are precisely @xmath144 , \underline{i } \in i_{d , n } \}$ ] , where @xmath145 .
schubert variety _
@xmath146 associated to @xmath128 is the zariski closure of the @xmath8-orbit @xmath147 $ ] with the canonical reduced scheme structure .
we have a bijection between @xmath148 and @xmath149 . the partial order on schubert varieties given by inclusion
induces a partial order ( called the _ bruhat order _ ) on @xmath150 ; namely , given @xmath151 , @xmath152 , @xmath153 we note the following facts for schubert varieties in the grassmannian ( see @xcite or @xcite for example ) : * * bruhat decomposition * : @xmath154 $ ] . * * dimension * : @xmath155 . * * vanishing property of a plcker coordinate * : @xmath156 a monomial @xmath157 , @xmath158 is said to be _ standard _ if @xmath159 such a monomial is said to be _
standard on the schubert variety @xmath160 _ , if in addition to ( [ standard ] ) , we have @xmath161 .
let @xmath162 .
let @xmath163 $ ] , the homogeneous coordinate ring for @xmath44 , for the plcker embedding . recall the following two results from standard monomial theory ( cf .
@xcite ) . [ rwbasis ]
the set of standard monomials on @xmath44 of degree @xmath40 is a basis for @xmath164 .
[ iwgen ] for @xmath162 , let @xmath165 be the ideal in @xmath166 $ ] generated by @xmath167
. then @xmath168/i_w$ ] .
let @xmath169 denote the unipotent lower triangular matrices of @xmath170 . under the canonical projection @xmath171)$ ] ,
@xmath169 maps isomorphically onto its image @xmath172 $ ] .
the set @xmath172 $ ] is called the _
opposite big cell _ in @xmath132 , and is denoted by @xmath2 .
thus , @xmath2 may be identified with @xmath173 thus we see that @xmath2 is an affine space of dimension @xmath174 , with @xmath175 as the origin ; further @xmath176 $ ] can be identified with the polynomial algebra @xmath177 $ ] .
to be very precise , denoting the elements of @xmath9 as in @xcite , we have @xmath178 , @xmath179 ; given @xmath180 , say @xmath181 , we identify @xmath182 with @xmath183 .
we denote by @xmath184 the reflection corresponding to @xmath185 , namely , the transposition switching @xmath186 and @xmath187 .
let @xmath188 .
we shall denote the plcker coordinate @xmath189 by @xmath190 .
let us denote a typical element @xmath191 by @xmath192 .
then @xmath190 is simply a minor of @xmath0 as follows .
let @xmath193 , and let @xmath194 be the largest entry @xmath195 .
let @xmath196 be the complement of @xmath197 in @xmath198 .
then this minor of @xmath0 is given by column indices @xmath199 and row indices @xmath200 ( here the rows of @xmath0 are indexed as @xmath201 ) .
conversely , given a minor of @xmath0 , say , with column indices @xmath202 , and row indices @xmath203 , then that minor is the evaluation of @xmath190 at @xmath0 , where @xmath193 may be described as follows : @xmath204 is the complement of @xmath205 in @xmath206 , and @xmath207 are simply the row indices ( again , the rows of @xmath0 are indexed as @xmath201 ) .
note that if @xmath208 , then @xmath209 evaluated at @xmath0 is 1 . in the above discussion
, therefore , we must consider the element 1 ( in @xmath176 $ ] ) as the minor of @xmath0 with row indices ( and column indices ) given by the empty set .
consider @xmath210 .
then @xmath211 on @xmath2 , we have @xmath212 , @xmath213 , @xmath214 , @xmath215 , @xmath216 , @xmath217 . note that each of the plcker coordinates is homogeneous in the local coordinates @xmath183 .
in view of the bruhat decomposition , in order to determine the multiplicity at a singular point @xmath219 , it is enough to determine the multiplicity of the @xmath7-fixed point in the @xmath8 orbit @xmath220 . in this section
, we shall discuss the behavior at a particular @xmath7-fixed point , namely the identity .
we define @xmath221 to be @xmath222 . since @xmath223 is open dense , and
@xmath224 , we have that @xmath225 . as a consequence of theorem [ iwgen ]
, @xmath226 is defined as an algebraic subvariety by the homogeneous polynomials @xmath227 ; further , @xmath228 corresponds to the origin .
thus we have that @xmath229,m_{id } ) = k[y(w)]$ ] . hence , @xmath230,m_{id } ) ) { } \nonumber\\ & & { } = \textrm{spec}(k[y(w ) ] ) = y(w).\end{aligned}\ ] ] for a monomial @xmath231 $ ] , define _ @xmath232 _ to be the multiset @xmath233 .
it follows immediately from the definition that multisupp gives a bijection between the monomials of @xmath176 $ ] and the multisets of @xmath39 , pairing the square - free monomials with the unisets .
let @xmath34 .
we call a monomial _ w - good _ if it maps under multisupp to an element of @xmath35 .
note that the w - good square - free monomials are precisely those which map to @xmath38 .
define a monomial order @xmath82 on @xmath176 $ ] in the following manner .
we say @xmath234 if @xmath235 , or if @xmath236 and @xmath237 .
note that this extends the partial order @xmath238 ( in the bruhat order ) .
the monomials are then ordered using the lexicographic order .
define the monomial ideal @xmath239 $ ] to be the ideal generated by @xmath240 , and let @xmath241/j_w$ ] .
with our ordering , @xmath242 is a commuting chain of reflections whose product is @xmath243 .
thus the non w - good monomials form a vector space basis for @xmath244 , and therefore the w - good monomials form a basis for @xmath245 . in view of ( [ tc ] ) and the above discussion , theorem [ main1 ] follows immediately from [ main ] @xmath246}(m ) = h_{a_w}(m ) \
, , m \in \mathbb{n}$ ] .
theorem [ main2 ] is also a consequence .
indeed , @xmath247 ) = \textrm{deg}(k[y(w ) ] ) = \textrm{deg}(a_w).\ ] ] since @xmath245 is an affine quotient of an ideal generated by square - free monomials , letting @xmath46 be the maximum degree of a square - free monomial in @xmath245 , we have ( cf .
@xcite ) @xmath248 \ , : \ , p \textrm { is a square - free w - good monomial } \textrm { and deg}(p)=m\}| \\&\,&\quad = |\{s \in s_w ' \ , : \
, the proof of lemma [ main ] relies on an inductive argument which shows directly that both functions agree for all positive integers @xmath40 .
note that @xmath249 = k[x(w)]_{(p_{id})}$ ] .
thus , as a consequence of theorem [ rwbasis ] , @xmath249 $ ] has a basis consisting of monomials of the form @xmath250 . if @xmath251 denotes the basis elements of degree @xmath40 , then @xmath252 .
letting @xmath253 be the degree of @xmath37 ( see section [ combin ] below for definition ) , as a consequence of standard monomial theory we have @xmath254 where @xmath255 , the union being taken over the divisors @xmath256 of @xmath44 ( cf .
@xcite ) .
we have that @xmath257 can be set - theoretically written as the integral linear combination of terms of the form @xmath258 and terms of the form @xmath259 .
further , it can be shown that @xmath260 where @xmath243 is given by @xmath261 .
( note that @xmath149 being a distributive lattice implies that for @xmath262 , @xmath263 is irreducible . )
thus , @xmath264 taking cardinalities of both sides of ( [ smred ] ) , we obtain @xmath265}(m+d ) = h_{k[y(w)]}(m ) + \!\ ! \sum_{w ' < w } \!a_{w ' \!}h_{k[y(w')]}(m+d).\ ] ] equivalently , @xmath246}$ ] satisfies the difference equation @xmath266 to prove lemma [ main ] , it suffices to show that @xmath267 satisfies ( [ diffeq ] ) for all @xmath268 , since it is a straightforward verification that @xmath246}(m)$ ] and @xmath267 have the same initial conditions .
as stated earlier , @xmath269 $ ] has as basis the w - good monomials of @xmath176 $ ] , which are in bijection with the elements of @xmath35 .
thus @xmath270}(m ) = |s_w(m)|$ ] , and it suffices to show that @xmath271 satisfies ( [ diffeq ] ) .
we can write @xmath272 where @xmath273 , the union being over the divisors @xmath256 of @xmath44 .
following the identical arguments used to deduce ( [ smh ] ) ( replacing `` @xmath274 '' by `` @xmath24 '' everywhere ) , one obtains @xmath275 for the same integers @xmath276 as in ( [ smh ] ) .
establishing an explicit bijection between @xmath277 and @xmath278 completes the proof , for then ( taking cardinalities of both sides of ( [ sred ] ) ) , one sees that @xmath267 satisfies ( [ diffeq ] ) for all @xmath279 . in view of
the discussion of flat deformations in section [ flatdef ] , lemma [ main ] also implies the set @xmath280 $ ] forms a grbner basis for the ideal it generates .
we call a multiset @xmath24 of @xmath36 a _ t - multipath _ , if the chainlength of @xmath24 is @xmath32
. if @xmath24 has no repeated elements ( i.e. it is a uniset ) , then we call it a _ t - unipath_. define @xmath281 to be a _
chain - maximal element of s _ if there is no element in @xmath24 strictly greater than @xmath282 which commutes with @xmath282 .
any t - multipath s can be written in the following manner as the union of @xmath32 nonintersecting 1-multipaths : if @xmath283 is the @xmath284 1-multipath , then @xmath285 is the multiset of chain - maximal elements ( including repetitions ) of @xmath286 ( for @xmath287 , where @xmath288 is defined to be the empty set ) . if the t - multipath s is a t - unipath , then each @xmath283 will be a 1-unipath .
fix @xmath34 .
there is a unique expression @xmath289 such that @xmath290 for all k , and all the reflections pairwise commute ; @xmath291 is called the _ degree _ of @xmath37 .
let @xmath292 .
then @xmath293 , where @xmath294 is a chain of commuting reflections .
thus @xmath295 .
let @xmath296 .
we say that a t - multipath @xmath24 is w - good if , when written as the union of weighted 1-multipaths @xmath297 as above , we have that the elements of @xmath298 are in @xmath299 . any multiset in @xmath39 is a t - multipath for some t ; it is said to be w - good if the corresponding t - multipath is w - good .
it can be seen that the combinatorial property that a multiset ( resp .
uniset ) @xmath24 of @xmath300 is w - good is equivalent to the group - theoretic property that @xmath301 ( resp .
@xmath302 ) . thus theorem [ main1 ] is equivalent to the assertion that @xmath303 is the number of w - good multisets of @xmath300 of degree m. letting @xmath46 be the maximum cardinality of a w - good uniset , theorem [ main2 ] is equivalent to the assertion that @xmath304 is the number of w - good unisets of cardinality @xmath46 .
let @xmath305 .
we have that @xmath306 is a chain of commuting reflections , and thus @xmath295 .
the diagram below shows the lattice @xmath307 , where the reflection @xmath308 is denoted by @xmath309 .
the set @xmath24 of reflections which lie along the three broken - line paths is an example of a @xmath37-good uniset of maximum cardinality .
in fact , any @xmath37-good uniset of maximum cardinality can be seen as the set of reflections lying on three paths in the lattice , satisfying the following properties : * one path starts and ends at `` x '' , the second at `` y '' , and the third at `` z '' .
* each path can move only down or to the right . * the paths
do not intersect .
thus the number of ways of drawing three such paths is @xmath304 .
let @xmath310 . define @xmath311 to be the multisets @xmath24 of @xmath300 , such that for every chain of commuting reflections @xmath312 , we have that @xmath313 ; define @xmath314 to be the unisets of @xmath300 having the same property .
for @xmath40 a positive integer , define @xmath315 @xmath316 | we give a closed - form formula for the hilbert function of the tangent cone at the identity of a schubert variety @xmath0 in the grassmannian in both group theoretic and combinatorial terms .
we also give a formula for the multiplicity of @xmath0 at the identity , and a grbner basis for the ideal defining @xmath1 as a closed subvariety of @xmath2 , where @xmath2 is the opposite cell in the grassmannian .
we give conjectures for the hilbert function and multiplicity at points other than the identity . |
different viral types have phylogenetic trees exhibiting different branching properties , with influenza and hiv being two extreme examples . in the influenza tree
a single type dominates for a long time with other types dying out quickly until suddenly a new type completely takes over and the old type dies out .
the hiv phylogeny is the complete opposite , with a large number of co existing types . in @xcite
a stochastic model is described and depending on the choice of parameters it can exhibit both types of dynamics .
we briefly describe the model after @xcite .
we only keep track of the number of different viral types at each time point @xmath0 .
let @xmath1 denote the number of distinct viral types at time @xmath0 . in the nomenclature of phylogenetics
@xmath1 counts the number of different species alive at time @xmath0 . at each time point
the birth rate is @xmath2 and the death rate is @xmath1 . if there is only one type alive then it can not die .
clearly @xmath1 is a markov chain with discrete state space and continuous time .
each virus type is described by a fitness value that is randomly chosen at its birth . if a death event occurs the type with smallest fitness dies .
this means that only the fitness ranks matter and so the exact distribution of a virus fitness will not play a role .
the main result of @xcite is the asymptotic behaviour of the dominating type , whether it is expected to remain the same for long stretches of time or change often .
this is summarized in theorem 1 @xcite , take @xmath3 .
if @xmath4 then @xmath5 while if @xmath6 then this limit is @xmath7 .
the proof of this theorem is based on considering successive visits to the state @xmath8 , in particular denote @xmath9 to be the ( random ) times between visits of the chain to @xmath10 and @xmath11 . in @xcite the latter random variable
is represented as @xmath12 where @xmath13 are independent mean @xmath8 exponential random variables and @xmath14 are the ( independent ) hitting times of the state @xmath8 conditional on starting in state @xmath15 .
descriptively @xmath14 is the @xmath16th return from state @xmath15 to the state of one virus type alive , since from @xmath8 the chain has to jump to two types .
the markovian nature of the process ensures that the @xmath14s are independent and identically distributed for distinct @xmath16s . in the proof of theorem 1
it is stated that the cumulative distribution function of @xmath14 , @xmath17 , satisfies , @xmath18 solved uniquely for , @xmath19 this gives the asymptotic behaviour ( lemma @xmath20 @xcite ) , @xmath21 from which the result of theorem 1 is derived when @xmath22 .
however eq . ( [ eqligintft ] ) does not take into account that this model differs from a classical birth death model where @xmath7 is the absorbing state .
we illustrate this in fig .
we can easily re numerate the state values , but the intensity values will differ between the two models . ) would be correct .
the numbers inside the circles are the states ( counting the number of virus type ) and the numbers above and below the arrows are the birth and death rates respectively of the state from which they come out.,title="fig : " ] + ) would be correct .
the numbers inside the circles are the states ( counting the number of virus type ) and the numbers above and below the arrows are the birth and death rates respectively of the state from which they come out.,title="fig : " ] correcting for this difference in intensity values one will still get the same asymptotics as in eq .
( [ eqlignt ] ) and hence the same result as in but with a more complicated proof .
below we present a correct proof of lemma @xmath20 @xcite in the case of @xmath22 , based on lemmas [ lemrenewal ] and [ lemftasympt ] . from this
the statement of theorem @xmath8 of @xcite follows .
[ lemrenewal ] let @xmath23 .
then @xmath24 solves the renewal equation @xmath25 in panel a of fig .
[ fgbd ] we can see a representation of the studied markov chain on the state space @xmath26 . due to the @xmath14s being independent for different @xmath16s we can study the distribution of @xmath27 and treat @xmath8 as an absorbing state .
let @xmath28 denote the probability of being in state @xmath29 at time @xmath0 , when one starts in state @xmath15 at time @xmath7 .
the system of differential equations describing the probabilities is , @xmath30 with initial conditions , @xmath31 let @xmath32 denote the generating function of the sequence @xmath33 , i.e. @xmath34 taking first derivatives we get the partial differential equation , @xmath35 with initial conditions @xmath36 following @xmath37 @xcite and using the substitution @xmath38 with @xmath39 we arrive at , @xmath40 evaluating the derivative of both sides with respect to @xmath41 at @xmath7 we find that the function @xmath24 must satisfy the following integral equation ( we can recognize it as a renewal equation ) , @xmath42 [ lemftasympt ] @xmath43 , the solution of the renewal equation ( [ eqrenewal ] ) , has the following properties , @xmath44 @xmath45 the proof is based on tauberian theory and we refer the reader to @xcite for details on this .
another approach would be to study the asymptotic behaviour of @xmath46 by renewal theory results ( see e.g. @xcite ) .
the laplace transform of a density function @xmath47 , denoted @xmath48 is , @xmath49 we will use the following theorem from @xcite , theorem @xmath15 xiii.5 , [ tauberian ] if @xmath50 is slowly varying at infinity and @xmath51 , then each of the relations @xmath52 @xmath53 implies the other .
@xmath54 defined as the solution to the renewal equation ( [ eqrenewal ] ) after differentiating will satisfy @xmath55 we calculate the laplace transforms of @xmath46 , @xmath56 and @xmath57 , @xmath58 we are interested in the behaviour of the transforms as @xmath59 , and for this we will use the well known property ( verifiable by the de lhspital rule ) of the exponential integral , @xmath60 to arrive at , @xmath61 as both the constant function @xmath62 and @xmath63 are slowly varying functions for @xmath59 the tauberian theorem allows us to conclude that , @xmath64
we will now use lemmas [ lemrenewal ] and [ lemftasympt ] to prove lemma 3 from @xcite . define as there , @xmath65 with @xmath66 . by lemma [ lemftasympt ]
we know that , @xmath67 we now need to check how @xmath68 behaves asymptotically .
we do not know what @xmath69 is but using the tauberian theorem and eq .
( [ twf ] ) from the proof of lemma [ lemftasympt ] we get that , @xmath70 therefore using integration by parts and eq .
( [ eqt2f ] ) , @xmath71 and so we arrive at @xmath72 the rest of the proof is a direct repeat of the one in @xcite and so we get ( as in @xcite ) that @xmath73 tends to @xmath8 in probability implying theorem 1 @xcite for @xmath22 . if we applied the same chain of reasoning to the model of panel b in fig .
[ fgbd ] starting off with the system of differential equations , the analogue of eq .
( [ partdiffeq ] ) would be a homogeneous partial differential equation , @xmath74 with initial conditions @xmath75 in agreement with @xmath76 .
it would therefore be an interesting problem to see what conditions are necessary on the nonhomogeneous part of eq .
( [ partdiffeq ] ) to still get the same asymptotic behaviour of the markov chain and what underlying model properties do these conditions imply .
we are grateful to wojciech bartoszek and joachim domsta for many helpful comments , insights and suggestions .
k.b . was supported by the centre for theoretical biology at the university of gothenburg , stiftelsen fr vetenskaplig forskning och utbildning i matematik ( foundation for scientific research and education in mathematics ) , knut and alice wallenbergs travel fund , paul and marie berghaus fund , the royal swedish academy of sciences , wilhelm and martina lundgrens research fund and stersjsamarbete scholarship from svenska institutet ( 00507/2012 ) . | birth and death models are now a common mathematical tool to describe branching patterns observed in real world phylogenetic trees .
liggett and schinazi ( 2009 ) is one such example .
the authors propose a simple birth and death model that is compatible with phylogenetic trees of both influenza and hiv , depending on the birth rate parameter .
an interesting special case of this model is the critical case where the birth rate equals the death rate .
this is a non trivial situation and to study its asymptotic behaviour we employed the laplace transform . with this
we correct the proof of liggett and schinazi ( 2009 ) in the critical case . |
the evaluation of radiative corrections in the minimal supersymmetric standard model ( mssm ) plays an important role in our understanding of its phenomenological implications .
this is of particular interest , the mssm being , at the moment , the most promising extension of the standard model ( sm ) for the strong and electroweak interactions .
finite corrections to supersymmetric ward identities , such as those associated with higgs boson masses and couplings @xcite , have a direct impact on the model predictions .
the remaining corrections have more subtle phenomenological implications , since they can be largely reabsorbed into redefinition of the many arbitrary parameters of the mssm .
this is no longer true , however , if one embeds the model into a more fundamental theory ( as superguts , supergravities or superstrings ) where some or all of the mssm parameters can be predicted . in this paper
we compute the one - loop corrections to the masses of the top quark , the stop squarks and the gluino , going beyond the leading - log approximation .
we present explicit analytical formulae , keeping in the feynman rules only the vertices proportional to the strong coupling constant @xmath1 and to the top yukawa coupling @xmath2 .
this approximation should reproduce the main contributions to the one - loop radiative corrections . comparing our full one - loop results with the renormalization - group - improved tree - level masses
, we find that there can be significant differences , also in the case of `` natural '' choices for the renormalization scale .
the corrections can be particularly relevant for the lightest stop eigenstate and for a very light gluino . in the top quark case ,
we find that the extra corrections due to loops of supersymmetric particles can be non - negligible compared to the pure qcd ones .
the paper is organized as follows : in this section , we introduce the notation by giving the top , stop and gluino tree - level mass formulae ; in section [ sec : loop ] we introduce the renormalization procedure , we recall the relation between running and pole masses , and we give the main points of our calculations ; in section [ sec : nume ] we discuss the numerical results and their phenomenological implications . the complete results for the one - loop - corrected top , stop and gluino masses are presented in appendix a. the explicit solutions of the one - loop renormalization group equations for all the parameters of relevance to our calculations , in the approximation where the only non - zero dimensionless couplings are @xmath1 and @xmath2 , are reported in appendix b. the mssm @xcite contains three generations of quark ( @xmath3 ) and lepton ( @xmath4 ) chiral superfields ( @xmath5 ) , and two higgs chiral superfields ( @xmath6 ) , with interactions specified by an @xmath7-parity - invariant superpotential and a collection of soft supersymmetry - breaking terms . as often done in the literature , we shall assume real mass parameters and neglect all yukawa couplings apart from the top quark one .
the tree - level expressions for the different masses and couplings in the mssm are well known : in the following , we shall use the notation and conventions of @xcite , including the reversal of the sign of @xmath8 in the squark sector as specified in the errata @xcite .
we now recall the tree - level expressions for the top and stop masses ( the tree - level gluino mass is just the explicit soft mass itself ) . with no loss of generality , the vacuum expectation values ( vevs ) of the higgs fields , @xmath9 ,
can be chosen to be real and positive .
the top quark mass , then , is given by : @xmath10 the mass matrix for the stop squarks is : @xmath11 where @xmath12 are soft mass terms , @xmath13 , @xmath8 is the superpotential mass parameter and , at the tree level , @xmath14 the eigenvalues of ( [ stopmatrix ] ) are : @xmath15 ^ 2 + m^2_t ( a_t - \mu \cot
\beta ) ^2 } .\end{aligned}\ ] ] the corresponding eigenstates are : @xmath16 where the rotation angle @xmath17 that diagonalizes the mass matrix is given by : @xmath18 } , \ ] ] with @xmath19 .
our conventional ordering of the eigenvalues ( namely , @xmath20 ) is obtained with the prescription that @xmath17 should have the same sign as the combination @xmath21 . summarizing , at the tree level the top , stop and gluino masses are functions of the following independent input parameters : @xmath22 and @xmath23 .
as we shall see in the next section , other parameters will appear via the one - loop corrections .
at the one - loop level , the pole masses of the top and the gluino are given by : @xmath24 \right|_{\slash{p } = m_t } , \\
\label{gluepolemass } \tilde m_g & = & \left .
[ \tilde m_g(q ) - \pi_{\tilde g}(\slash{p},q ) ] \right|_{\slash{p } = \tilde m_g}.\end{aligned}\ ] ] in the above equations , @xmath25 is the renormalization scale ; @xmath26 are the running masses in some renormalization scheme , while @xmath27 are the real parts of the renormalized self - energies in the same scheme .
up to higher - order corrections , the pole masses do not depend on the scale or on the renormalization scheme .
to evaluate the corrections to the pole masses of the stops , we need to study the corrected inverse propagator matrix @xmath28 . in the ( @xmath29 ) basis , where @xmath30 are the tree - level eigenstates ,
the expression for the inverse propagator is : @xmath31 at the one loop , the stop pole masses are : @xmath32 in eqs .
( [ toppolemass ] ) , ( [ gluepolemass ] ) , ( [ stopmass1 ] ) and ( [ stopmass2 ] ) , the running masses can be calculated using the tree - level functional forms ( presented in section [ sec : intro ] ) , with all the parameters considered as @xmath25-dependent . to evaluate the one - loop self - energies of top , stops and gluino , we work in the approximation where all dimensionless couplings other than @xmath1 and @xmath2 are set to zero , which should take into account the most relevant one - loop corrections to the studied observables .
the feynman graphs that represent the self - energies of top , stops and gluino in this approximation are reported in figs .
[ topfeyn ] , [ stopfeyn ] and [ gluefeyn]b and [ gluefeyn]c must be counted twice , due to the fact that the gluino is a majorana fermion . ] .
the self - energies for the top ( @xmath33 ) and gluino ( @xmath34 ) read : @xmath35 while for the stops the self - energies @xmath36 can be written as : @xmath37 they have been grouped in subsets of graphs , to make the results more transparent and to simplify the numerical study that will be presented in the next section .
the masses of the particles circulating in each subset are controlled mainly by one input parameter , e.g. the contributions of the graphs with higgs and higgs / squark loops to the stop self - energies ( fig .
[ stopfeyn]d ) can be studied by varying @xmath38 , while neutralino and chargino loops ( figs .
[ stopfeyn]e and [ stopfeyn]f ) depend mainly on @xmath8 .
the self - energies have been calculated both in the feynman and in the landau gauge , with the same results . as usual , in order to explicitly preserve supersymmetry , we have adopted the @xmath39 scheme @xcite .
the complete results for the one - loop self - energies of the top , the two stops and the gluino are reported in appendix a. it can be seen by looking at the formulae that some new parameters must be introduced in order to calculate the full one - loop expressions .
the independent parameters that must be added to the ones that appear in the tree - level relations are @xmath40 and @xmath41 , where @xmath42 is the physical mass of the cp - odd neutral higgs boson and @xmath41 is the soft mass term for the right sbottom squark .
the masses of the other higgs bosons ( @xmath43 and @xmath44 ) and the cp - even mixing angle @xmath45 can be deduced with tree - level relations from @xmath46 and @xmath47 .
the masses of the sbottoms can be obtained from @xmath48 and @xmath47 .
all these tree - level relations can be found in the literature .
we have verified that the inclusion of terms proportional to @xmath49 and @xmath50 ( namely , @xmath51 and @xmath52 ) in the tree - level relations for the masses of the particles circulating in the loops does not significantly modify the results .
for this reason , we have neglected them in the tree - level relations for the charginos and neutralinos mass matrices .
this further approximation greatly simplifies the numerical implementation of the general formulae reported in appendix a. the charginos and neutralinos eigenvalues in this approximation are : @xmath53 while the rotation matrices are : @xmath54 in this approximation it can be seen , by looking at the formulae reported in appendix a , that the gaugino soft masses @xmath55 do not appear explicitly . therefore they have not been included in the input parameters set . in our approximation the electroweak gauge couplings @xmath49 and @xmath50 ( equivalent to @xmath56 and @xmath57 ) do not evolve under the renormalization group equations .
they have been fixed at the values @xmath58 and @xmath59 .
another fixed input parameter is the @xmath60 physical mass , @xmath61 gev .
the other two couplings ( @xmath62 ) have been chosen in order to give , at the electroweak scale , a value for the top pole mass that lies approximately in the experimental range .
we stress that all the input parameters must be interpreted as mssm parameters : e.g. @xmath63 is the mssm strong coupling constant , and not the sm one . in order to compare our full one - loop calculations with the results obtained in the leading - log approximation via the renormalization group equations ,
we have used the explicit solution of the rges that can be found in their general form in @xcite . in our approximation
we must neglect terms proportional to dimensionless couplings other than @xmath64 also in the rges , which therefore can be explicitly solved .
the details of the rges solutions are reported in appendix b. it must be noted that in those formulae another two parameters ( deducible from the already discussed set ) appear , @xmath65 and @xmath66 . for consistency , their expression in terms of the input parameters has been calculated at the one - loop level . to extract from the input parameters the one - loop vev @xmath65 , we have calculated the running @xmath39 mass @xmath67 from the physical @xmath60 mass @xmath68 using the following relation : @xmath69 in the already obtained expression for the one - loop self - energy @xmath70 @xcite , we have kept only terms @xmath71 to implement our approximation and @xmath50 , for this will result in putting @xmath72 . ] .
the feynman graphs that contribute to the @xmath60 self - energy in this approximation are reported in fig .
[ zetafeyn ] .
the decomposition of @xmath70 on the subsets of graphs is : @xmath73.\ ] ] the one - loop expression is presented for completeness in appendix a , jointly with our new one - loop results .
the one - loop vev @xmath65 is then obtained from @xmath67 and @xmath23 with the tree - level relation ( [ zetamass ] ) .
this procedure corresponds to the minimization of the one - loop effective potential : with this definition of the vevs , the one - loop tadpole terms disappear ( as the tree - level tadpoles disappear as a consequence of the tree - level scalar potential minimization ) . to extract from the input parameters the one - loop value for @xmath66 , we have used the one - loop expressions that relate @xmath74 and @xmath75 presented in eqs .
( 22),(23 ) and ( 24 ) of @xcite .
in the case of the stop squarks we have verified , as a check of the calculation , that the quadratic divergence of the sum of the full set of graphs is zero , as it should be in a softly broken supersymmetric theory . as a further check ,
we have verified analytically for all the studied masses that the implicit @xmath25-dependence of the running mass cancels with the explicit @xmath25-dependence of the one - loop self - energy .
the logarithmic dependence on @xmath25 for the renormalized gluino mass has also been checked by a comparison with the results in @xcite .
the one - loop correction to the top quark mass ( including the finite part ) in the @xmath39 scheme for the terms proportional to @xmath1 has been compared with @xcite . as a consistency check
, we have verified that if the off - diagonal contributions @xmath76 in eq .
( [ inverse ] ) are kept the one - loop - corrected pole masses do not significantly change .
in this section we describe in some detail the numerical results obtained for the one - loop - corrected masses of the top , the two stops and the gluino . in fig .
[ qq_all ] we report the scale - dependence of the @xmath39 running masses ( dashed lines ) and of the one - loop - corrected pole masses ( solid lines ) .
our representative choice of parameters , assigned conventionally at the scale @xmath77 gev , is : @xmath78 gev ; @xmath79 ; it can be seen that , in all four cases , the strong @xmath25-dependence of the running mass @xmath80 is significantly reduced .
this is an expected feature , due to the analytic cancellation of the logarithmic dependence in @xmath25 between the running expression and the one - loop self - energy .
it must be noted that the residual scale - dependence present in figs . [ qq_all]b and [ qq_all]c can be explained in the following way : using in the numerical calculations the neutralino and chargino masses reported in eqs .
( [ charginomass ] ) and ( [ neutralinomass ] ) , we introduce a partial non - cancellation of the scale - dependence between @xmath81 and @xmath82-dependence of the different sets of feynman graphs that appear in fig .
[ stopfeyn ] , we have seen that the scale - dependence is actually reintroduced adding graphs ( e ) and ( f ) to the others . ] . due to the improved scale - independence
, one can check if the usual procedure adopted to choose the optimal scale to evaluate the running mass ( @xmath83 ) can lead to a wrong calculation of the one - loop effects .
this is actually the case , as can be seen for example for the gluino or the lightest stop mass .
now we can start a detailed analysis of the complete results , comparing them with the running masses obtained at the `` natural '' scale @xmath83 .
we find that in many cases the largest corrections to the running masses come from the loops in which a gluino is circulating .
therefore we report , for all four masses under consideration , the dependence of the one - loop - corrected mass on the gluino mass .
we present also the dependence on other parameters in various interesting cases . 1 . in fig .
[ top_all ] we report the one - loop - corrected mass for the top . we have verified that the top mass corrections are almost independent from @xmath42 , which has then been kept fixed at @xmath84 gev .
the corrections have been calculated at the `` natural '' scale of @xmath85 gev .
we have studied two main situations : the case in which the two stop states are almost degenerate and the case in which the lightest stop state is `` very '' light ( between the experimental lower bounds of @xmath86 gev and 100 gev ) . to obtain degenerate stop states , we have chosen universal soft mass terms ( @xmath87 ) for the squark and @xmath88 gev .
this case is illustrated in fig .
[ top_all]a , for four representative values of @xmath89 , ( @xmath90 gev ) . in the same situation
, we have studied the dependence on @xmath47 .
the results are reported in fig .
[ top_all]b , with @xmath91 gev , @xmath92150 . to obtain a light stop state ,
we have still chosen universal soft mass terms ( @xmath87 ) for the squark , but @xmath93 .
the dependence of the top mass on @xmath42 in this situation is reported in fig .
[ top_all]c , with @xmath94200600 gev . in these last two figs .
@xmath95 gev .
it can be seen that in all the cases presented , the corrections to the running top mass are quite stable and not very sensitive to the variation of the parameters .
the results obtained for @xmath96 gev are reported in fig .
[ top_all]d .
the corresponding lightest stop mass is approximately @xmath97 gev . in this last figure , we have studied in detail the single contribution of the different classes of feynman graphs to the complete result , in the case of a `` very light '' @xmath98 .
the corrections have been compared with the qcd correction ( represented by the solid line that lies in the middle ) , which is the main sm contribution in the approximation @xmath99 .
the qcd correction to the running mass at @xmath100 gev is slightly less than 10 gev .
the dotted lines represent the different contributions added to the qcd corrections . starting from the qcd line , we can see the charginos , neutralinos , higgses and gluino contributions .
it can be seen that the largest correction ( to the qcd result ) comes from the gluino loops .
we stress that , in the different cases examined , the mssm corrections can reach half ( or more ) of the qcd correction .
also , the complete one - loop results can significantly differ from the running ones .
+ 2 . in fig .
[ st1_all ] we report the results for the `` heaviest '' stop state @xmath101 .
also for this particle we have studied the two cases of almost degenerate stop states and of a `` very light '' @xmath98 . in fig .
[ st1_all]a we consider the case of degenerate stop states at three different values of the universal soft mass term @xmath102 gev . as usual the dashed lines represent the running masses .
the three results have been obtained at the `` natural '' scales @xmath103 gev . in figs . [
st1_all]b and [ st1_all]d we have reported on the dependence on @xmath42 and @xmath8 in the case of a `` very light '' @xmath98 at the scale @xmath104 gev ( with @xmath105 gev , @xmath106 gev ) .
in particular , in [ st1_all]b @xmath107 gev and @xmath94200600 gev while in [ st1_all]d @xmath108 gev and @xmath109200600 gev . in fig .
[ st1_all]c , is reported on the dependence of @xmath110 on the gluino mass in the same situation , with @xmath111 gev and @xmath112 gev . in this case , we have studied also the contributions coming from the different loops .
the horizontal dotted lines are ( from the bottom ) the neutralinos , charginos , higgses , sbottoms and stops contributions , while the dotted line that follows the complete result is the gluino contribution .
it can be easily seen that also in this case the most important contribution comes from the gluino loops .
actually , in all the cases that we have observed , the corrections to the @xmath110 running mass are not very large .
we have seen that for almost degenerate running states , the one - loop - corrected masses can be interchanged ( namely , @xmath98 can become the heaviest stop state ) .
+ 3 . in fig .
[ st2_all ] we report the results for @xmath113 with the same choices of parameters as for @xmath110 . in fig .
[ st2_all]a , it can be seen that @xmath98 has a behaviour quite similar to @xmath101 : actually , the one - loop corrections can interchange the two states ( @xmath98 can become the heaviest one ) .
also for @xmath114 , the corrections to the running mass ( evaluated at the same `` natural '' scales ) are quite small . in figs .
[ st2_all]b , [ st2_all]c and [ st2_all]d we have studied the case of a `` very light '' @xmath98 , at the `` natural '' scale @xmath115 gev .
it can be seen in all three figures that the one - loop corrections significantly modify the renormalization - group - improved tree - level results .
the most interesting situation is reported in fig .
[ st2_all]c . at the `` natural '' scale of @xmath116 gev
, we have chosen a universal soft mass term @xmath117 gev , in order to get a sufficiently heavy running mass , @xmath118 gev .
it can be seen from the figure that the one - loop - corrected mass is significantly different from the running mass , and that can easily fall below the experimental lower bound of @xmath86 gev also for a non - super - heavy gluino mass .
this implies that in the case of a light stop state the running - mass result must be used ( also at the `` natural '' scale ) with great care . as usual , we have studied the different contributions to this case .
starting from the bottom , the horizontal dotted lines are the contributions coming from neutralinos , charginos , sbottoms , stops and higgses loops . the dotted line with the same shape as the complete result represents the gluino contribution .
+ 4 . in fig .
[ glue_all ] we report the results for the one - loop - corrected gluino mass in two different cases : a heavy gluino and a massless gluino .
our results agree with the one - loop corrections calculated in @xcite , and contain as a particular case the results of @xcite for a gluino massless at tree level . in [ glue_all]a and
[ glue_all]b we have studied the dependence on @xmath8 and @xmath47 of the corrections to a heavy gluino mass , for a running mass @xmath91 gev at the `` natural '' scale @xmath104 gev .
it has been found that the corrections increase with increasing scalar soft mass terms @xmath119 , and with decreasing @xmath47 .
they can be of order 1030% of the running mass , for stop masses not exceeding 1 tev . in figs .
[ glue_all]c and [ glue_all]d we report results for the massless case ( for which we have used @xmath120 ) . here
the situation is reversed , and the corrections increase with decreasing scalar soft mass terms .
it is well known that for a massless gluino the corrections at one loop are proportional to the mass differences between the squarks .
this has been verified by giving non - universal soft mass terms , keeping the average scale @xmath89 at a small value .
we have seen that , at the `` natural '' scale @xmath121 gev , the corrections to a zero mass gluino can reach 3 or 4 gev . comparing our results with those presented in @xcite , we note that we have explored a different region of the parameters space and @xmath122 , we can avoid the experimental bounds induced on @xmath8 by the non - observation at lep of charginos lighter than 50 gev that were considered in @xcite . ] .
i kindly thank f. zwirner for continuous help during the completion of this work and for introducing me to this subject .
i would also like to thank a. brignole and g. ridolfi for useful discussions and suggestions .
in this appendix we report all the explicit results for the one - loop radiative corrections to the pole masses of the stops , of the gluino and of the top quark , obtained in the approximation where all the dimensionless couplings other than @xmath1 and @xmath2 are set to zero .
the corresponding feynman rules have been deduced from the rules reported in @xcite . it must be noticed that ( in the last of @xcite ) a factor @xmath123 is missing in the vertices between two neutral higgses and two squarks .
this factor , due to a combinatorial factor of 2 , cancels if the external legs are the two neutral higgses giving the feynman rule reported in the same reference .
the same happens to the two - gluon / two - squark vertex . in the formulae we have made use of the following integrals : @xmath124 \log \left(\frac{t(1,2,3)}{q^2}\right ) \\
g(1,2,3 ) & = & { \cal r}e\int^1_0 dz [ m^2_3z(1-z ) ] \log \left(\frac{t(1,2,3)}{q^2}\right ) \\
k(1,2,3 ) & = & { \cal r}e\int^1_0 dz [ m^2_3z(1-z ) - m^2_1(1-z ) - m^2_2z ] \log \left(\frac{t(1,2,3)}{q^2}\right ) \\ l(1,2,3 ) & = & { \cal r}e\int^1_0 dz [ m^2_3 ( z-2)^2 ] \log \left(\frac{t(1,2,3)}{q^2}\right ) \\
m(1,2,3 ) & = & { \cal r}e\int^1_0 dz [ m_3 z ] \log \left(\frac{t(1,2,3)}{q^2 } \right ) \\
n(1,2,3 ) & = & { \cal r}e\int^1_0 dz [ m_3(1- z ) ] \log \left(\frac{t(1,2,3)}{q^2 } \right)\end{aligned}\ ] ] where @xmath125 .
the analytical expressions for these integrals can be found following @xcite .
they all depend on : @xmath126 where @xmath127 the other integrals have the following analytical expressions : @xmath128 \\ & + & \frac{1}{2 } \left[m^2_1 + m^2_2 - \left(\frac{m^2_1 - m^2_2}{m_3}\right)^2\right ] f(1,2,3 ) \\
g(1,2,3 ) & = & \frac{m^2_3}{18 } -\frac{1}{6 } \left[m^2_1 \log\frac{m^2_1}{q^2 } + m^2_2 \log\frac{m^2_2}{q^2 } \right ] + \frac{2}{3}h(1,2,3 ) \\ & - & \frac{1}{6 } [ m^2_1 + m^2_2 - m^2_3]f(1,2,3 ) \\
k(1,2,3 ) & = & g(1,2,3 ) - h(1,2,3 ) \\
l(1,2,3 ) & = & m^2_3 \left(4 - \frac{3m^2_1}{m^2_1-m^2_2}\right ) f(1,2,3 ) + \frac{3m^2_3}{m^2_1 - m^2_2}h(1,2,3 ) - g(1,2,3 ) \\ m(1,2,3 ) & = & \frac{m_3 m^2_1}{m^2_1 - m^2_2 } f(1,2,3 ) - \frac{m_3}{m^2_1-m^2_2 } h(1,2,3 ) \\ n(1,2,3 ) & = & m_3 f(1,2,3 ) - m(1,2,3)\end{aligned}\ ] ] all these integrals are symmetric under the permutation @xmath129 , made exception for @xmath130 and for @xmath131 . it has been stressed @xcite that if some entries of the integral @xmath132 are zero , the right expressions are not deducible by just taking the limit from these formulae .
this happens because , if @xmath133 has some zero entries , some of the poles in the logarithm may disappear .
here we report the main particular cases , for completeness : @xmath134 \\
f(1,1,0 ) & = & \log\frac{m^2_1}{q^2 } .\end{aligned}\ ] ] the particular cases for the other integrals are deducible by taking the limit for the reported expressions and using the particular expressions for the @xmath132 integral .
this is not true for the @xmath135 integrals in the case @xmath136 : but also in the pathological case their decomposition over the main integrals @xmath137 is easy to obtain .
below are reported the expressions for the one - loop radiative corrections to the pole masses of the stops , the gluino and the top quark . in these expressions
, @xmath138 is the number of colours ( @xmath139 ) , @xmath140 and @xmath141 .
moreover , @xmath142 and @xmath143 . we have used a subscript @xmath144 for the up - type quarks and squarks different from top and stop , and a subscript @xmath145 for all the down - type quarks and squarks . in the following , ( @xmath146 )
will be abbreviated as ( @xmath147 ) . 1 . self - energy of the top quark ( see fig . [ topfeyn ] )
: @xmath148 \\ i \pi^{\tilde g}_t(\slash{p } ) & = - \frac{i \delta^{\alpha\beta}}{16 \pi^2 } g^2_s c_2(r ) & \left [ m(\tilde g,\tilde t_1,p ) - 2\tilde m_g s_t c_t f(\tilde g,\tilde t_1,p ) \right.\\ & & + \left .
m(\tilde g,\tilde t_2,p ) + 2\tilde m_g s_t c_t f(\tilde g,\tilde t_2,p ) \right ] \\
i \pi^h_t(\slash{p } ) & = - \frac{i \delta^{\alpha\beta}}{16 \pi^2 } h^2_t & \left\ { \frac{1}{2}\sin^2\alpha \left [ n(h^0,t , p ) + m_t f(h^0,t , p ) ) \right ] \right . \\ & & + \frac{1}{2}\cos^2\alpha \left[n(h^0,t , p ) + m_t f(h^0,t , p ) \right ] \\ & & + \frac{1}{2}\cos^2\beta \left [ n(a , t , p ) - m_t f(a , t , p)\right ] \\ & & + \frac{1}{2}\sin^2\beta \left [ n(0,t , p ) - m_t f(0,t , p)\right ] \\ & & + \left . \frac{1}{2 } \left [ \cos^2\beta m(h^{\pm},b , p ) + \sin^2\beta m(0,b , p ) \right ] \right\ } \\
i \pi^{\tilde \chi^0}_t(\slash{p } ) & = - \frac{i \delta^{\alpha\beta}}{16 \pi^2 } h^2_t & \left\ { \frac{1}{2 } \sum_j \left [ \mid n_{j4}\mid^2 m(\tilde \chi^0_j , \tilde t_1,p ) + \tilde m_{\chi^0_j } s_t c_t ( n^2_{j4 } + n^{\star^2}_{j4 } ) f(\tilde \chi^0_j , \tilde t_1,p ) \right ] \right . \\ & & + \left . \frac{1}{2 } \sum_j \left [ \mid n_{j4}\mid^2 m(\tilde \chi^0_j , \tilde t_2,p ) - \tilde m_{\chi^0_j } s_t c_t ( n^2_{j4 } + n^{\star^2}_{j4 } ) f(\tilde \chi^0_j , \tilde t_2,p ) \right ] \right\ } \\
i \pi^{\tilde \chi^+}_t(\slash{p } ) & = - \frac{i \delta^{\alpha\beta}}{16 \pi^2 } h^2_t & \left\ { \frac{1}{2 } \sum_i \left [ c^2_b \mid v_{i2}\mid^2 m(\tilde \chi^+_i , \tilde b_1,p ) + s^2_b \mid v_{i2}\mid^2 m(\tilde \chi^+_i , \tilde b_2,p ) \right ] \right\}\end{aligned}\ ] ] 2 .
self - energies of the stops ( see fig . [ stopfeyn ] ) : 1 .
i \pi^{\tilde t}_{11 } & = \frac{i \delta^{\alpha\beta}}{16 \pi^2 } & \left \ { g^2_s c_2(r ) \left [ \frac{1}{6 } p^2 + \frac{1}{2 } \tilde m^2_{t_1 } - 4 s^2_t c^2_t ( \tilde m^2_{t_1 } - \tilde m^2_{t_2 } ) - 2 k(0,\tilde t_1,p ) \right.\right.\\ & & + \left .
l(0,\tilde t_1,p ) + ( 1 - 4 s^2_t c^2_t ) k(\tilde t_1,\tilde t_1,0 ) + 4 s^2_t c^2_t k(\tilde t_2,\tilde t_2,0 ) \right ] \\ & & + h^2_t \left [ 2 ( n+1 ) s^2_t c^2_t ( \tilde m^2_{t_1 } - \tilde m^2_{t_2 } ) + \tilde m^2_{t_2 } \right .
\\ & & + \left.\left .
2 ( n+1 ) s^2_t c^2_t k(\tilde t_1,\tilde t_1,0 ) + ( c^4_t + s^4_t - 2 n s^2_t c^2_t ) k(\tilde t_2,\tilde
t_2,0 ) \right ] \right \ } \\ i \pi^{\tilde b}_{11 } & = \frac{i \delta^{\alpha\beta}}{16 \pi^2 } & h^2_t \left\ { s^2_t ( c^2_b \tilde m^2_{b_1 } + s^2_b \tilde m^2_{b_2 } ) + s^2_t \left [ c^2_b k(\tilde b_1,\tilde b_1,0 ) + s^2_b k(\tilde b_2,\tilde b_2,0)\right ] \right\ } \\
i \pi^{h}_{11 } & = \frac{i \delta^{\alpha\beta}}{16 \pi^2 } & h^2_t \left [ \frac{1}{2 } m^2_{h^0 } \sin^2 \alpha + \frac{1}{2 } m^2_{h^0 } \cos^2 \alpha + \frac{1}{2 } m^2_a \cos^2 \beta + s^2_t m^2_{h^{\pm } } \cos^2 \beta \right . \\ & & + \frac{1}{2 } \sin^2 \alpha k(h^0,h^0,0 ) + \frac{1}{2 } \cos^2 \alpha k(h^0,h^0,0 ) \\ & & + \frac{1}{2 } \cos^2 \beta k(a , a,0 ) + s^2_t \cos^2 \beta k(h^{\pm},h^{\pm},0 ) \\ & & - 2 m^2_t ( v^{h^0}_{11})^2 f(h^0,\tilde t_1,p ) -2 m^2_t ( v^{h^0}_{12})^2 f(h^0,\tilde t_2,p ) \\ & & - 2 m^2_t ( v^{h^0}_{11})^2 f(h^0,\tilde t_1,p ) -2 m^2_t ( v^{h^0}_{12})^2 f(h^0,\tilde t_2,p ) \\ & & - \frac{1}{2 } \sin^2\beta ( v^a_{12})^2 f(a,\tilde t_2,p ) -\frac{1}{2 } \sin^2\beta ( v^{g^0}_{12})^2 f(0,\tilde t_2,p ) \\ & & - \sin^2\beta ( v^{h^{\pm}}_{11})^2 f(h^{\pm},\tilde b_1,p ) - \sin^2\beta ( v^{h^{\pm}}_{12})^2 f(h^{\pm},\tilde b_2,p ) \\ & & - \left.\sin^2\beta ( v^{g^{\pm}}_{11})^2 f(0,\tilde b_1,p ) - \sin^2\beta ( v^{g^{\pm}}_{12})^2 f(0,\tilde b_2,p ) \right ] \\
i \pi^{\tilde \chi^0}_{11 } & = \frac{i \delta^{\alpha\beta}}{16 \pi^2 } & h^2_t \left [ \frac{1}{3 } p^2 - m^2_t - \sum_j \mid n_{j4 } \mid^2 \tilde m^2_{\chi^0_j } \right.\\ & & - 4 \sum_j \mid n_{j4}\mid^2 k(\tilde \chi^0_j , t , p ) - 2 \sum_j \mid n_{j4}\mid^2 g(\tilde \chi^0_j , t , p ) \\ & & + \left . 2 s_t c_t \sum_j ( n^2_{j4 } + n^{\star^2}_{j4 } ) m_t \tilde m_{\chi^0_j } f(\tilde \chi^0_j , t , p ) \right ] \\
i \pi^{\tilde \chi^+}_{11 } & = \frac{i \delta^{\alpha\beta}}{16 \pi^2 } & h^2_t \left [ \frac{1}{3 } s^2_t p^2 - s^2_t m^2_b - s^2_t \sum_i \mid v_{i2 } \mid^2 \tilde m^2_{\chi^+_i } \right .
\\ & & - \left .
4 s^2_t \sum_i \mid v_{i2}\mid^2 k(\tilde \chi^+_i , b , p ) - 2 s^2_t \sum_i \mid v_{i2}\mid^2 g(\tilde \chi^+_i , b , p ) \right]\end{aligned}\ ] ] 2 .
i \pi^{\tilde t}_{22 } & = \frac{i \delta^{\alpha\beta}}{16 \pi^2 } & \left \
{ g^2_s c_2(r ) \left [ \frac{1}{6 } p^2 + \frac{1}{2 } \tilde m^2_{t_2 } + 4 s^2_t c^2_t ( \tilde m^2_{t_1 } - \tilde m^2_{t_2 } ) - 2 k(0,\tilde t_2,p ) \right.\right.\\ & & + \left .
l(0,\tilde t_2,p ) + ( 1 - 4 s^2_t c^2_t ) k(\tilde t_2,\tilde t_2,0 ) + 4 s^2_t c^2_t k(\tilde t_1,\tilde t_1,0 ) \right ] \\ & & + h^2_t
\left [ - 2 ( n+1 ) s^2_t c^2_t ( \tilde m^2_{t_1 } - \tilde m^2_{t_2 } ) + \tilde m^2_{t_1 } \right .
\\ & & + \left.\left .
2 ( n+1 ) s^2_t c^2_t k(\tilde t_2,\tilde t_2,0 ) + ( c^4_t + s^4_t - 2 n s^2_t c^2_t ) k(\tilde
t_1,\tilde t_1,0 ) \right ] \right \ } \\ i \pi^{\tilde b}_{22 } & = \frac{i \delta^{\alpha\beta}}{16 \pi^2 } & h^2_t \left\ { c^2_t ( c^2_b \tilde m^2_{b_1 } + s^2_b \tilde m^2_{b_2 } ) + c^2_t \left [ c^2_b k(\tilde b_1,\tilde b_1,0 ) + s^2_b k(\tilde b_2,\tilde b_2,0)\right ] \right\ } \\
i \pi^{h}_{22 } & = \frac{i \delta^{\alpha\beta}}{16 \pi^2 } & h^2_t \left [ \frac{1}{2 } m^2_{h^0 } \sin^2 \alpha + \frac{1}{2 } m^2_{h^0 } \cos^2 \alpha + \frac{1}{2 } m^2_a \cos^2 \beta + c^2_t m^2_{h^{\pm } } \cos^2 \beta \right . \\ & & + \frac{1}{2 } \sin^2 \alpha k(h^0,h^0,0 ) + \frac{1}{2 } \cos^2 \alpha k(h^0,h^0,0 ) \\ & & + \frac{1}{2 } \cos^2 \beta k(a , a,0 ) + c^2_t \cos^2 \beta k(h^{\pm},h^{\pm},0 ) \\ & & - 2 m^2_t ( v^{h^0}_{21})^2 f(h^0,\tilde t_1,p ) -2 m^2_t ( v^{h^0}_{22})^2 f(h^0,\tilde t_2,p ) \\ & & - 2 m^2_t ( v^{h^0}_{21})^2 f(h^0,\tilde t_1,p ) -2 m^2_t ( v^{h^0}_{22})^2 f(h^0,\tilde t_2,p ) \\ & & - \frac{1}{2 } \sin^2\beta ( v^a_{12})^2 f(a,\tilde t_1,p ) -\frac{1}{2 } \sin^2\beta ( v^{g^0}_{12})^2 f(0,\tilde t_1,p ) \\ & & - \sin^2\beta ( v^{h^{\pm}}_{21})^2 f(h^{\pm},\tilde b_1,p ) - \sin^2\beta ( v^{h^{\pm}}_{22})^2 f(h^{\pm},\tilde b_2,p ) \\ & & - \left.\sin^2\beta ( v^{g^{\pm}}_{21})^2 f(0,\tilde b_1,p ) - \sin^2\beta ( v^{g^{\pm}}_{22})^2 f(0,\tilde b_2,p ) \right ] \\
i \pi^{\tilde \chi^0}_{22 } & = \frac{i \delta^{\alpha\beta}}{16 \pi^2 } & h^2_t \left [ \frac{1}{3 } p^2 - m^2_t - \sum_j \mid n_{j4 } \mid^2 \tilde m^2_{\chi^0_j } \right.\\ & & - 4 \sum_j \mid n_{j4}\mid^2 k(\tilde \chi^0_j , t , p ) - 2 \sum_j \mid n_{j4}\mid^2 g(\tilde \chi^0_j , t , p ) \\ & & - \left .
2 s_t c_t \sum_j ( n^2_{j4 } + n^{\star^2}_{j4 } ) m_t \tilde m_{\chi^0_j } f(\tilde \chi^0_j , t , p ) \right ] \\
i \pi^{\tilde \chi^+}_{22 } & = \frac{i \delta^{\alpha\beta}}{16 \pi^2 } & h^2_t \left [ \frac{1}{3 } c^2_t p^2 - c^2_t m^2_b - c^2_t \sum_i \mid v_{i2 } \mid^2 \tilde m^2_{\chi^+_i } \right . \\ & & - \left . 4 c^2_t \sum_i \mid v_{i2}\mid^2 k(\tilde \chi^+_i , b , p ) - 2 c^2_t \sum_i \mid v_{i2}\mid^2 g(\tilde \chi^+_i , b , p ) \right]\end{aligned}\ ] ] 3 .
i \pi^{\tilde t}_{12 } & = \frac{i \delta^{\alpha\beta}}{16 \pi^2 } & \left\ { g^2_s c_2(r ) \cos 2 \phi_t \left [ - s_t c_t ( \tilde m^2_{t_1 } - \tilde m^2_{t_2 } ) \right.\right .
\\ & & - \left .
2 s_t c_t ( k(\tilde t_1 , \tilde t_1 , 0 ) - k(\tilde t_2 , \tilde t_2 , 0 ) ) \right ] \\ & & + h^2_t \left [ ( n+1)s_t c_t \cos 2 \phi_t ( \tilde m^2_{t_1 } - \tilde m^2_{t_2 } ) \right.\\ & & + \left . \left .
( n+1 ) s_t c_t \cos 2\phi_t ( k(\tilde t_1,\tilde t_1,0 ) - k(\tilde t_2,\tilde t_2,0 ) ) \right ] \right\ } \\
i \pi^{\tilde b}_{12 } & = \frac{i \delta^{\alpha\beta}}{16 \pi^2 } & h^2_t \left\ { s_t c_t ( c^2_b \tilde m^2_{b_1 } + s^2_b \tilde m^2_{b_2 } ) \right . \\ & & + \left .
s_t c_t \left [ c^2_b k(\tilde b_1,\tilde b_1,0 ) + s^2_b k(\tilde b_2,\tilde b_2,0)\right ] \right\ } \\
i \pi^h_{12 } & = \frac{i \delta^{\alpha\beta}}{16 \pi^2 } & h^2_t \left [ m^2_{h^{\pm } } \cos^2 \beta s_t c_t + \cos^2 \beta s_t c_t k(h^{\pm},h^{\pm},0 ) \right .
\\ & & - 2 m^2_t ( v^{h^0}_{11}v^{h^0}_{12 } ) f(h^0,\tilde t_1,p ) -2 m^2_t ( v^{h^0}_{12}v^{h^0}_{22 } ) f(h^0,\tilde t_2,p ) \\ & & - 2 m^2_t ( v^{h^0}_{11}v^{h^0}_{12 } ) f(h^0,\tilde t_1,p ) -2 m^2_t ( v^{h^0}_{12}v^{h^0}_{22 } ) f(h^0,\tilde t_2,p ) \\ & & - \sin^2 \beta ( v^{h^{\pm}}_{11}v^{h^{\pm}}_{12 } ) f(h^{\pm},\tilde b_1,p ) - \sin^2 \beta ( v^{h^{\pm}}_{12}v^{h^{\pm}}_{22 } ) f(h^{\pm},\tilde b_2,p ) \\ & & - \left .
\sin^2 \beta ( v^{g^{\pm}}_{11}v^{g^{\pm}}_{12 } ) f(0,\tilde b_1,p ) - \sin^2 \beta ( v^{g^{\pm}}_{12}v^{g^{\pm}}_{22 } ) f(0,\tilde b_2,p ) \right ] \\
i \pi^{\tilde \chi^0}_{12 } & = \frac{i \delta^{\alpha\beta}}{16 \pi^2 } & h^2_t \left [ 2 \sum_j ( c^2_t n^{\star^2}_{j4 } - s^2_t n^2_{j4 } ) m_t \tilde m_{\chi^0_j } f(\tilde \chi^0_j , t , p ) \right ] \\
i \pi^{\tilde \chi^+}_{12 } & = \frac{i \delta^{\alpha\beta}}{16 \pi^2 } & h^2_t \left [ \frac{2}{3}s_t c_t p^2 - 2 s_t c_t m^2_b - 2 s_t c_t \sum_i \mid v_{i2 } \mid^2 \tilde m^2_{\chi^+_i } \right .
\\ & & - \left .
2 s_t c_t \sum_i \mid v_{i2}\mid^2 ( k(\tilde \chi^+_i , b , p ) + g(\tilde \chi^+_i , b , p ) ) \right ] \\\end{aligned}\ ] ] + in these expressions , we have used the following definitions for the higgs squark squark vertices in the ( @xmath152 ) basis : @xmath153 the vertices @xmath154 can be obtained by @xmath155 , replacing @xmath156 and @xmath157 .
self - energy of the gluino ( see fig .
[ gluefeyn ] ) : @xmath158 \\ i \pi^{\tilde t}_{\tilde g } ( \slash{p } ) & = & -\frac{i g^2_s \delta^{ab } } { 16 \pi^2 } \sum^3_{i=1 } \left [ n(\tilde t_{1_i},t_i , p ) - 2 m_{t_i } s_{t_i } c_{t_i } f(\tilde t_{1_i},t_i , p ) + n(\tilde t_{2_i},t_i , p ) + 2 m_{t_i } s_{t_i } c_{t_i } f(\tilde t_{2_i},t_i , p ) \right ] \\
i \pi^{\tilde b}_{\tilde g } ( \slash{p } ) & = & -\frac{i g^2_s \delta^{ab } } { 16 \pi^2 } \sum^3_{i=1 } \left [ n(\tilde b_{1_i},b_i , p ) - 2 m_{b_i } s_{b_i } c_{b_i } f(\tilde b_{1_i},b_i , p ) + n(\tilde b_{2_i},b_i , p ) + 2 m_{b_i } s_{b_i } c_{b_i } f(\tilde b_{2_i},b_i , p ) \right]\end{aligned}\ ] ] in this expression , @xmath159 is the generation index , and @xmath160 stand for top - type and bottom - type .
self - energy of the @xmath60 vector boson ( see fig . [ zetafeyn ] ) : @xmath161\end{aligned}\ ] ] in these expressions we have used : @xmath162 where @xmath163 represents top- and bottom - type quarks , neutrinos and charged leptons ( with @xmath164 and @xmath165 ) .
this result has already been obtained @xcite .
we have reported it here for completeness , in the approximated form in which terms proportional to @xmath166 are neglected .
in this appendix we report the explicit solutions of the one - loop renormalization group equations for the parameters relevant to our calculations , in the approximation where the only non - vanishing dimensionless couplings are @xmath1 and @xmath2 . in the following , @xmath167 , where @xmath168 is an arbitrary reference scale .
the rges of interest can be extracted from the general formulae of @xcite . it must be stressed that , in accordance with our sign and phase convention @xcite , the correct equations that must be solved correspond to the equations reported in the second paper of @xcite with the substitution @xmath169 . following the notation of @xcite
, we introduce a set of parameters : @xmath170 where @xmath171 .
it can be seen that , in our approximation , the integral @xmath132 can be easily solved . introducing the variable @xmath172
the explicit solutions for a first set of equations are : @xmath173 \\
\tilde m_g(t ) & = & \tilde m_g(0 ) z_3 \ .\end{aligned}\ ] ] the explicit solutions for the soft masses rges depend on the quantity : @xmath174 \nonumber \\ & + & \frac{1}{2 } \tilde m^2_g(0)x \left [ t \frac{e}{f } \frac{16}{3}\frac{\alpha_s(0)}{4 \pi } t z_3 - \left(t \frac{e}{f } - 1 \right)^2 x \right ] \ .\end{aligned}\ ] ] then , we obtain : @xmath175 we recall that the soft mass terms for the squarks of other generations ( @xmath176 , with @xmath177 ) and ( @xmath178 , with @xmath5 ) for the sleptons have been considered identical to @xmath179 , for simplicity . j. f. gunion , h. e. haber , g. l. kane and s. dawson , _ the higgs hunter s guide _ ( addison - wesley , reading , ma , 1990 ) ; errata , hep - ph/9302272 . j. f. gunion and h. e. haber , nucl . phys . * b 272 * ( 1986 ) 1 , * b 278 * ( 1986 ) 449 and * b 307 * ( 1988 ) 445 ; errata , hep - ph/9301205 .
j. rosiek , phys . rev . *
d 41 * ( 1990 ) 3464 ; errata , hep - ph/9511250 .
w. siegel , phys .
* b 84 * ( 1979 ) 193 ; a. brignole , phys .
* b 281 * ( 1992 ) 284 .
a. brignole , j. r. espinosa , m. quirs and f. zwirner , phys . lett .
* b 324 * ( 1994 ) 181 . y. yamada , phys .
* b 316 * ( 1993 ) 25 ; phys .
72 ( 1994 ) 25 .
j. bagger , k. matchev and d. pierce , hep - ph/9501378 .
s. p. martin and m. t. vaughn , phys .
* 318 * ( 1993 ) 331 ; | we compute the one - loop radiative corrections to the masses of the top quark , the stop squarks and the gluino in the minimal supersymmetric standard model .
we include the loops controlled by the strong coupling constant and the top yukawa coupling , neglecting those controlled by the @xmath0 gauge couplings and by other yukawa couplings .
we find that the significant scale - dependence of the renormalization - group - improved tree - level expressions is almost completely removed . even for natural choices of the renormalization scale , corrections can be numerically relevant .
our results should allow more reliable predictions to be extracted from candidate fundamental theories beyond the mssm .
= 16.5 cm = 22.5 cm = -2 cm = -1 cm /#1 # 1 # 1 # 1 cern - th/95 - 287 + rome1 - 1122/95 + 0.5 cm * one - loop corrections * * to the top , stop and gluino masses in the mssm * 0.8 cm * a. donini * theory division , cern , 1211 geneva 23 , switzerland , dip . di fisica , universit degli studi di roma `` la sapienza '' and infn , sezione di roma , p.le a. moro 2 , 00185 rome , italy cern - th/95 - 287 + november 1995 0.5 cm |
those conceptual approaches to a subject that offer both wider applicability , and reduced algorithmic complexity , might be called `` deep simplifications '' .
minkowski s approach to special relativity via the metric equation is a classic example .
the first impressions of `` superfluous erudition''@xcite have now been eclipsed , by uses for minkowski s perspective in developing general relativity@xcite as well as first introductions to space - time@xcite .
similarly , the communication - theory insights of shannon@xcite led jaynes@xcite in the middle of this century to clarify the distinction in thermal physics between `` describing the dice '' ( i.e. physical description ) , and `` taking the best guess '' ( the `` gambling theory '' part ) .
thus for example , the 1st through 3rd laws of thermodynamics help `` describe the dice '' , while the zeroth law , as well as the boltzmann ( canonical ) and gibbs ( grand canonical ) factors , are `` dice - independent '' tools of statistical inference@xcite .
hence we can ( with increasing help from undergraduate texts@xcite ) deepen understanding ( including the physical intuition of computer science and biology students on matters of `` code '' ) by disclosing that entropy @xmath7 is a measure of statistical uncertainty , expressible using information units as well as in j / k . from this
, it follows that temperature ( i.e. the reciprocal of uncertainty slope @xmath8 ) can be measured in units of energy per unit information ( e.g. room temperature @xmath9 @xmath10 ev per nat ) .
thus heat capacities ( i.e. @xmath11 ) , normally understood in units of energy per degree kelvin , may find themselves measured in units of information alone ! as we will show ,
this is part of a larger trend in statistical physics to shift the focus from temperature ( or its reciprocal ) to the physical dependence of multiplicity ( or entropy ) on variables conserved in the interaction between complex systems .
but , what is the _ physical _ meaning of a heat capacity without reference to historical temperature units ?
bits of what ?
an answer to this question ( something any student might ask ) does not appear to be common knowledge among physics teachers , so we outline an answer here .
we show further why total thermal energy over kt _ for any system _ serves at once : ( i ) as the instantaneous exponent of energy in the expression for accessible state multiplicity , ( ii ) as a measure of the number of bits of ( micro - canonical ) uncertainty added per two - fold increase in thermal energy , and ( iii ) _ for quadratic systems _ as the number of degrees freedom over two .
the dice of thermal physics are usually physical systems capable of accomodating thermal energy ( as well as other quantities that may be conserved , like volume and particles ) in a _ multiplicity _ of ways .
this multiplicity is itself the key to understanding , particularly when systems are seen from the ( micro - canonical ) vantage point of the conserved quantities _ per se_. for example ,
many gases , liquids , and solids behave over some of their temperature range as though multiplicity ( @xmath12 ) is proportional to @xmath13 , where @xmath1 is thermal energy , @xmath14 is the number of molecules , and @xmath15 is the number of ways to store thermal energy per molecule .
such systems are called `` quadratic '' , since the proportionality results from a `` sum of squares '' connection between @xmath1 and the state coordinates involved in storing energy .
application of gambling theory to physical systems begins with the question : where might we expect to find a conserved quantity @xmath16 which has been randomly ( as far as we know ) shared between systems for so long that prior information about the whereabouts of @xmath16 is irrelevant ? in the jargon of the field , this is the same as asking : where is @xmath16 likely to be _
after equilibration _ ?
the science of decision - making in the presence of uncertainty ( i.e. statistical inference or `` gambling theory '' ) suggests that the best bet is the distribution of @xmath16 that can happen in the most ways , provided there is no reason to prefer one way over another . in the jargon of probability theory , this is the recommendation : assume equal _ a priori _ probabilities , when evidence to the contrary is not available .
for example , if systems a and b have a total energy @xmath1 to share between them , then the best bet after equilibration will be that value of @xmath17 that has the largest total multiplicity @xmath18 . setting to zero the derivative of @xmath12 with respect to @xmath19 ,
one finds that this maximum requires that @xmath20 , and hence that @xmath21 be the same for both systems ! these derivatives ( and most others in this paper ) are taken under `` microcanonical constraints '' , i.e. they are partial derivatives with other extensive quantities ( like volume or number of particles ) held constant .
information theorists , of course , define the logarithm of multiplicity as uncertainty via the equation @xmath22 , and measure it in \{nats , bits , bytes , or j / k } if @xmath23 is \{@xmath24,@xmath25,@xmath26,@xmath27 } , respectively .
thus the best bet for any two systems sharing a conserved quantity @xmath16 , in the absence of information to the contrary , is that @xmath16 will rearrange itself between the two systems until each system s _ uncertainty slope _ ( @xmath28 )
has reached a common value .
this is a quantitative version of the zeroth law of thermodynamics , based purely in the science of statistical inference , which applies to _ any systems _ sharing conserved quantities . of course , when energy @xmath1 is the quantity shared randomly between systems , the uncertainty slope @xmath29 is the reciprocal temperature or _
_ @xmath30 .
hence temperature is a property which signals the propensity of a system for sharing of energy thermally .
for example , calculating the uncertainty slope for quadratic systems from the multiplicity given above yields the widely useful equipartition relation : @xmath31 . when @xmath32 is the quantity shared randomly between systems , the uncertainty slope @xmath33 is the _ free - expansion coefficient _ equal to @xmath34 at equilibrium@xcite . for an ideal gas , @xmath35 . solving this for @xmath36 yields the ideal gas equation of state @xmath37 .
when @xmath14 is the quantity shared randomly , the uncertainty slope @xmath38 is the _ chemical affinity _ ,
equal to @xmath39 at equilibrium . from this , for example , reaction equilibrium constants may be calculated .
this quantitative version of the zeroth law applies to all thermal systems which equilibrate , including spin systems ( like magnets ) capable of population inversions and hence negative absolute temperatures .
moreover , as a theorem of statistical inference not involving energy at all , it applies also to thermally unequilibrated systems sharing other conserved quantities ( even money , for example ) , provided the only prior information we have is how the multiplicity of ways that quantity can be distributed depends on the amount of that conserved quantity to begin with !
if we have other kinds of information , such as knowledge of a system s temperature but not its total energy , then the broader class of maximum entropy strategies in statistical inference ( e.g. the canonical and grand ensembles ) predict the distribution of outcomes we can expect there as well .
a closer look shows that the statistical definition of temperature above can be rewritten as : @xmath40 the quantity in curly brackets is the log - log derivative of multiplicity with respect to thermal energy .
we can also think of this as the `` instantaneous exponent '' of energy in the expression of multiplicity as energy to some power , or as the slope of the multiplicity versus energy curve on a log - log plot . rearranging the equation yields something that looks very much like the familiar equipartition theorem , except that the relation applies to all thermal systems under conditions of maximum ignorance ( i.e. at equilibrium ) :
@xmath41 for quadratic systems , it is easy to see that our log - log derivative is nothing more than half the number of degrees of freedom ( @xmath42 ) . for any system
, however , @xmath43 measures the instantaneous energy exponent , as well as the number of _ nats of information lost about the state of the system per e - fold increase in thermal energy _ of the system.the last term in the equality string simply notes that the value is independent of the base @xmath3 of the logarithms used , provided the same base is used in numerator and denominator .
thus we can also think of @xmath43 as the number of bits of information lost per @xmath44-fold increase in thermal energy , or more generally the number of base-@xmath3 units of information lost per @xmath3-fold increase in thermal energy . in our search for the meaning of heat capacity in natural units ,
this is our first big clue .
before we move on , we should also point out something that applies if our energy origin has been chosen so that @xmath45 as @xmath46 , something we _ might _ expect for a measure of thermal energy . in terms of the no - work ( e.g. constant volume ) heat - capacity @xmath47 , we can write : @xmath48 with the middle equality applying only for systems not in a population inversion , so that absolute temperature @xmath4 and @xmath49 .
thus when absolute temperature is positive , @xmath43 is a _ heat capacity average _ over temperatures ranging between @xmath50 and absolute zero .
we ve shown here that the log - log derivative of multiplicity , with respect to energy , has a simple information theoretic interpretation , and is elegantly given in natural units by @xmath51 as well . from the perspective of an experimentalist , however , it has one glaring disadvantage : _ it s numeric value depends on our choice for the zero of thermal energy . _ to illustrate the problem , consider the cooling of water until it becomes ice . as water cools initially ,
the temperature drop per unit energy removed is roughly constant .
one might easily say : `` this looks like a quadratic system with about 18 degrees of freedom per molecule '' , so @xmath43 must be about 9 bits per 2-fold increase in thermal energy . then
, at the freezing point , the temperature stops dropping as energy continues to be removed , suggesting a quadratic system with nearly infinite degrees of freedom !
once all is frozen , of course , temperature continues its drop , this time suggesting a quadratic system with about 8 degrees of freedom per molecule , or @xmath43 closer to 4 bits per 2-fold increase in thermal energy !
since @xmath2 may change little during this experiment , how can @xmath5 be jumping around so much ?
the answer of course is that these inference follow not by measuring total energy @xmath1 , but only changes in energy .
moreover , in the process our preferred zero of thermal energy has been shifting about .
we can see the effects of this more explicitly if we plot energy versus temperature for water , as shown in the lower left panel of figure 1 .
the question then is , can we modify our estimate for the log - log derivative of thermal energy so as to reflect only data on temperature changes over a limited energy range ? such a quantity might allow us to probe the ways that thermal energy is being accomodated specifically , one energy range at a time .
the no - work ( e.g. constant volume ) instantaneous heat capacity , in natural units , can be written in terms above as : @xmath52 = \left [ 1+t\frac \partial { \partial t}\right ] \xi = t% \frac{\partial \ln \omega } { \partial t}=t\frac{\partial ( s / k)}{\partial t}% \text{. } \label{heatcapacity}\ ] ] here @xmath50 is absolute temperature in any units you like , and the partials are taken with work parameters ( like volume ) held constant .
this quantity has an interesting property . if we define thermal energy @xmath1 as a difference between total energy @xmath53 and a specified `` zero thermal energy '' origin @xmath54 , i.e. as @xmath55 , then it is easy to see that @xmath56 is independent of our choice for @xmath57 .
thus , although @xmath43 obviously depends on one s choice of @xmath54 , @xmath58 does not . to see what @xmath58 actually measures ,
let s suppose that we have a quadratic system whose multiplicity obeys @xmath59 .
it then follows simply that @xmath60 , @xmath61 , @xmath62 . thus @xmath63 estimates not @xmath64 but @xmath65 , where @xmath66 is the `` true origin '' of thermal energy for this quadratic system .
this is illustrated in figure 2 , which plots for an ideal monatomic gas ( the classical quadratic system ) the same quantities plotted for water in fig .
if the system is not simply quadratic ( e.g. if it has phase changes , modes of energy storage which freeze out , etc . ) , then @xmath63 is simply a local estimate of thermal energy over @xmath2 , under the quadratic assumption .
thus @xmath67 modifies the log - log derivative of multiplicity with respect to energy , by combining it with its rate of increase per e - fold change in temperature , to yield an estimate of @xmath65 , where @xmath66 is a zero of thermal energy determined by assuming that the locally - measured log - log derivative of multiplicity is constant down to @xmath68 .
this observation also provides a different perspective on the mechanism by which heat capacity blows up during a phase change .
thermal energy over @xmath2 ( or the log - log derivative of multiplicity ) of course should have no singularities in it , since both energy and @xmath69 are expected to be finite for finite systems .
it is thus the 2nd term in the two - term expression for heat capacity above , namely the temperature derivative , that provides the instability .
discontinuous shifts in the locally - inferred energy zero ( @xmath66 ) on which heat capacity is based during a phase change thus , via this second term , also cause heat capacity to become singular .
returning to figure 1 , this raises the interesting question : _ does the thermal energy of steam increase , decrease , or go negative , when it condenses to water ?
_ from the above , we can see that it of course decreases ( perhaps even goes negative ) if one s `` zero of thermal energy '' is held constant , since steam loses the latent heat of vaporization when it condenses , bringing it s total energy down .
however , because the specific heat of water at boiling is higher than that for steam , energy of random motion measured with respect to our locally - inferred zero of thermal energy ( e.g. at 100c ) actually goes up ! in other words , a small part of the binding energy , liberated when water molecules fall into the potential well of their neighbors , goes to /em increase the energy of random motion in the condensed phase relative to that available to particles in uncondensed gas !
the observations above suggest that introductory texts might consider highlighting @xmath63 per molecule in natural units for common substances , with no apology for the fact that it is near but not exactly half - integral in many cases .
after all , thermal energy may not have the same access to all molecules , all of the time .
this quantity nonetheless provides deep insight into the relationship between uncertainty and thermal energy . in shifting the focus from historical temperature units to the multiplicities which underlie our inferences , we can say that both @xmath43 and @xmath63 measure bits of uncertainty per 2-fold increase in energy for _ all _ physical systems , with respect to their respective choices for energy origin . in this sense , they represent physical quantities like degrees freedom , but with wider applicability .
after all , _
degrees freedom _ presumes not only multiplicities that are linear with energy on a log - log plot ( like a high - temperature einstein solid ) , but it also presumes quadratic energies ( i.e. energies proportional to a sum of squares of some `` randomly - occupied '' coordinates of state ) . the idea that `` every active coordinate gets @xmath70 '' , as we show below , even more strongly resists extension to systems with one or more entropy maxima .
we begin , however , with a non - quadratic example of less drastic proportion .
the debye heat capacity of a solid is one case where @xmath63 depends strongly on temperature@xcite . in the debye low temperature limit
, one has @xmath71 , so that @xmath72 , while @xmath73 . here of course , @xmath74 is the debye temperature related to the density and speed of sound in the solid .
note that in this limit , only a quarter of @xmath63comes from equipartition ( @xmath51 ) , the remaining three quarters from the time derivative of @xmath75 ( in effect , from the unfreezing of new modes of energy accomodation ) .
as you can see from figure 1 , such unfreezing is associated with a lowering of the thermal energy zero locally referenced by the heat capacity .
thus attempts to infer @xmath43 from the heat capacity by assuming that @xmath76 yield a 4-fold overestimate of the number of degrees of freedom !
this overestimate decreases as temperatures work themselves up to and beyond the debye temperature @xmath74 of the solid , as illustrated in figure 1 for temperatures well above @xmath74 .
there in the high temperature limit , @xmath77 , as one expects from a classical lattice model above with @xmath78 .
a system more challenging to the traditional interpretation of `` degrees freedom '' is that for a system of @xmath14 half - integral spins ( i.e. a two - state paramagnet ) of orientation energy @xmath79 .
i like it because , as dan schroeder says@xcite , `` it forces us to think primarily in terms of entropy rather than temperature '' .
begin with any system whose energy - storing coordinates ( e.g. displacements in a potential field ) in practice have an upper limit on the amount of energy they ll accomodate . as long as energy is low enough that no single coordinate approaches the maximum value , then we may well find behavior very much like that of the systems discussed above .
however , when individual coordinate energies begin to approach their maximum ( this happens quickly for two - state paramagnets whose coordinates accomodate but one unit of energy ) , things change fundamentally . in particular , there will be but one way ( neglecting degeneracies ) for the system to store the maximum amount of energy ( namely when each of the coordinates is fully energized ) . with but one way to accomodate either minimum energy or maximum energy , multiplicities will approach 1 at both endpoints of the continuum . since large systems may have many ways to store intermediate amounts of energy , multiplicity ( and entropy ) as a function of system energy will have a maximum ( or _ maxima _ ) somewhere between . at such maxima
, @xmath80 will be zero , while on their high - energy side @xmath81 will be negative ( signaling `` population - inverted '' states not accessible by thermal contact with reservoirs at positive absolute temperature ) .
taylor - expanding to second order about the energy ( @xmath82 ) of such multiplicity maxima gives @xmath83 in the neighborhood as @xmath84 , where @xmath85 .
hence for energies near @xmath82 , @xmath86 . thus deviations from @xmath82 are negative at positive absolute temperature , and ( for small deviations at least ) are proportional to reciprocal temperature @xmath87 , as illustrated in figure 3 .
of course , this law was discovered by pierre curie in experimental study of magnetization , and bears his name . for two - state paramagnets , @xmath88 and @xmath89 , where @xmath90 is the energy of alignment per spin ( magnetic moment times magnetic field strength ) .
thus for @xmath1 near @xmath91 , @xmath92 , while @xmath93 . since @xmath94 for these systems may be positive or negative , @xmath43 will be negative for some @xmath95 values regardless of our choice of the thermal energy zero ! although negative `` degrees freedom '' may cause discomfort for some , a negative value for @xmath43 should disturb no one since , to paraphrase a related comment by schroeder@xcite , there s no law of physics guaranteeing that there will not be fewer ways to distribute energy , as more energy is added .
in fact , as we now know ,
temperature and reciprocal temperature are simply different forms for the lagrange multiplier that characterizes a system s willingness to share thermal energy@xcite . systems , like these spin systems , capable of taking on ( and sharing energy from ) negative absolute temperature states show clearly that for them reciprocal temperature has more fundamental significance , and that the `` absolute zeros '' of temperature ( approached from negative or positive directions ) are indeed at opposite ends of a continuum@xcite .
but if reciprocal temperature is more fundamental , our instantaneous no - work heat capacity should be no less simply connected to the log - log derivative via reciprocal temperature .
rearrangement of the equation above shows that indeed this is the case .
@xmath96 = \left [ 1-\beta \frac \partial { % \partial \beta } \right ] \left\ { \frac{\partial ( \ln \omega ) } { \partial \left ( \ln e\right ) } \right\ } = -\beta \frac{\partial \ln \omega } { \partial \beta } = -\beta \frac{\partial ( s / k)}{\partial \beta } \text{.}\ ] ] had we historically adopted as our measure of `` willingness to share energy '' some other power of the uncertainty slope , say @xmath97 where @xmath98 , the instantaneous heat capacity would have remained proportional to the log - log derivative of multiplicity with respect to that measure as well .
this version of the relation now lets us simplify our perspective on entropy maxima . in the continuum ( stirling )
approximation for spin system accessible states @xmath99 , and measuring energy from the `` low - energy side '' , equation ( [ temperature ] ) yields @xmath100 where @xmath101 , so that @xmath102 and @xmath103 ^ 2e^xn$ ] .
here of course , @xmath104 takes on positive and negative values , ranging from around @xmath105 to @xmath106 respectively for orientation energies @xmath1 with allowed values from @xmath107 to @xmath108 . as you can see from the plot in fig .
3 , again @xmath63 overestimates @xmath75 at low temperatures ( high values of @xmath104 ) , although the estimate becomes exact when @xmath104 decreases to around @xmath109 ( the solution of @xmath110 ) .
after this @xmath63 _ underestimates _
@xmath51 , which begins to decrease as @xmath111 decreases and @xmath50 increases from this point .
all of this switches again when @xmath112 passes through zero , since @xmath63 remains positive while the change in uncertainty per @xmath113-fold increase in energy ( @xmath114 ) becomes negative since uncertainty about the system state _ decreases _ with added energy past this point .
also , of course , average heat capacity goes to zero and no longer equals @xmath51 , since the average must be obtained piecewise when temperature ( unlike reciprocal temperature ) breaches the discontinuity from plus infinity to minus infinity .
thus in addition to information units for heat capacity , we gain from this approach a way to visualize the limits of equipartition , and minimize consternation over negative degrees of freedom ( e.g. for spin and virial systems ) as well .
in short , we ve looked here at natural ( as distinct from historical ) units for the common thermodynamic quantities , so that we might explore the possibility that common uses of @xmath50 and @xmath115 , as measures of `` willingness to share thermal energy '' , have inherited their present emphasis partly because they predate our present understanding of multiplicity ( the @xmath116 in @xmath117 on josiah willard gibb s tombstone ) .
heat capacity is a particularly knotty concept in this regard , since for most of us it has always been a change in energy `` per degree kelvin '' . in fundamental units ,
if heat capacity has any dimensions at all they are those of the base-@xmath3 information units , since `` change in energy per unit change in energy per bit '' leaves us with nothing but bits in the bargain .
we point out a simple interpretation for the result , namely that thermal energy @xmath1 , divided by temperature @xmath118 , is fundamentally @xmath119 , i.e. the log - log derivative of multiplicity with respect to energy ( e.g. a measure of the bits of uncertainty increase per two - fold increase in energy ) .
the instantaneous no - work heat capacity in this context is @xmath120 , an estimate of @xmath121 with an energy - zero inferred from the `` local slope '' of the log - log plot .
we show that these two quantities bear a simple relationship to each other , regardless of the variable ( e.g. @xmath50 or @xmath69 ) chosen to keep track of a system s willingness to share energy thermally .
the former of the quantities , namely @xmath121 , plays the role of `` degrees freedom over two '' in quadratic systems , but is dependent on the energy zero , and regardless can take on negative values in systems with entropy maxima .
its limitations are those of the concept of equipartition itself .
the latter quantity , namely @xmath63 , provides deep insight into the ways a system accomodates _ new _ thermal energy .
because these quantities are defined in terms of state multiplicity and the conserved variable being shared ( in this case energy ) , and relatively independent of the form chosen for the lagrange multiplier in the problem ( e.g. temperature in historical units ) , their _ analogs _ in problems that involve the sharing of other conserved quantities ( e.g. volume , particles , or even dollars ) may be easier to recognize and put to use as well .
this work has benefited indirectly from support by the u.s .
department of energy , the missouri research board , as well as monsanto and memc electronic materials companies .
it has benefited most , however , from the interest and support of students at um - st .
louis .
e. f. taylor , the boundaries of nature , special and general relativity and quantum mechanics : a second course in physics , oersted medal lecture , _ the american journal of physics _ * 66 * , 369 - 376 ( may 1998 ) . c. e. shannon , _ bell system tech .
j. _ * 27 * , 379 , 623 ( 1949 ) ; these papers are reprinted in c. e. shannon and w. weaver , * the mathematical theory of communication * ( university of illinois press , urbana , 1949 ) . | information theory this century has clarified the 19th century work of gibbs , and has shown that natural units for temperature kt , defined via @xmath0 , are energy per nat of information uncertainty .
this means that ( for _ any _ system ) the total thermal energy @xmath1 over @xmath2 is the log - log derivative of multiplicity with respect to energy , and ( for _ all _ b ) the number of base-@xmath3 units of information lost about the state of the system per @xmath3-fold increase in the amount of thermal energy therein . for `` un - inverted '' ( @xmath4 ) systems
, @xmath5 is also a temperature - averaged heat capacity , equaling `` degrees - freedom over two '' for the quadratic case . in similar units
the work - free differential heat capacity @xmath6 is a `` local version '' of this log - log derivative , equal to bits of uncertainty gained per 2-fold increase in temperature .
this makes @xmath6 ( unlike @xmath5 ) independent of the energy zero , explaining in statistical terms its usefulness for detecting both phase changes and quadratic modes . from umstl - cme-94a09pf |
star formation in the milky way and other galaxies is described as a ( mass and size ) scale - free , hierarchical process , in which turbulent gas forms large - scale structures with a mass distribution following a power - law .
in essence , such a scale - free process leads to a mass and size fractal distribution . as a consequence ,
young stellar groupings are clustered according to hierarchical patterns , with the great star complexes ( associated with the @xmath12 superclouds ) at the largest scales and the ob associations and subgroups , small loose groups , clusters and cluster subclumps ( e.g. @xcite ) at the smallest . in several galaxies ,
the interstellar gas appears to follow a fractal structure ranging from the subpc ( @xmath13 the current resolution limit ) to the kpc scales ; if star formation occurs preferentially at the densest regions , stars should form following such patterns ( e.g. @xcite and references therein ) . in this context ,
star clusters , formed at the core ( i.e. the densest regions ) of giant molecular clouds , can be taken as the unavoidable star formation product in a hierarchically structured gas ( e.g. @xcite ) .
a similar picture , in which star clusters are present in dense cores , emerges from numerical simulations that follow in time the collapse of gas clouds ( e.g. @xcite ) , which also occurs when effects of radiative feedback and magnetic fields are included ( @xcite ) . .
[ fig1 ] in a hierarchical scenario , the turbulent gas forms large - scale structures ( clusters and loose groups ) with a mass distribution following a power - law of negative slope , i.e. @xmath14 , with @xmath15 , consistent with the mass distribution functions measured in several galaxies ( @xcite ) .
recent studies came up with robust evidence indicating that star - forming regions are indeed hierarchically structured , for instance in the nearby spiral galaxies m33 ( @xcite ) , m51 ( @xcite ) , and ngc628 ( @xcite ) , the local group dwarf irregular galaxy ngc6822 ( @xcite ) , the galactic disk ( @xcite ) , and the gould belt ( @xcite ) .
given the relative proximity , the magellanic clouds are an excellent environment to investigate the above issues .
for instance , @xcite found that the average age difference between pairs of lmc clusters increases as a function of their distance , which implies hierarchical star formation coupled to evolutionary effects .
the angular correlation of lmc stellar populations for separations between 2 ( @xmath16pc ) and 40 ( @xmath17pc ) also implies large - scale hierarchical structure in current star formation ( @xcite ) .
the character of the lmc hi structure as a function of scale , the filamentary and patchy structures of the high- and low - emission regions respectively , suggest that most of the ism is fractal , presumably the result of pervasive turbulence , self - gravity , and self - similar stirring @xcite ) .
more recently , @xcite found a highly substructured and rapidly evolving distribution in the lmc stars .
they suggest that all of the original structure is erased in @xmath18myr ( approximately the lmc crossing time ) , with small - scale structures mixing first .
similar conclusions apply to the smc , in which stars appear to have formed with a high degree of ( fractal ) sub - structure , possibly imprinted by the turbulent nature of the parent gas ; these structures are subsequently erased by random motions in the galactic potential on a time - scale of a crossing time through the galaxy ( @xcite ) . in this paper
we investigate the degree of spatial correlation among the different kinds of lmc and smc extended structures listed in the updated catalogue of @xcite , together with its relation to star formation .
we also study properties of their size distribution functions .
only two wide - apart age ranges are used for spatial correlation purposes : _ ( i ) _ very young objects ( not older than @xmath19myr , and probably younger than @xmath20myr ) , which encompass clusters related to nebular emission and associations related or not to emission , as classified and catalogued from sky survey plates by @xcite and references therein , and _
( ii ) _ old clusters ( older than @xmath21myr ) . according to our definition ,
the dynamical age of the very young clusters is lower ( sect .
[ 2pcf ] ) than the crossing time ( of the host galaxy ) , while for the old ones it corresponds to several crossing times , which is important for interpreting the spatial correlation in different time periods .
clusters within the wide age range @xmath22myr are not used in the spatial correlation analysis ( sect .
[ 2pcf ] ) . only the magellanic clouds have so far such a deep , homogeneous information on star clusters , associations and nebulae .
exceptions are some neighbouring dwarf galaxies that have been surveyed and are _ ( i ) _ featureless ( ursa minor ) , _ ( ii ) _ contain a few globular clusters ( fornax ) or , _
( iii ) _ star - forming events like in the clouds ( e.g. ngc6822 - @xcite ) .
this paper is organised as follows . in sect .
[ upcat ] we briefly discuss the updated magellanic system catalogue . in sect . [ osc ] we describe the selection criteria for star clusters older than the hyades . in sect .
[ distribfunc ] we discuss the size ( and mass , for star clusters ) distribution functions of the different classes of objects . in sect .
[ 2pcf ] we examine the spatial correlation of the different structures by means of two - point correlation functions . concluding remarks
are given in sect .
[ conclu ] .
properties of the updated mc catalogue are fully discussed in @xcite .
we recall here the basic statistical properties . taking the lmc , smc and the bridge together ,
the updated catalogue contains , respectively , 3740 classical star clusters , 3326 associations , 1445 emission nebulae , and 794 hi shells and supershells . with the recent additions and cross - identifications
, @xcite contains about 12% more objects than those in @xcite and @xcite together .
lcrcccrcccrccl & & & & & & + + class & age & n & @xmath1&@xmath23 & & n & @xmath1&@xmath23 & & n & @xmath1&@xmath23 & comments + & ( myr ) & & ( pc ) & & & & ( pc ) & & & & ( pc ) + c&any & 2268 & 38 & @xmath24 & & 456 & 30 & @xmath25 & & 2724 & 38 & @xmath26 & ordinary cluster + cn&@xmath4 & 81 & 25 & @xmath27 & & 9 & 8 & @xmath28 & & 90 & 25 & @xmath29 & cluster in nebula + ca&@xmath30 & 738 & 21 & @xmath31 & & 110 & 11 & @xmath32 & & 848 & 21 & @xmath33 & cluster similar to assoc .
+ ac&@xmath34 & 1185 & 32 & @xmath35 & & 60 & 14 & @xmath36 & & 1245 & 32 & @xmath37 & assoc .
similar to cluster + nc&@xmath38 & 183 & 16 & @xmath39 & & 72 & 8 & @xmath40 & & 255 & 16 & @xmath41 & nebula w / prob
cluster + clusters & & 4455 & 38 & @xmath42 & & 707 & 30 & @xmath43 & & 5162 & 38 & @xmath44 & c+cn+ca+nc+ac + a & @xmath45 & 1476 & 171 & @xmath46 & & 130 & 292 & @xmath47 & & 1606 & 292 & @xmath48 & ordinary association + an & @xmath4 & 217 & 262 & @xmath49 & & 39 & 62 & @xmath50 & & 256 & 265 & @xmath51 & association w / nebular traces + na & @xmath38 & 817 & 472 & @xmath52 & & 169 & 283 & @xmath53 & & 986 & 472 & @xmath54 & nebula w / embedded assoc .
+ dan@xmath55dnc@xmath56 & @xmath38 & 77 & 400 & @xmath57 & & 33 & 144 & @xmath58 & & 110 & 400 & @xmath59 & decoupled structures + non - clusters & & 2587 & 472 & @xmath60 & & 371 & 288 & @xmath61 & & 2958 & 472 & @xmath62 & a+an+na+dan+dcn + snr & & 52 & 78 & @xmath63 & & 22 & 38 & @xmath64 & & 74 & 78 & @xmath65 & supernova remnants + hi shells & & 124 & 472 & @xmath66 & & 545 & 482 & @xmath67 & & 794 & 477 & @xmath68 & hi shells and supershells + table notes .
n is the number of objects ; @xmath23 is the power - law slope ( @xmath69 ) fitted to the large radii range ( sect .
[ distribfunc ] ) ; @xmath1 is the maximum radius measured in each class .
( @xmath70 ) - small cluster or association in large nebula . dcn and dan : the nebular and stellar component of the objects can be distinguished on sky survey plates . especially in view of the spatial correlation analysis ( sect .
[ 2pcf ] ) , in this paper we restrict the object selection to the lmc and smc , not including bridge or extended wing structures . a census of the lmc and smc extended structures is provided in table [ tab1 ] , separated according to object class and including the probable age range .
we note that , based on similarities observed in the size distributions ( sect .
[ distribfunc ] ) , in the present paper we include the ac and nc classes ( relatively young objects ) into the cluster classification , thus resulting in a higher number of such objects than quoted in @xcite .
besides the latter two classes , the cluster classification also contains the c , cn and ca classes . as non - clusters ( structures mostly associated with star formation environments ) we take the a , an , na , dan and dnc classes ( see table [ tab1 ] notes ) .
the snr and hi shells are not used because they are object classes apart and their size distribution functions are significantly different from those of the clusters and non - clusters ( fig . [ fig4 ] ) .
figure [ fig1 ] ( top panels ) shows the angular distribution of the 4455 lmc clusters ( left ) and 2587 non - clusters ( right ) .
both kinds of structures trace well - known lmc ( and smc ) structures ( e.g. @xcite and references therein ) .
it is also clear that the non - clusters appear to present a high degree of spatial correlation , with most of them tightly clumped together .
this applies as well to the clusters , but to a lesser degree , because young and old clusters present significantly different levels of spatial self - correlation , the latter being essentially non - correlated ( sect .
[ 2pcf ] ) .
when the angular sizes are considered ( bottom panels ) , we see that most structures are arranged according to complex patterns , with sub - structures located inside larger ones .
the identification , characterisation and spatial distribution of old star clusters in the clouds has been a major concern throughout decades ( e.g. @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) . by old or red star clusters we mean those older than the hyades ( @xmath71myr ) , or intermediate age clusters ( iacs ) up to classical globular cluster ages .
we adopted the definition of old star cluster by @xcite and @xcite .
magellanic cloud clusters about this age appear to show dynamically evolved surface density profiles ( e.g. @xcite ; @xcite ; @xcite ) .
since the cluster age distribution function drops significantly with age ( see , e.g. , fig . 2 of @xcite for the magellanic clouds ) ,
the presently adopted old cluster definition encompasses a statistically more significant sub - sample ( sect .
[ cnstru ] ) than what would result for , e.g. clusters older than 1gyr .
clusters are expected to mix up by random motions under the galactic potential on a time - scale of a crossing time that , for the smc , is of the order of 75myr ( @xcite ) , and about twice that value for the lmc .
thus , the old clusters as defined above have ages that correspond to several crossing times of the respective galaxy , and any memory of the clumpy structures where they were born should have been erased . in this sense , they can be used as probes of the long - term behaviour of the cluster spatial correlation ( sect .
[ 2pcf ] ) .
@xcite and @xcite discovered luminous and intermediate luminosity clusters in the smc using plate material . @xcite and @xcite discovered intermediate and low luminosity clusters , while @xcite discovered even fainter ones by means of 4 m telescope plates .
@xcite discovered low luminosity clusters on sky survey plates , while @xcite discovered low luminosity clusters by means of ccd imaging .
@xcite identified 35 luminous old clusters by means of non - calibrated cmds , of which 11 were discoveries . @xcite and @xcite discovered most of the luminous and intermediate - luminosity clusters in the lmc , the latter work being dedicated to the outer parts .
@xcite discovered intermediate - luminosity clusters , while @xcite low - luminosity ones with 4m - telescope plates .
@xcite discovered low - luminosity clusters in the outer parts .
@xcite discovered additional intermediate and low - luminosity clusters , while @xcite discovered a large number of low luminosity clusters on sky survey plates .
@xcite discovered low - luminosity clusters in the lmc with ccd observations .
@xcite and references therein have cross - identified these catalogues and a number of other studies , and is particularly suitable as a starting point for a deeper new survey such as the visible and infrared survey telescope for astronomy ( vista ) .
it also provides a mean to properly acknowledge previous discoveries and to unambiguously establish new cluster findings .
@xcite serendipitously detected two faint clusters in an lmc bar field using hst .
the clusters have masses comparable to those of galactic open clusters and ages in the range @xmath72myr .
the clusters are extremely faint on dss and xdss images , which suggests that the clouds might harbour an important open cluster counterpart population .
besides being a powerful tool to explore probable red brighter and intermediate - luminosity star clusters in the clouds ( table [ tab2 ] ) , vista will be essential also to detect such a possible population of open cluster counterparts , and estimate their age distribution . based on the broad - band ubvr photometry of @xcite , @xcite derived absolute values of age and mass for a sample of lmc star clusters to study the cluster formation rate , their characteristic disruption time - scale and the cluster mass function in different mass ranges .
the same sample was used for further investigation of the cluster formation rate and the disruption time - scale by @xcite .
the same method was applied to a sample of smc clusters by @xcite to study the _ infant mortality_. our approach in this paper differs in several ways , since we intend to build statistically significant samples of clusters ( as well as associations and emission nebulae ) characterised by very different age ranges .
we compiled ages from the literature later than 1988 , as determined from cmds .
there are 85 and 202 old clusters with ages derived from cmds for the smc and lmc respectively , and they are provided in table [ tab2 ] is available only in electronic format . here
we provide an excerpt , for illustrative purposes . ] .
columns 1 to 8 of this table contain the same information as the general catalogue ( @xcite ) .
we now introduced additional columns that provide the age determination method ( col .
9 ) , @xmath73 ( col .
10 ) , and the relevant references for the age ( col .
note that several references are compilations themselves , so more references are therein .
lrrcccccccl designations&@xmath74$]&@xmath75$]&class&a&b&pa&classification & method&log(age)&ref + & ( @xmath76)&()&&()&()&()&&&(yr ) + ( 1 ) & ( 2 ) & ( 3 ) & ( 4 ) & ( 5 ) & ( 6 ) & ( 7 ) & ( 8) & ( 9 ) & ( 10 ) & ( 11 ) + + am-3,eso28sc4 & 23:48:59 & -72:56:43 & c & 0.90 & 0.90 & - & old iac & cmd & 9.74 & r3 + l1,eso28sc8 & 0:03:54 & -73:28:19 & c & 4.60 & 4.60 & - & globular cluster & cmd & 9.95 & r12 + " & & & & & & & & cmd & 9.88 & r26 + l2 & 0:12:55 & -73:29:15 & c & 1.20 & 1.20 & - & & pla & red & r55 + l3,eso28sc13 & 0:18:25 & -74:19:07 & c & 1.00 & 1.00 & - & & pla & red & r36 + k1,l4,eso28sc15 & 0:21:27 & -73:44:55 & c & 2.20 & 2.20 & - & & cmd & 9.49 & r18 + bologna a & 0:21:31 & -71:56:07 & c & 0.80 & 0.80 & - & sup 47 tucanae & cmd & iac & r56 + l5,eso28sc16 & 0:22:40 & -75:04:29 & c & 1.10 & 1.10 & - & old iac & cmd & 9.61 & r18 + + ngc1466,sl1,lw1,eso54sc16,kmhk1 & 3:44:33 & -71:40:17 & c & 3.50 & 3.50 & - & globular cluster & cmd & 10.17 & r41 + sl2,lw2,kmhk2 & 4:24:09 & -72:34:23 & c & 1.60 & 1.60 & - & & pla & red & r55 + kmhk3 & 4:29:34 & -68:21:22 & c & 0.80 & 0.80 & - & & pla & red & r55 + ngc1629,sl3,lw3,eso55sc24,kmhk4 & 4:29:36 & -71:50:18 & c & 1.70 & 1.70 & - & & col & red & r5 + hs8,kmhk5 & 4:30:39 & -66:57:25 & c & 0.80 & 0.80 & 30 & & pla & red & r55 + sl4,lw4,kmhk7 & 4:32:38 & -72:20:27 & c & 1.70 & 1.70 & - & & cmd & 9.23 & r6 + kmhk6 & 4:32:48 & -71:27:30 & c & 0.60 & 0.55 & 80 & & pla & red & r55 + table notes .
cols . 5 and 6 : semimajor axes @xmath77 and @xmath78 ; col . 7 : position angle ; col . 9 : old age method . col . 10 : @xmath79 when available , or age class .
references ( col .
11 ) : r3 : @xcite ; r5 : @xcite ; r6 : @xcite - relative ages ; r12 : @xcite ; r18 : @xcite ; r26 : @xcite ; r36 : @xcite , @xcite ; r41 : @xcite ; r55 : present paper - red ( old ) cluster by plate inspection ; r56 : @xcite .
we employed observed ( @xcite ) , reddening - corrected ( @xcite ) integrated colours , and swb types to identify old clusters ( typically swb ivb or later , @xcite ) , for clusters that still lack cmd ages .
we also included results from integrated spectroscopy ( @xcite ) .
by inspection of dss and xdss images we excluded clusters with apparent contamination by relatively bright stars , concerning integrated colours and spectra .
we found 41 and 117 old clusters in the smc and lmc respectively from integrated colours ( table [ tab2 ] ) . finally , following @xcite and @xcite , we examined blue and red dss and xdss images , and eso film sky survey plates to identify red clusters .
it is remarkable how the red smc clusters by @xcite his types t1 and t2 have been confirmed as old clusters by means of deep cmds .
brck disposed of u plates to help the classification .
we dispose of blue and red plates , where it was basically possible to recognise clusters with brighter stars red from the rgb or bluer ms stars .
also , blue clusters have as rule more irregular angular distributions .
most clusters that we examined by this simple method are in the outer parts of the lmc . by means of integrated colours
, @xcite found that the outer lmc disk appears to be essentially composed of old clusters .
our goal here is to provide a sample of probable red clusters suitable to correlation function tests , and to isolate that sample for cmd studies in view of vista and other large telescopes .
there are 19 and 203 clusters , respectively in the smc and lmc , that are probably old ( red ) from our plate inspections .
table [ tab2 ] also includes rather populous clusters that have cmds , integrated colours or plate diagnostics pointing to a blue - red transition cluster , that occurs around 500 myr .
the sudden or perhaps rather smooth integrated colour change is expected from the so - called agb and rgb phase transitions ( e.g. @xcite and references therein ) .
the present sample is a new one for such purposes . to minimise ambiguous age determinations ,
we have not included the blue - red transition clusters in our spatial correlation study , although they are , in principle , also old enough for dynamical purposes . also , we note that @xcite decontaminated the clusters containing atypical bright stars superimposed .
we examined all clusters showing red colours from @xcite and @xcite on sky survey plates , and excluded those that appeared to be dominated by one or a few bright stars .
we may have excluded some faint intrinsic old clusters with one or a couple bright agb stars , but such stars are rare even in populous magellanic cloud clusters ( e.g. @xcite ) . in summary , there are 522 star clusters in the lmc that can be currently considered as old as , or older than the hyades .
the smc contains 145 such cases .
considering the lmc and smc together , the total sample of old / red clusters corresponds to a fraction of @xmath80 of the cluster - like structures ( table [ tab2 ] ) .
details on the angular distribution of this sub - sample are shown in fig .
the old cmd sample shows a well - defined lmc bar , while the smc probably shows a thick edge - on disk ( @xcite and references therein ) .
red integrated colours complement these samples mostly for fainter clusters .
the lmc plate sample corresponds essentially to the outer disk .
we emphasise that the present sharp inner border of the old sample is an artifact , but not the outer ring structure , as can be seen in the most recent census of clusters and related objects ( @xcite ) .
the outer lmc disk ring is a real feature , probably produced as a consequence of the last lmc / smc encounter that took place @xmath81myr ago ( @xcite ) .
this structure is present in the uniform plate survey by @xcite and in that by @xcite .
the magnitude - limited integrated photometry of lmc clusters by @xcite also showed this structure for the oldest age group . in the present study ,
essentially all known red clusters in the outer lmc are included in that locus .
the geometries of the sub - samples were established by each survey , but have apparently not affected the correlation functions , as shown by the tests with different old cluster subsamples ( sect .
[ 2pcf ] ) .
in their surveys , @xcite , @xcite , and @xcite employed film copies of the eso schmidt telescope red survey and the uk schmidt telescope serc - j ( blue band ) survey in australia .
the red plates trace emission nebulae by means of h@xmath23 .
the limiting magnitudes are r=21.5 and bj=22.5 , respectively .
thus the detection limit of stars in clusters is very deep , especially in j. however , only the new generation of ccd surveys will permit to quantify the completeness of those samples compiled or discovered by our group , and certainly to explore an as yet undetected population of fainter objects .
the spatial distribution of interstellar gas follows a fractal structure ranging over many scales , from the subparsec at the smallest to the cluster and star complexes at the largest .
this suggests that , if stars are formed mostly in the densest regions , they should also form in fractal patterns ( e.g. @xcite and references therein ) .
indeed , the power - law nature of the size distribution function has been observed in galactic giant molecular clouds ( @xcite ) and cloud clumps ( @xcite ) .
the end result of this process is that the cluster size distribution should follow a declining power - law with size , a behaviour that has been observed in several galaxies ( e.g. @xcite ; @xcite ; @xcite ) .
our first approach in the investigation of the hierarchical structures in the clouds is by means of the size ( i.e. absolute radius , ) distribution functions @xmath82 .
we build @xmath83 individually for all classes of objects listed in table [ tab1 ] based on the apparent major and minor axes given in the updated catalogue , together with the cloud distances @xmath84kpc and @xmath85kpc ( e.g. @xcite ) . although the disks of both clouds are inclined with respect to the line of sight , the effect of the distance correction on the absolute radius distribution is small , to within the error bars ( see app .
[ apxa ] ) .
the radius distribution functions for the cluster - like structures ( fig . [ fig3 ] ) are characterised by a steep decline for @xmath86pc , which corresponds to about 0.25 . as we discuss in app .
[ apxb ] , observational incompleteness probably accounts for the shape of @xmath83 in the small - size range , @xmath87pc .
the maximum radius ( @xmath1 ) reached by the cluster - like structures in the lmc is @xmath88pc , and @xmath89pc in the smc .
power - law fits ( @xmath90 ) to the incompleteness - unaffected range are obtained with rather steep slopes , @xmath91 , especially for the ( statistically ) well - defined distributions .
the values of @xmath1 and @xmath23 are given in table [ tab1 ] .
figure [ fig4 ] shows @xmath83 for the remaining object classes . except for the snr ,
the distributions are qualitatively similar to those in fig .
[ fig3 ] , with a decline for large radii .
however , compared to the cluster - like classes , the power - law slopes are significantly shallower ( @xmath92 ) and the maximum radii are @xmath93 times as large , reaching @xmath94pc in the lmc and @xmath95pc in the smc .
the incompleteness - related turnover for the a , an , and na classes occurs at the same radii as that in the cluster - like objects .
the hi shells , on the other hand , have a turnover at @xmath96pc ( @xmath97 ) , which might reflect a real effect , not related to completeness .
based on similarities of @xmath1 and @xmath23 , we define the c , ca , cn , nc , and ac classes as cluster - like structures , while a , an , na , dnc , and dan as non - clusters .
their composite radius distributions are shown in fig .
[ fig5 ] , together with the power - law fit .
as expected from the above discussion , the cluster - like slopes for the lmc , smc , and lmc@xmath98smc distributions ( @xmath99 ) are significantly steeper than the corresponding ones derived for the non - clusters ( @xmath100 ) .
figure [ fig5 ] also shows the distributions obtained by adding all the structures , including the snr and hi shells . while most of the individual features are preserved , the smc profile , on the other hand , now requires two different power - laws to be described .
the slopes in the radius distribution of the non - clusters are consistent with those measured for hii regions in spiral galaxies ( @xcite ) .
we derive the size distribution functions of the young and old star cluster population .
we take the cn ( age @xmath4myr ) and nc ( age @xmath38myr ) classes ( table [ tab1 ] ) to represent the young star clusters .
the old ( age @xmath7myr ) ones were selected according to the criteria discussed in sect .
[ osc ] .
significant differences are observed in the size distribution functions ( figure [ fig5 ] ) , especially in the lmc .
the distribution function of the young clusters falls off with radius at a steeper rate than the old ones , reaching a maximum size ( @xmath101pc ) less than half of that reached by the old ones ( @xmath88pc ) . in the statistically more significant distributions of the lmc and smc combined ( right panels ) , the young clusters fall off with the slope @xmath5 , while the old ones have @xmath8 .
the differences in @xmath1 and slope probably reflect the several @xmath102yr of dynamical evolution of the old clusters , a consequence of which is an expansion of the outer parts ( e.g. @xcite ) of the clusters that survive the infant mortality ( e.g. @xcite ) phase , both in the the smc ( @xcite ) and lmc ( @xcite ) . ] .
the size distribution functions of the young and old clusters fall off at a steeper rate ( @xmath103 ) than the non - clusters ( @xmath104 ) .
hierarchically structured gas is expected to form star clusters with mass distributed according to a power - law of the form @xmath105 , with @xmath15 ( e.g. @xcite ) .
below we apply a simple method to analytically transform the cluster radius distribution function into a mass distribution .
we wish to test if our sample of lmc and smc clusters basically follow the above mass distribution .
we caution that our approach to the mass distribution is a simplification , since we do not take into account individual mass - to - light (
m / l ) ratios , which are known to vary considerably between young and old clusters ( e.g. @xcite ; @xcite ; @xcite ) .
however , the presently extracted sample ( @xcite ) is by far ( @xmath106 ) dominated by young clusters and , thus , large variations of the m / l ratio are not expected .
besides , instead of computing individual masses from integrated luminosity , we use scaling relations that apply well to a wide variety ( in terms of age , mass , and size ) of galactic star clusters to transform one kind of distribution function into another . in this sense
, we expect that the adopted radius to mass transformation is representative of the average m / l ratio of the clusters .
we start by assuming a spherical star cluster with a mass radial density profile that can be described by a king - like function @xmath107 , where @xmath108 is the surface mass density at the cluster centre and is the core radius .
we also consider that essentially all stars are contained within @xmath109 , where is the cluster radius .
the spatial mass density of such a structure can be computed from inversion of abell s integral , @xmath110^{3/2}.\ ] ] thus , the cluster mass can be computed from @xmath111 @xmath112.\ ] ] galactic star clusters with ages from a few myr to @xmath113gyr , masses within @xmath114 , and radii within @xmath115pc , have the relation between and well approximated by @xmath116 ( e.g. @xcite ) .
comparable ratios are observed in lmc and smc star clusters ( e.g. @xcite ; @xcite ; @xcite ) . under these assumptions
we have @xmath117 .
this equation , together with central mass densities in the range @xmath118 , accounts for the distribution of cluster mass and core radius ( @xcite ) .
then , the transformation of the radius distribution to mass is given by @xmath119 , and the average cluster mass - density is a declining function of the cluster radius , @xmath120 . for a given
, the radius to mass scalings depend only on @xmath108 as @xmath121 and @xmath122 that , for different values of @xmath108 , preserve the shape of the mass distribution , only changing the mass values .
we use the above scaling relations to transform the cluster radius distribution functions ( fig .
[ fig5 ] ) into mass distributions , @xmath123 , separately for the lmc and smc , and the combination of both , lmc@xmath98smc .
we also consider the age ranges @xmath4myr , @xmath7myr , and clusters of all ages combined .
the average value ( galactic star clusters - @xcite ) of the central mass density , @xmath124 , is used to compute the mass distributions ( fig .
[ fig6 ] ) ; fit parameters are given in table [ tab3 ] .
the mass distribution of the very young clusters in both clouds falls off at a steeper rate towards large masses than that of the old ones , which is consistent with a mass - dependent disruption time - scale ( e.g. @xcite ) .
also , the slopes are steeper in the lmc than in the smc . these slopes are consistent with those of the mass distributions of star clusters in different galaxies ( @xcite ) .
crccccrccccrccc & & & & & + + age range & n & @xmath125 & @xmath126&@xmath127 & & n & @xmath125 & @xmath126&@xmath127 & & n & @xmath125 & @xmath126&@xmath127 + ( myr ) & & ( ) & ( ) & & & & ( ) & ( ) & & & & ( ) & ( ) & + @xmath4&201 & @xmath128&@xmath129&@xmath130&&47&@xmath128&@xmath131 & @xmath132&&198&@xmath128&@xmath129&@xmath133 + @xmath7&435&@xmath134&@xmath135&@xmath136&&135&@xmath128&@xmath137 & @xmath138&&552&@xmath134&@xmath137&@xmath139 + all ages&3700&@xmath128&@xmath135&@xmath140&&629&@xmath128&@xmath137 & @xmath141&&4271&@xmath128&@xmath137&@xmath142 + table notes .
n is the number of clusters used to fit the mass range @xmath143 with the power - law @xmath144 .
@xmath125 and @xmath126 computed for @xmath145 ; they scale linearly with @xmath108 .
the mass distributions of the very young clusters are characterised by a maximum mass of @xmath146 ( lmc ) and @xmath147 ( smc ) , while for the old ones it is @xmath148 in both clouds .
the latter values are considerably higher than the maximum mass of typical galactic open clusters ( e.g. @xcite ) .
the decline in the number of clusters with mass below @xmath125 ( table [ tab3 ] ) is probably related to observational incompleteness in the detection of small clusters ( sect .
[ distribfunc ] ) . for comparison purposes
we also show in fig .
[ fig6 ] the mass distributions for the lmc@xmath98smc clusters , as well as those corresponding to clusters of all ages , in which the basic features of the individual distributions are preserved .
interestingly , the maximum mass of the lmc clusters younger than @xmath149myr ( @xcite ) , and smc ones younger than @xmath93myr ( @xcite ) , is about 2.4 times higher than that of the very young lmc and smc clusters ( table [ tab3 ] ) . for lmc clusters younger than @xmath150gyr ( @xcite ) and smc ones younger than @xmath151gyr ( @xcite ) , the ratio increases to @xmath152 .
although the somewhat different age ranges , consistency between both sets of @xmath126 values can be reached with the central mass densities @xmath153 and @xmath154 , respectively for the very young and old clusters .
the somewhat higher values of @xmath108 in the mcs clusters , with respect to the galactic open clusters , is consistent with the relative cluster mass ranges encompassed by the lmc ( @xcite ) , smc ( @xcite ) , and milky way ( @xcite ) mass distributions . finally ,
if we take into account variations of m / l with age for the dominant ( in number ) young clusters ( e.g. @xcite ) , the actual mass values for clusters younger than @xmath155myr would be @xmath156 lower than the average-(m / l ) estimates above , and @xmath157 higher for those with age within @xmath158myr .
given that the number of clusters decreases with age ( see , e.g. @xcite , for the age - distribution of the mc clusters ) , our mass estimates in each bin of the mass distributions ( fig .
[ fig6 ] ) may be somewhat overestimated .
in a hierarchical scenario , young star clusters are expected to preserve some memory of the physical conditions prevailing in their birthplace .
because of random motions along many orbits under the galactic potential , the spatial distribution of old star clusters , on the other hand , should be very little reminiscent of the primordial one . according to this scenario ,
the frequency of young star clusters lying relatively close to each other - and to star - forming structures - should be higher than for the old ones . based on the 590 lmc clusters ( @xmath80 of the present sample size - table [ tab1 ] ) catalogued by @xcite with ages derived from the ubv colours by @xcite
, @xcite found that the average age difference between pairs of clusters increases with the separation , which they interpreted as resulting from star formation that is hierarchical in space and time . a similar result - and interpretation -
was found by @xcite for pairs of open clusters in the milky way disk .
we investigate this point further by means of the degree of spatial correlation among groups of objects characterised by different age ranges and sizes .
we use the young and old star clusters defined in sect .
[ yosc ] .
consider two groups of objects , @xmath159 and @xmath160 .
for each object in @xmath159 , we compute the angular separation with respect to all objects in @xmath160 . after applying the same procedure to all objects in @xmath159
, we build the two - point - correlation function ( 2pcf ) , which measures the fractional number of pairs @xmath161 that lie within a given separation @xmath162 and @xmath163 , @xmath164 . according to this definition ,
the 2pcf is simply the angular separation distribution function .
artificial 2pcfs built with samples of points that emulate both the geometry and object distribution of the lmc and smc are used to check the statistical significance of the spatial correlations . in the simulations we randomly select the right ascension ( @xmath23 ) and declination ( @xmath165)coordinates of a given point within the actual ranges spanned by each cloud ( fig .
[ fig1 ] ) and with the same number - frequency as the observed ones . irrespective of the adopted geometry
, a random distribution of objects would produce a number of neighbours within a given separation @xmath162 that increases as @xmath166 , at least for a maximum separation @xmath167 ( which should scale with the angular size of the simulated field ) .
thus , the 2pcfs should increase with @xmath162 as @xmath168 .
indeed , the 2pcfs derived with the simulations ( fig . [ fig7 ] ) present the expected dependence with separation for @xmath169 ( lmc ) and @xmath170pc ( smc ) . beyond these values
both the measured and simulated 2pcfs consistently drop , as a consequence of the limited size of the clouds . as a first step we compute the spatial self - correlation functions , in which @xmath171 , for the lmc and smc ( fig .
[ fig7 ] ) . compared to the simulated 2pcfs ,
the non - clusters ( top panels ) present a relatively high degree of spatial self - correlation for separations smaller than @xmath172pc ( lmc ) and @xmath173pc ( smc ) .
young ( age @xmath4myr ) clusters present a high degree of spatial self - correlation , from small to large scales ( middle panels ) , @xmath174pc ( lmc ) and @xmath173pc ( smc ) .
old ( age @xmath7myr ) clusters , on the other hand , have a very low degree of spatial self - correlation , restricted to separations @xmath175pc ( lmc ) and @xmath176pc ( smc ) .
interestingly , the same pattern is obtained with the 2pcfs computed for the clusters older than 1gyr , with the age determined from cmds .
some degree of spatial correlation at small separations among old clusters is expected , since the clouds contain binary and/or merger star clusters preferentially of comparable ages ( e.g. @xcite ; @xcite ; @xcite ) . in summary ,
young clusters have a probability of being clustered together significantly higher than old ones , both in the lmc and smc .
the dynamical age of clusters older than @xmath177myr corresponds to @xmath178 crossing times in the lmc , and @xmath179 in the smc , while for the young ones ( age @xmath4myr ) it is @xmath180 and @xmath181 , respectively for the lmc and smc .
given that a single crossing time is necessary to smear out most of the primordial structural pattern ( @xcite ) , the above conclusion is consistent with the old clusters having been mixed up by the random motions under the galactic potential along several @xmath102 years , while the young ones still trace most of the birthplace pattern .
now we test the degree of spatial correlation of the young and old clusters with the non - clusters ( star - formation environments ) .
as expected from the self - correlation analysis , the young clusters are highly correlated with the non - clusters ( bottom panels ) .
the old clusters , on the other hand , appear to have some spatial correlation with the non - clusters only at the very - small scales , @xmath182pc ( lmc ) and @xmath183pc ( smc ) .
part of this correlation may be due to projection effects on the bar . for larger separations
the 2pcfs can be accounted for by the random distribution of objects .
now we examine the spatial self - correlation among clusters and non - clusters of different radius ranges .
based on the respective radius distribution functions ( fig .
[ fig5 ] ) , we take @xmath184pc as the boundary between small and large clusters ; for the non - clusters we take the boundary at @xmath185pc . the derived correlation functions ( fig .
[ fig8 ] ) indicate that the small clusters are more spatially correlated than their large counterparts .
the same applies to the non - clusters .
again , this picture is what should be expected from a hierarchical structure .
finally , we investigate the effective separation of the young and old clusters with respect to the non - clusters .
we first compute the effective separation ( @xmath186 ) between a cluster and a non - cluster , which we define as the ratio of the angular distance ( @xmath162 , converted to the absolute scale ) to the radius of the non - cluster ( @xmath187 ) , @xmath188 . in this way , a cluster that is located inside a non - cluster has @xmath189 .
two - point correlation functions built with all the @xmath186 between samples @xmath159 and @xmath160 thus provide a measure of the clustering among objects in both samples . for comparison
we use simulated 2pcfs built for object samples ( young and old clusters and non - clusters ) with coordinates selected as described in sect .
we now also include absolute radii separately for each sample , randomly taken from the observed distributions ( fig .
[ fig5 ] ) .
the resulting 2pcfs are shown in fig .
consistently with the analyses of the previous sections , young clusters in both clouds present a high degree of clustering with the non - clusters , especially for small effective separations ( @xmath189 ) , but reaching as well high values of @xmath186 .
in all scales , the clustering degree of the old clusters with respect to the non - clusters , on the other hand , can be accounted for by a random distribution of old clusters .
the above results are consistent with a strong hierarchical structuring of the young star clusters in both clouds , including their time evolution effects .
in broad lines , when star formation occurs in turbulent gas , large - scale structures are expected to be produced following a power - law mass distribution ( @xmath190 - @xcite ) , and with hierarchically clustered young stellar groupings ( e.g. @xcite ; @xcite ) . in the present paper
we address the above issue by investigating the degree of spatial correlation among sets of lmc and smc extended structures , characterised by different properties , and its relation to star formation .
based on the catalogue of @xcite , we built sub - samples that basically contain star clusters ( young and old ) and nebular complexes ( and their stellar associations ) .
the latter structures are related to star - forming regions ; for simplicity , we refer to them as non - clusters . in all cases ( figs .
[ fig3]-[fig5 ] ) , the radius distribution functions follow a power - law ( @xmath0 ) decline for large radii with slopes that depend on object class ( and age ) . taking both clouds combined ,
the non - clusters fall - off with a slope @xmath2 and reach sizes of @xmath3pc .
old ( age @xmath7myr ) clusters present the somewhat steeper slope @xmath8 , while the young ( age @xmath4myr ) ones have the steepest slope @xmath5 .
the maximum size reached by clusters is less than @xmath191 of the non - clusters , with the old ones reaching a size @xmath192 bigger than the young ones .
the differences in slope and maximum size between the young and old clusters can be accounted for by long - term dynamical effects acting on the clusters . by means of a radius to mass
scaling ( sect .
[ mdf ] ) , we show that the mass distribution of the lmc and smc clusters follows @xmath105 ( fig .
[ fig6 ] ) , with @xmath15 . within the uncertainties ( table [ tab3 ] )
, this value agrees with the slope expected in a hierarchical scenario ( @xcite ) . also , the mass distribution for clusters younger than @xmath93myr falls off towards large masses faster than the clusters older than @xmath193myr . according to the two - point correlation functions ( sect .
[ 2pcf ] ) , the lmc and smc star clusters younger than @xmath93myr present a very high degree of spatial correlation among themselves and , especially , with the non - clusters ( fig . [ fig7 ] ) .
clusters older than the hyades ( @xmath7myr ) on the other hand , appear to have been heavily mixed up , probably because their ages correspond to several galactic crossing times and the strong perturbations associated with the lmc and smc encounters ( e.g. @xcite ) .
when the analysis is restricted to clusters older than 1gyr - with the age determined from cmds , the same conclusions are obtained .
considering two different radius ranges , we show that small clusters ( @xmath10pc ) and non - clusters ( @xmath11pc ) are spatially self - correlated , while the large ones are not ( fig .
[ fig8 ] ) . also , young clusters in both clouds present a very high degree of spatial clustering with the non - clusters , which does not occur with the old ones ( fig .
[ fig9 ] ) . the above results , expressed in terms of the spatial and size distribution of extended structures in the lmc and smc , are fully consistent with a hierarchical star - formation scenario , in which star complexes are part of a continuous star - formation hierarchy that follows the gas distribution .
similar conclusions drawn from different methods and samples of objects have been obtained for the lmc ( e.g. @xcite ; @xcite ; @xcite ; @xcite ) and the smc ( e.g. @xcite ; @xcite ) . vista and other large telescopes will certainly uncover a large number of faint clusters in the magellanic clouds , with masses comparable to the galactic open clusters . the same for embedded clusters .
cmds will provide accurate ages for them , as well as for many luminous , intermediate - luminosity , and low - luminosity clusters already catalogued .
also in this context , the catalogue by @xcite will be an essential tool , for having gathered and cross - identified the small and large structures in the magellanic clouds .
the catalogue is also useful to help establish discoveries .
the present work has provided as well a sub - catalogue of old and probable - old clusters , which can be useful also for vista studies .
we thank important suggestions made by the referee , dr .
richard de grijs .
we thank dr .
deidre hunter for providing us access to the catalogue of integrated photometry of smc and lmc clusters .
we acknowledge partial support from cnpq ( brazil ) .
aaronson m. & mould j. 1982 , apjs , 48 , 161 ahumada a.v . , clari j.j .
, bica e. & dutra c.m .
2002 , a&a , 393 , 855 balbinot e. , santiago b.x .
, bica e. & bonatto c. 2009 , mnras , 396 , 1596 bastian n. , gieles m. , lamers h.j.g.l.m . ,
scheepmaker r.a .
& de grijs r. 2005a , a&a , 431 , 905 bastian n. , gieles m. , efremov yu .
& lamers h.j.g.l.m .
2005b , a&a , 443 , 79 bastian n. , ercolano b. , gieles m. , rosolowsky e. , scheepmaker r.a . , gutermuth r. & efremov yu .
2007 , mnras , 379 , 1302 bastian n. , gieles m. , ercolano b. & gutermuth r. 2009 , mnras , 392 , 868 bate m.r . 2009 , mnras , 392 , 1363 bekki k. & chiba m. 2007 , pasa , 24 , 21 bellazzini m. , pancino e. & ferraro f.r .
2005 , a&a , 435 , 871 van den bergh s. 1981 , a&as , 46 , 79 bica e. , arimoto n. & alloin d. 1988 , a&a , 202 , 8 bica e. & schmitt h. 1995 , apjs , 101 , 41 bica e. , clari j.j . , dottori h. , santos jr . j.f.c . & piatti a.e .
1996 , apjs , 102 , 57 bica e. , schmitt h. , dutra c.m . &
oliveira , h.l .
1999 , aj , 117 , 238 bica e. & dutra c.m .
2000 , aj , 119 , 1214 bica e. , bonatto c. , dutra c.m . , & santos jr . , j.f.c .
2008 , mnras , 389 , 678 bica e. , santos jr .
schmidt a.a .
2008 , mnras , 391 , 915 bonatto c. & bica e. 2009a , mnras , 392 , 482 bonatto c. & bica e. 2009b , mnras , 397 , 1915 brck m.t . 1975 ,
mnras , 173 , 327 brck m.t .
1976 , orroe , 1 , 1 bruzual g. & charlot s. 2003 , mnras , 344 , 1000 carvalho l. , saurin t.a .
, bica e. , bonatto c. & schmidt a.a . 2008 , a&a , 485 , 71 charlot s. & bruzual g. 1991 , apj , 367 , 126 chiosi e. , vallenari a. , held e.v . , rizzi l. & moretti a. 2006 , a&a , 452 , 179 crowl h.h . , sarajedini a. , piatti a.e . , geisler d. , bica e. , clari j.j .
, santos jr .
2001 , aj , 122 , 220 da costa g.s . 1999 ,
iaus , 190 , 446 danks a.c .
1982 , a&a , 106 , 4 dieball a. , mller h. & grebel e.k .
2002 , a&a , 391 , 547 efremov yu.n .
1995 , aj , 110 , 2757 efremov yu.n . &
elmegreen b.g .
1998 , mnras , 299 , 588 elias f. , alfaro e.j . & cabrera - cao j. 2009 , mnras , 397 , 2 elmegreen b.g . & efremov yu.n .
1996 , apj , 466 , 802 elmegreen b.g . & falgarone e. 1996 , apj , 471 , 816 elmegreen d.m . &
salzer j.j . 1999 ,
aj , 117 , 764 elmegreen b.g . , kim s. & staveley - smith l. 2001 , apj , 548 , 749 elmegreen d.m . , kaufman m. , elmegreen b.g . ,
brinks e. , struck c. , klari m. & thomasson m. 2001 , aj , 121 , 182 elmegreen b.g .
2006 , in _ mass loss from stars and the evolution of stellar clusters _ ,
a. de koter , l. smith , r. waters , asp conf . ser .
13 , in press , astroph/0610679 elmegreen b.g .
, elmegreen d.m .
, chandar r. , whitmore b. & regan m. 2006 , apj , 644 , 879 elmegreen b.g . & elmegreen d.m .
2001 , apj , 121 , 1507 elmegreen b.g .
2008 , in _
globular clusters - guides to galaxies _ , eso astrophysics symposia .
isbn 978 - 3 - 540 - 76960 - 6 .
springer berlin heidelberg , 2009 , p. 87 elson r.a . &
1988 , aj , 96 , 1383 freeman k.c . , illingworth g. & oemler a. jr . 1983 , apj , 272 , 488 friel e.d .
1995 , ara&a , 33 , 381 de la fuente marcos r. & de la fuente marcos c. , 2009 , apj , 700 , 436 geisler d. , bica e. , dottori h. , clari j.j .
, piatti a.e . &
santos jr . j.f.c .
1997 , aj , 114 , 1920 gieles m. , bastian n. & ercolano b. 2008 , mnras , 391l , 93 girardi l. , chiosi c. , bertelli g. & bressan a. 1995 , a&a , 298 , 87 girardi l. , rubele s. & kerber , l. 2009 , mnras , 384 , 74 glatt k. , grebel e.k .
, sabbi e. , gallagher j.s .
, nota a. , sirianni m. , clementini g. , tosi m. 2008 , aj , 136 , 1703 goodwin s.p .
& bastian n. 2006 , mnras , 373 , 752 graff d.s . , gould a.p . , suntzeff n.b . , schommer r.a . &
hardy e. 2000 , apj , 540 , 211 de grijs r. & anders p. 2006 , mnras , 366 , 295 de grijs r. & goodwin s.p .
2008 , mnras , 383 , 1000 de grijs r. & goodwin s.p .
2009 , in _ iau symposium 256 _ , van loon j.t .
& oliveira j.m .
, cambridge univ .
press grocholski a.j . ,
cole a.a . ,
sarajedini a. , geisler d. & smith v.v .
2006 , aj , 132 , 1630 harris j. & zaritsky d. 1999 , aj , 117 , 2831 hodge p.w .
1960 , apj , 131 , 351 hodge p.w . 1982 , apj , 256 , 447 hodge p.w . & sexton , j.a . 1966 , aj , 71 , 363 hodge p.w . &
wright f.w .
1974 , aj , 79 , 858 hodge p.w .
1986 , pasp , 98 , 1113 hodge p.w .
1988 , pasp , 100 , 1051 hughes s.m.g . ,
wood p.r . &
reid n. 1991 , aj , 101 , 1304 hunter d.a .
, elmegreen b.g .
, dupuy t.j . & mortonson m. 2003 , aj , 126 , 1836 janes k.a . &
phelps r.l .
1994 , aj , 108 , 1773 karampelas a. , dapergolas a. , kontizas e. , livanou e. , kontizas m. , bellas - velidis i. & vlchez j.m .
2009 , a&a , 497 , 703 kerber l.o .
& santiago b.x .
2005 , a&a , 435 , 77 kerber l.o . ,
santiago b.x . &
brocato e. 2007 , a&a , 462 , 139 khalisi e. , amaro - seoane p. & spurzem r. 2007 , mnras , 374 , 703 king i. 1962 , aj , 67 , 471 kontizas m. , morgan d.h . , hatzidimitriou d. & kontizas e. 1990 , a&as , 84 , 527 kron g.e .
1956 , pasp , 68 , 125 lamers h.j.g.l.m . , gieles m. , bastian n. , baumgardt h. , kharchenko n.v . & portegies zwart s. 2005 , a&a , 441 , 117 leitherer c. , schaerer d. , goldader j.d .
, gonzlez - delgado r.m .
, robert c. , kune d.f .
, de mello d.f . , devost d. , heckman t.m . 1999 ,
apjs , 123 , 3 lindsay e.m .
1958 , mnras , 118 , 172 livanou e. , kontizas m. , gonidakis i. , kontizas e. , maragoudaki f. , oliver s. , efstathiou a. & klein u. 2006 , a&a , 451 , 431 livanou e. , gonidakis i. , kontizas e. , klein u. , kontizas m. , kester d. , fukui y. , mizuno n. & tsalmantza p. 2007
, aj , 133 , 2179 lyng g. & westerlund b.e .
1963 , mnras , 127 , 31 mackey a.d . &
gilmore g.f .
2003a , mnras , 338 , 85 mackey a.d . & gilmore g.f .
2003b , mnras , 338 , 120 mackey a.d . &
gilmore g.f .
2004 , mnras , 352 , 153 matteucci a. , ripepi v. , brocato e. & castellani v. 2002 , a&a , 387 , 861 milone a.p .
, bedin l.r . , piotto g. & anderson j. 2009 , a&a , 497 , 755 mould j.r . , jensen j.b .
& da costa g.s .
1992 , apjs , 82 , 489 mucciarelli a. , origlia l. , ferraro f.r .
, maraston c. & testa v. 2006 , apj , 646 , 939 oey m.s .
, parker j.s . ,
mikles v.j .
& zhang x. 2003 , aj , 126 , 2317 olsen k.a.g . & salyk c. 2002 , aj , 124 , 2045 olszewski e.w .
, harris h.c .
, schommer r.a . &
canterna r. 1988 , aj , 95 , 84 parmentier g. & de grijs r. 2008 , mnras , 383 , 1103 parisi m.c .
, grocholski a.j . , geisler d. , sarajedini a. & clari j.j .
2009 , aj , 138 , 517 piatti , a.e .
, santos jr .
, clari j.j . ,
bica e. , sarajedini a. & geisler d. 2001 , mnras , 325 , 792 piatti , a.e .
, sarajedini a. , geisler d. , seguel j. & 2005 , mnras , 358 , 1215 piatti a.e .
, sarajedini a. , geisler d. , clark d. & seguel j. 2007a , mnras , 377 , 300 piatti a.e .
, sarajedini a. , geisler d. , gallart c. & wischnjewsky m. 2007b , mnras , 381 , l84 piatti a.e .
, sarajedini a. , geisler d. , gallart c. & wischnjewsky m. 2007c , mnras , 382 , 1203 piatti a.e . ,
geisler d. , sarajedini a. & gallart c. 2009 , a&a , 501 , 585 pietrzynski g. , udalski a. , kubiak m. , szymanski m. , wozniak p. & zebrun k. 1998 , aca , 48 , 175 pietrzynski g. , udalski a. , kubiak m. , szymanski m. , wozniak p. & zebrun k. 1999 , aca , 49 , 521 pietrzynski g. & udalski a. 2000 , aca , 50 , 355 piskunov a.e .
, schilbach e. , kharchenko n.v .
, rser s. & scholz r .- d .
2007 , a&a , 468 , 151 rafelski m. & zaritsky d. 2005 , aj , 129 , 2701 rochau b. , gouliermis d.a .
, brandner w. , dolphin a.e . & henning , t. 2007 , apj , 664 , 322 sabbi e. , sirianni m. , nota a. , tosi m. , gallagher j. , meixner m. , oey m.s .
, walterbos r. , pasquali a. , smith l.j . & angeretti l. 2007 , aj , 133 , 44 santiago b.x .
, elson r.a.w . , sigurdsson s. & gilmore g.f .
1998 , mnras , 295 , 860 schaefer b.e .
2008 , aj , 135 , 112 schommer r.a .
, suntzeff n.b . , olszewski e.w .
& harris h.c .
1992 , aj , 103 , 447 shapley h. & lindsay e.m .
1963 , iraj , 6 , 74 stanimirovi s. , staveley - smith l. & jones p.a .
2004 , apj , 604 , 176 walsh a.j .
, bourke t.l . & myers p.c .
2006 , apj , 637 , 860 williams j.p . , blitz l. & stark a.a .
1995 , apj , 421 , 252 [ lastpage ]
the lmc and smc disks are inclined with respect to the line of sight , and this should introduce some variations in the absolute sizes across the disks , as compared to the no - inclination approach adopted in sect .
[ distribfunc ] . to investigate the inclination effect on the radius distribution functions we assume @xmath194 ( e.g. @xcite ) and @xmath195 ( e.g. @xcite ) as the inclination of the lmc and smc disks with respect to the line of sight .
then , the absolute size of each cluster was recomputed for its corrected distance . the inclination - corrected radius distribution function ( for the combined lmc@xmath98smc clusters )
is shown in fig .
[ fig11 ] , in which the un - corrected distribution ( fig .
[ fig5 ] , panel f ) is also shown for comparison purposes .
we conclude that differences are small , essentially within the error bars .
this can be accounted for by the relatively small distance corrections ( with respect to the adopted cloud distances ) , and that corrections affect objects both in the near and far sides in the opposite sense . on average ,
when a large number of objects is considered , the near and far - side corrections tend to self - compensate .
another effect that might introduce variations on absolute cluster size is the triaxial nature of the clouds .
the smc , for instance , may have a line - of - sight depth of 6 - 12kpc ( e.g. @xcite ) .
thus , depth corrections would be of the same order as those related to inclination .
since extended structures are the focus of this work , the surface brightness ( sb ) incompleteness - which is expected to affect the radius distribution functions - should be taken into account .
basically , for a given luminosity , a more extended object will have on average a lower sb , and thus may not be detected by depth - limited surveys .
we examine this effect by means of a sample of @xmath196 artificial star clusters whose luminosity and radius distributions are described by @xmath197 and @xmath198 , respectively . as discussed in @xcite ,
these analytical functions describe star clusters and hii regions . to reproduce the input radius and luminosity distributions ,
the simulated radius and luminosity are computed from @xmath199^{1/2.3}$ ] and @xmath200 $ ] , where @xmath201 , @xmath202 , @xmath203 , and @xmath204 are the minimum and maximum radii and luminosities , and @xmath205 and @xmath206 are random numbers in the range @xmath207 $ ] .
the radius and luminosity of a given cluster are independently assigned .
this process allows that clusters of the same size ( and mass ) - but different ages - may have different luminosities , as expected from the fading lines associated with the stellar evolution .
also , clusters with any radius within ( @xmath208 ) are allowed to have any luminosity within ( @xmath209 ) .
then , the sb is computed in the usual way , @xmath210 , in arbitrary units .
the results are summarised in fig .
[ fig12 ] . in general
, the sb distribution among the simulated clusters agrees with the expected relation of decreasing sb with cluster radius ( panel a ) .
the two power - laws that describe the radius and luminosity distributions are reflected on the shape of the sb distribution ( panel b ) , which first ( beginning at the smallest and most luminous clusters ) increases exponentially towards lower sbs , reaches a maximum , and falls off exponentially towards the largest and less luminous clusters .
based on this distribution we arbitrarily apply cuts for clusters with @xmath211 and @xmath212 and compute the corresponding radius distributions ( panel c ) .
clearly , the sb cuts preserve the power - law character of the radius distribution .
the major effect is a steepening of the slope . indeed , while the complete radius distribution is a power - law of slope @xmath213 , the sb - restricted distributions have @xmath214 , and @xmath215 , respectively for @xmath211 and @xmath212 . in summary ,
sb - related incompleteness affects the radius distributions preferentially at the large - clusters tail , having little effect on the small clusters . also , it preserves the power - law character of the radius distribution and , due to the preferential effect on large clusters , it produces a steepening of the distributions .
thus , the decrease in the observed radius distributions towards small clusters ( figs .
[ fig3 ] - [ fig5 ] ) can not be accounted for by sb incompleteness , and appears to be linked to an observational effect . in this sense
, vista will be important also to explore the structure of small clusters in the clouds , and to investigate the shape of the radius distribution at the small scales . | we investigate the degree of spatial correlation among extended structures in the lmc and smc . to this purpose
we work with sub - samples characterised by different properties such as age and size , taken from the updated catalogue of bica et al . or gathered in the present work .
the structures are classified as star clusters or non - clusters ( basically , nebular complexes and their stellar associations ) .
the radius distribution functions follow power - laws ( @xmath0 ) with slopes and maximum radius ( @xmath1 ) that depend on object class ( and age ) .
non - clusters are characterised by @xmath2 and @xmath3pc , while young clusters ( age @xmath4myr ) have @xmath5 and @xmath6pc , and old ones ( age @xmath7myr ) have @xmath8 and @xmath9pc .
young clusters present a high degree of spatial self - correlation and , especially , correlate with star - forming structures , which does not occur with the old ones .
this is consistent with the old clusters having been heavily mixed up , since their ages correspond to several lmc and smc crossing times . on the other hand , with ages corresponding to fractions of the respective crossing times
, the young clusters still trace most of their birthplace structural pattern .
also , small clusters ( @xmath10pc ) , as well as small non - clusters ( @xmath11pc ) , are spatially self - correlated , while their large counterparts of both classes are not .
the above results are consistent with a hierarchical star - formation scenario for the lmc and smc .
[ firstpage ] ( _ galaxies _ : ) magellanic clouds |
the dynamical coupling between the baryons and the cosmic microwave background ( cmb ) during the epoch of cosmological recombination ( @xmath5 ) is commonly assumed to be mediated only through thomson scattering ( peebles & yu 1970 ) .
this coupling becomes ineffective when hydrogen recombines , at around the same time when the universe becomes transparent , and hence it leaves a clear signature on the temperature anisotropies of the cmb ( white , scott , & silk 1994 , and references therein ) .
forthcoming satellite missions , such as map in 2001 or planck in 2007 , offer the sensitivity to probe deviations from this simple model at the percent level ( scott 1999 ; yu , spergel , & ostriker 2001 ) . in this _ letter _ , i examine the potential significance of the coupling between the cosmic gas and the cmb due to resonant line transitions of neutral atoms . for such transitions , the integral of the absorption cross - section over frequency , @xmath6 is typically many orders of magnitude larger than for thomson scattering , where @xmath7 is the absorption oscillator strength ( rybicki & lightman 1979 ) .
for example , the transition of hydrogen provides an average cross - section which is seven orders of magnitude larger than the thomson value , when averaged over a frequency band as wide as the resonant frequency itself [ @xmath8 .
since the abundance of atoms in excited states is highly suppressed by large boltzmann factors at the recombination redshift , we consider only transitions from the ground state . for helium , these transitions correspond to photon frequencies that are too far on the wien tail of the cmb spectrum and only couple to a negligible fraction of the radiation energy density .
hence , we focus our discussion on hydrogen and lithium . in 2 , we derive the resonant line opacity and drag force for atoms that move relative to a blackbody radiation field . in 3 , we consider the cosmological implications of these results for hydrogen and lithium . finally ,
4 summarizes the main conclusions of this work . throughout the paper ,
we adopt the density parameters @xmath9 for matter and @xmath10 for baryons , and a hubble constant @xmath11 .
the effect of the cosmological constant can be neglected at the redshifts of interest here .
consider a single atom moving at a non - relativistic peculiar velocity @xmath12 relative to an isotropic blackbody radiation field of temperature @xmath13 .
because of its motion , the atom sees a dipole anisotropy in the radiation temperature , @xmath14 $ ] , where @xmath15 is the angle relative to its direction of motion ( peebles 1993 ) .
we assume that the atom couples to the radiation through a resonant line of frequency @xmath16 with @xmath17 .
hence , the line resonates in the wien tail of the spectrum , for which the radiation energy density per unit frequency ( in erg @xmath18 hz@xmath19 ) is given by , @xmath20 where we have used the fact that @xmath21\ll 1 $ ]
. absorption of a resonant photon by the atom is followed by isotropic emission in the atom s rest frame , and so the net drag force ( momentum transfer per unit time ) is given by , @xmath22 since the absorption cross - section is sharply peaked around @xmath16 , we may use equation ( [ eq : int_s ] ) to get , @xmath23 for an ensemble of atoms of density @xmath24 , which are all moving at the same peculiar velocity @xmath25 , the force per unit volume is given by @xmath26 .
if the resonant atoms are embedded within a hydrogen plasma of total ( free@xmath27bound ) proton density @xmath28 and are strongly coupled to the plasma through binary particle collisions , then the cosmological equation of motion for the gas as a whole reads , @xmath29 or equivalently , @xmath30 where @xmath31 is the hubble expansion rate , and @xmath32 is the characteristic time over which the peculiar velocity of the gas is damped due to the drag force on the resonant atoms , @xmath33 here , @xmath34 is the resonant wavelength , @xmath35 is the proton mass , and @xmath36 , where @xmath37 is the atomic weight of the resonant atoms . in the cosmological context , @xmath38 is the cmb temperature ( mather et al .
both @xmath13 and @xmath39 are functions of cosmic time , and so equation ( [ eq : acc ] ) admits the solution , @xmath40 \exp\left\{-\int { dt /t_{\rm a}(t)}\right\}$ ] , where @xmath41 and @xmath42 is the cosmological scale factor . in general , the right - hand - side of equation ( [ eq : acc ] ) should include other terms which source the peculiar velocity field , such as gravitational or pressure forces due to density inhomogeneities .
the significance of the drag force can be calibrated in terms of the hubble time , @xmath43 , through the product @xmath44 .
for the redshifts of interest , @xmath45 .
the assumption of an isotropic radiation background is only valid if the moving fluid element is _ optically - thin _ to resonant scattering . in the opposite regime of high
opacity , the background photons are isotropized locally through multiple scatterings and exert a reduced drag force on an atom embedded deep inside the moving fluid element . the sobolev optical depth for resonant scatterings in an expanding and nearly uniform universe is given by ( sobolev 1946 ; dellantonio & rybicki 1993 ) , @xmath46 where @xmath47 is the transition probability per unit time for spontaneous emission between the two energy levels ( in s@xmath19 ) , and @xmath48 is the ratio between the statistical weights of the excited and ground states .
the optical depth is dominated by the velocity gradient of the hubble flow , because gravitationally
induced peculiar velocities perturb this flow only slightly at the early cosmic times of interest here . for @xmath49 , the flux of resonant photons inside the moving fluid element is reduced by a factor of @xmath50 ( rybicki & hummer 1978 ) , and so the drag force in equation ( [ eq : int_f ] ) is lowered by the same factor .
this reduction has dramatic consequences for the resonant transition of neutral hydrogen , as we show next .
for the @xmath51@xmath52 transition of hydrogen , @xmath53 , @xmath54 , @xmath55 , and @xmath56 , yielding an exceedingly large sobolev optical depth , @xmath57^{3/2}$ ] , where @xmath58 denotes the neutral fraction of hydrogen .
the thermally broadened line has an effective cross section of @xmath59 ( peebles 1993 ) and a corresponding mean - free - path of @xmath60 , at @xmath61 . only on smaller scales
can viscosity damp velocity gradients effectively . on the much larger scales where the cmb anisotropies are measurable
, the drag time in equation ( [ eq : t_a ] ) is lengthened by a factor of @xmath62 , which makes it very much longer than the hubble time .
the situation is similar for deuterium .
balmer line transitions of hydrogen have a small optical depth , but the occupation probability of their lower level is suppressed by an exceedingly large boltzmann factor , @xmath63 . the resulting drag force is small , even when including the enhancement in the flux beyond the blackbody value , due to the photons released during recombination ( @xmath64 photon per hydrogen atom ; see dellantonio & rybicki 1993 for the distorted spectrum ) .
lithium has a closed @xmath65 shell with two electrons , and one electron outside this shell .
the transition between the ground state ( @xmath66 ) and the first excited state ( @xmath52 ) has @xmath67 and @xmath68 , with @xmath69 and @xmath70 for the @xmath71@xmath72 transition , and @xmath73 , @xmath74 for the @xmath71@xmath75 transition ( radzig & smirnov 1985 ; yan & drake 1995 ) .
we therefore get from equations ( [ eq : t_a ] ) and ( [ eq : tau_a ] ) , @xmath76 and @xmath77 figure 1 depicts the redshift dependence of the drag time in equation ( [ eq : t_li ] ) and compares it to the drag time due to thomson scattering in the recombining plasma , @xmath78 where @xmath79 is the ionization fraction of hydrogen .
we calculated @xmath80 from the updated version of the standard recombination model ( seager , sasselov , & scott 2000 ) . despite the low abundance of lithium ,
its resonant drag force is not much lower than the thomson drag force at @xmath0 , when roughly half of the lithium atoms are expected to have recombined ( palla , galli , & silk 1995 ; stancil , lepp , & dalgarno 1996 , 1998 ) .
the net effect of the lithium drag is highly sensitive to the recombination history of lithium and hydrogen .
previous discussions ( e.g. , stancil et al . 1996 , 1998 ) did not include the delaying effect of photons on the recombination history of hydrogen ( peebles 1993 ) , and therefore underestimated the free electron fraction and correspondingly the neutral lithium abundance at @xmath81 .
another process with an opposite sign that was omitted , involves the excitation of neutral lithium from its ground level by the cmb radiation field . for this effect , it is important to take account of the lyman and balmer series distortions of the cmb spectrum ( dellantonio & rybicki 1993 ) , which redshift into resonance at lithium recombination .
in fact , the lithium chemistry in the early universe might provide a way for inferring the existence of these distortions .
finally , the recombination history of @xmath82be should also be followed since a substantial fraction of the @xmath82li forms through electron - capture in the @xmath82be produced by big - bang nucleosynthesis ( t. walker , private communication ) .
since the half - life of neutral @xmath82be is only 53 days , this conversion can be regarded as instanteneous on the cosmological timescale , as soon as @xmath82be recombines .
careful calculations of the neutral lithium fraction as a function of redshift are required in order to quantify the imprint of the lithium drag on the cmb anisotropies ( dalgarno , loeb , & stancil 2001 ) .
the lithium chemistry could also be affected by a large drift velocity between the li i fluid and the hydrogen plasma . to examine this effect , we consider the equation of motion for the li i fluid ( burgers 1969 ) , @xmath83 here @xmath84 is the collision rate of a li atom with the gas , where @xmath85 , and @xmath86 $ ] .
the index @xmath87 runs over all other particle species in the gas which are assumed to move at a common bulk velocity @xmath25 because they are not subject to the resonant drag force .
the geometric cross - section of li i atoms , @xmath88 , couples them to the rest of the gas on a time scale @xmath89 , much shorter than the hubble time , @xmath90 at @xmath61 .
hence , we may ignore the terms on the left - hand - side in equation ( [ eq : li_eom ] ) , as well as the perturbed gravitational force which is of the same order .
the remaining terms give the solution , @xmath91 where @xmath92 .
substituting @xmath93 , we find that @xmath94^{3.5}~{\rm s^{-1}}$ ] , and that @xmath95 at @xmath96 , after lithium recombination . at higher redshifts , the assumption that @xmath97 seems not to apply and one may wonder whether the drag time in equation ( [ eq : acc ] )
should be multiplied by a factor of @xmath98 , since the drag force acts on @xmath99 and not on @xmath25 . in this context
it is crucial to realize that the li i nuclei do not maintain their identity as neutral atoms ; in fact , they continuously form and get ionized on a timescale much shorter than a hubble time .
the cross - section for coulomb collisions of li@xmath100 is far greater than that of li i , @xmath101^{-2}~{\rm cm^2}$ ] , where @xmath102 is the coulomb logarithm .
the fraction of time spent by a li nucleus in an ionized state is the ionization fraction , @xmath103 , and so @xmath104 should be replaced by @xmath105 $ ] and the drag force term should be multiplied by @xmath106 in equation ( [ eq : li_eom ] ) .
we consequently find that the substantial ionization fraction of the lithium fluid ( stancil et al .
1996 , 1998 ) diminishes its drift velocity relative to the gas at all relevant redshifts , @xmath107 .
hence , non - thermal relative velocities may be ignored in the li i chemistry . at @xmath81 ,
the mean - free - path for resonant scattering by lithium exceeds the relevant spatial scales for the cmb anisotropies ( @xmath108@xmath109 ) , even if the lithium abundance is taken to be somewhat higher than the value , @xmath110 , predicted by the latest deuterium measurements and standard big - bang nucleosynthesis ( burles , nollett , & turner 2001 ; see also walker et al .
1991 ; smith , kawano , & malaney 1993 ) .
note that values as high as @xmath111 were suggested by models of inhomogeneous big - bang nucleosynthesis ( applegate & hogan 1985 ; sale & mathews 1986 ; mathews et al .
1990 ) , and would result in @xmath112 for velocity gradients on sufficiently small scales with @xmath113 . if a substantial fraction of lithium recombines by @xmath0 ( as predicted by palla et al . 1995 or
stancil et al . 1996 , 1998 ) , then the flux of the original anisotropies will be suppressed by the absorption factor , @xmath114 , at observed wavelengths below @xmath115 @xmath116 m , for which @xmath117 .
the resonant scattering effect is different from that caused by thomson scattering at the epoch of reionization ( hu & white 1997 ) , in that it has a larger optical depth and it occurs within a much thinner shell of gas ( for a sufficiently narrow band of observed wavelengths ) .
consequently , it should induce new first - order anisotropies [ multiplied by @xmath118 due to the coherence of the velocity field in the thin scattering shell .
the dominant contribution to the new anisotropies would come from the doppler effect .
in contrast to reionization , there should be no severe cancellations of the doppler effects from line - of - sight velocity fluctuations at large wavenumbers , because the last scattering surface ( or `` visibility function '' ) for the resonance is extremely thin .
the increase in amplitude of sub - horizon fluctuations between @xmath119 and @xmath120 would enhance the new anisotropies relative to the original ones and introduce a distortion of order unity in the original anisotropy spectrum on these scales .
the anisotropy spectrum would be modified at multipole moments @xmath3 , around and below the angular scale of the first acoustic peak ( which is a factor of @xmath121 smaller than the horizon scale at @xmath122 ) , where the original anisotropy amplitude reaches its maximum ( @xmath123 ) .
resonant scattering would also result in enhanced polarization anisotropies on the same angular scales ( chandrasekhar 1960 ) . since the scattering is done by lithium
, the anisotropies will reflect the peculiar velocity of the lithium fluid which , as shown above , is not expected to deviate significantly from the velocity of the gas as a whole .
since the temperature fluctuations are in the wien tail of the cmb spectrum , they translate to brightness fluctuations ( in @xmath124 ) of a much larger contrast , @xmath125 where @xmath126 is the redshift being probed and we have substituted @xmath127 . hence the first acoustic peak would correspond to brightness fluctuations of up to @xmath128 . at a given observed wavelength , the redshift thickness of the scattering shell
is only determined by two contributions : ( i ) the possibility that a photon at this wavelength was either absorbed by the @xmath71@xmath75 transition with @xmath129 or at a slightly lower redshift by the @xmath71@xmath72 transition with @xmath130 ( radzig & smirnov 1985 ) , yielding @xmath131 = 2.2\times 10^{-5}$ ] ; and ( ii ) the thermal width of the line , the photon does not need to redshift away from the line center by more than the line width in order to escape to infinity . ] , @xmath131 = ( \delta \lambda/\lambda)_{\rm th } \approx 10^{-5}$ ] . in practice
, the minimum thickness would be limited by the band width of the detector .
the maximum band width and minimum angular resolution should be tuned to match the expected coherence length of the peculiar velocity field at @xmath0 , which corresponds to @xmath132 and @xmath133 or equivalently @xmath134 .
observations at different far infrared ( fir ) wavelengths would probe different thin slices of structure in the early universe . the anisotropy pattern would vary gradually as a function of wavelength and `` 3d tomography '' is in principle possible , by which one may probe correlations in the structure along the line - of - sight .
the feasibility of this measurement might be severely compromised by the anisotropies of the fir foreground .
unfortunately , @xmath135 m is just the wavelength where the cmb intensity blends into the extragalactic fir foreground ( see fig . 1 in scott et al .
however , the contribution of discrete sources ( see fig . 6 in knox
2001 ) could be separated out through observations at shorter wavelengths with a higher angular resolution .
furthermore , continuum foreground emission would result in similar anisotropies for all wavelengths within a band width @xmath136 , while the cmb anistropies would vary considerably across such a band since its width corresponds to @xmath137 .
subtraction of anisotropy maps at slightly different wavelengths within such a band can therefore be used to isolate out the cmb component ( as long as these wavelengths do not overlap with strong emission lines from the foreground sources ) . for @xmath138
, the anisotropies should have the same power - spectrum as measured by map or planck at much longer wavelengths .
the predictable correlations between the temperature and polarization patterns of the cmb and their general statistical properties could be used to further enhance the signal - to - noise ratio ( zaldarriaga & loeb 2001 ) .
obviously , these difficult observations need to be made through holes in the galactic dust emission at high galactic latitudes .
the 6708 resonant transition of neutral lithium provides a previously unexplored coupling between the baryons and the cmb after cosmological recombination .
the lithium opacity is substantial at observed wavelengths @xmath139\mu$]m , where @xmath140 is the redshift at which @xmath141 of the lithium recombines . at these wavelengths , the cmb anisotropies on sub - degree angular scales
would be significantly altered , due to the finite optical depth for resonant scattering , @xmath142^{3/2}$ ] .
this scattering would generate new temperature and polarization anisotropies at mutipole moments @xmath3 .
the detection of these anisotropies depends on the prospects for cleaning the expected anisotropies of the fir foreground ( knox et al .
2001 ) , and may be feasible through the subtraction of anisotropy maps for pairs of slightly different wavelengths . the relevant wavelength range overlaps with the highest frequency channel of the planck mission ( @xmath143 m ) and with the proposed baloon borne explorer of diffuse galactic emissions ( edge ) which will survey 1% of the sky in 10 wavelength bands between @xmath144@xmath145 m with a resolution ranging from @xmath146 to @xmath147 ( see table 1 in knox et al .
2001 ) . however , a new instrument design with multiple narrow bands ( @xmath148 ) at various wavelengths in the range @xmath149@xmath135 m and with angular resolution @xmath150 , is necessary in order to optimize the detection of the lithium signature on the cmb anisotropies .
the author thanks alex dalgarno , daniel eisenstein , george rybicki , and matias zaldarriaga for useful discussions .
this work was supported in part by nasa grants nag 5 - 7039 , 5 - 7768 , and by nsf grants ast-9900877 , ast-0071019 . | i show that neutral lithium couples strongly to the cosmic microwave background ( cmb ) through its 6708 resonant transition after it recombines at @xmath0 . at observed wavelengths of @xmath1 m ,
the cmb anisotropies are significantly altered since the optical depth for resonant scattering by neutral lithium is substantial , @xmath2 .
the scattering would suppress the original anisotropies , but will generate strong new anisotropies in the cmb temperature and polarization on sub - degree scales ( @xmath3 )
. observations at different wavelengths in this spectral regime can probe different thin slices of the early universe .
the anisotropy noise contributed by continuum radiation from foreground far infrared sources could be taken out by subtracting maps at slightly different wavelengths .
detection of the above effects can be used to study structure at @xmath4 and to constrain the primordial abundance and recombination history of lithium . |
the excellent quality of the data generated by _ chandra _ for bright snrs in our galaxy is showcased by the false color images that have become perhaps the mission s most recognizable and widespread visual result ( figure 1 ) . for a snr with an angular diameter of @xmath3 like cas
a , _ chandra _ can resolve more than @xmath4 individual regions .
deep exposures of bright objects usually provide enough photon statistics to obtain a good spectrum from most of these regions at the moderate spectral resolution of the acis ccd detectors ( @xmath5 ) .
this usually allows to detect k - shell emission from abundant elements like o , si , s , ar , ca , and fe . the density of the plasma inside snrs is low enough for the ages of young objects like cas a or tycho to be smaller than the ionization equilibrium timescale .
the x - ray emitting plasma heated by the shocks is therefore in a state of nonequilibrium ionization , or nei @xcite .
this means that the ionization state of any given fluid element is determined not only by its electron temperature @xmath6 , but also by its ionization timescale @xmath7 , where @xmath8 is the electron number density and @xmath9 is the time since shock passage .
the thermal x - ray spectrum from a young snr is thus intimately related to its dynamic evolution through the individual densities and shock passage times of each fluid element .
this has important implications for the ejecta emission , because of the large differences in chemical composition across the sn debris , and the fact that the electron pool is completely dominated by the contributions from high - z elements , making @xmath8 a strong function of the ionization state @xcite . under these circumstances ,
the quantitative analysis of x - ray spectra from young snrs becomes a challenging endeavor . in order to derive magnitudes that are relevant to sn physics , like the kinetic energy @xmath10 or the ejected mass of each chemical element
, it is necessary to build a hydrodynamic model of the entire snr .
one must know when each fluid element was shocked , what its chemical composition is , how much of the sn ejecta is still unshocked at the present time , etc .
individual fitting of each magnitude becomes impractical , because they are all related to each other , and in order to understand the x - ray spectrum of a particular snr , one has to understand of the object as a whole .
further complications stem from two sources .
one is a technical , but important , problem : the uneven quality of atomic data in x - ray emission codes for nei plasmas @xcite .
the other is of a more fundamental nature : the large uncertainties in the physics of collisionless shocks , in particular the amount of ion - electron temperature equilibration at the shock transition @xcite and the impact of cosmic ray acceleration on the dynamics of the plasma @xcite .
most of the analysis of the x - ray emission from snrs is done following one of two radically different approaches .
the first approach is to forgo the complex interaction between snr dynamics and x - ray spectrum , and fit the emission of the entire snr or individual regions using ready - made spectral models .
the most widely used are plane - parallel shock nei models @xcite , which fit the spectra by varying @xmath6 , some parameter related to @xmath7 , and a set of chemical abundances .
this has the advantages of simplicity and flexibility , but the number of free parameters is large , the quality of the fits is often poor , and the results are hard to interpret in the framework of sn physics and progenitor scenarios .
the other approach is to model the full hd evolution of the snr _ ab initio _ , starting from a grid of sn explosion models and ism or csm configurations , calculate the nei processes in the shocked plasma , and produce a set of synthetic x - ray spectra that can then be compared to the observations .
this approach is less flexible , and usually does not allow for spectral fits in the usual sense ( i.e. , based on the @xmath11 statistic ) , but it is considerably more powerful in that the observations and the physical scenarios that are being tested can be compared in a more direct way .
the first hd+nei models for snrs were built to analyze the data from early x - ray missions like _ einstein _ and _ exosat _ ( e.g , @xcite ) .
modern efforts produced for _ chandra _ and _ xmm - newton _ data can be found in @xcite and @xcite .
type ia sne are believed to be the thermonuclear explosion of a c+o wd that is destabilized when its mass becomes close to the chandrasekhar limit by accretion of material from a binary companion .
after the central regions ignite , the burning front propagates outwards , consuming the entire star and leading to a characteristic ejecta structure , with @xmath12 of fe - peak nuclei ( mostly the @xmath13ni that powers the light curve ) in the inside and about an equal amount of intermediate mass elements ( mostly si , s , ar , and ca ) in the outside .
the amount of @xmath13ni can be as high as @xmath14 in the brightest type ia sne , and as low as @xmath15 in the dimmest type ia sne , but the large scale stratification seems to be retained in all cases @xcite .
state - of - the - art three - dimensional explosion models still can not reproduce these basic features in a self - consistent way , mostly because hydrodynamic instabilities tend to destroy the stratification of nucleosynthetic products on very short timescales @xcite .
this relatively simple structure of type ia sn ejecta is responsible for the uniformity of the light curves and spectra that makes them useful as distance indicators for cosmology .
the still poorly understood details of the multi - dimensional explosion mechanism might provide the diversity within the uniformity , leading to bright and dim events @xcite .
the fact that bright ia sne appear to explode preferentially in star - forming galaxies @xcite suggests some connection between the properties of the progenitors and the explosion physics , but since the identity of the progenitors has never been clearly established , this connection remains mysterious @xcite .
depending on the nature of the wd companion , sn ia progenitors are divided into single degenerate ( the companion is a normal star , @xcite ) and double degenerate ( the companion is another wd @xcite ) . finding clear evidence supporting one of these two scenarios has become a central problem in stellar astrophysics , because unknown evolutionary effects associated with the progenitors may introduce systematic trends that could limit the precision of cosmological measurements based on sn ia @xcite .
most ( but not all ) type ia snrs reflect the relative uniformity and simplicity of their birth events .
even a cursory glance at the _ chandra _ images of tycho ( figure 1 , @xcite ) or sn1006 @xcite reveals strikingly symmetrical objects , without large anisotropies in the ejecta emission , in marked contrast to the turbulent nature of cc snrs .
the morphology and x - ray spectra of snrs with known ages and a well - established type ia classification also indicate that the progenitors do not modify their surroundings in a strong way - in particular , there is no evidence for large wind - blown cavities around the explosion sites @xcite .
this is relevant because the current paradigm for type ia progenitors in the single degenerate channel predicts fast , optically thick outflows from the wd surface @xcite that would leave behind such cavities .
we have to conlcude that either most type ia sne do not have single degenerate progenitors , or these fast outflows are not always present .
in fact , the dynamical and spectral properties of young type ia snrs like tycho or [email protected] are consistent with an interaction with the warm phase of the ism @xcite , and so is the detailed morphology of the blast wave in objects where it can be studied with sufficient resolution , like sn1006 @xcite a notable exception to this is the kepler snr .
although the recent deep ( 750 ks ) _ chandra _ exposure of kepler left little doubt that the snr is of type ia @xcite , its morphology shows a strong bilateral asymmetry ( figure 1 ) , and optical observations reveal clear signs of an interaction with a dense , n - enriched csm @xcite .
this suggests that either the progenitor of kepler s sn or its binary companion might have been relatively massive , creating a bow - shock shaped csm structure as the system lost mass and moved against the surrounding ism @xcite .
this complex csm interaction makes hd+nei modeling of the kepler snr challenging , and the estimation of the fundamental sn parameters difficult .
most of the detailed models for kepler were published when the snr was widely believed to be of cc origin @xcite .
it is important that these models be revised in light of the secure ia origin of the snr to determine the mass of @xmath13ni synthesized and the pre - explosion mass - loss rate from the progenitor .
obtained by matching the angular sizes of the reverse shock ( rs , dashed plots ) and contact discontinuity ( cd , solid plots ) to a bright ia sn model ( ddta , blue plots ) , and a dim ia sn model ( ddtg , red plots ) .
the values of @xmath17 in the hd+nei models of @xcite that provide the best match to the x - ray emission from the sn ejecta are indicated by the striped vertical band .
the value of @xmath17 required by the cr - modified shock models of @xcite is indicated by a vertical dash - dotted line .
the estimated value of the distance @xmath18 from @xcite is indicated by the horizontal solid line , with the dotted horizontal lines marking the upper and lower limits . ]
the hd+nei modeling approach has had more success in objects with simpler dynamics , like the tycho snr .
one - dimensional hd+nei models with a uniform ambient medium can reproduce the fundamental features of both the dynamics and the integrated x - ray spectrum of tycho , provided that the ejecta is stratified , with an fe content similar to what would be expected from a sn ia of normal brightness , and the ambient medium density is @xmath19 ( @xcite , figures 2 and 3 ) .
this density estimate merits a few comments .
although it is clear that cosmic ray acceleration has a strong impact on the dynamics of the forward shock @xcite , the fact remains that the microphysics of the process is not well understood , and cosmic ray backreaction is usually not included in full hd+nei models for the ejecta emission in snrs ( see @xcite ) .
instead , the assumption is made that the dynamics of the shocked ejecta are not severely affected by cosmic ray acceleration at the blast wave , which seems reasonable in light of the properties of the reverse shock ( see discussion in section 3 of @xcite ) . in any case
, it is important to emphasize that the value of @xmath17 required by the ejecta emission in the tycho snr also reproduces the angular sizes of the reverse shock and contact discontinuity at the correct distance to the snr ( figure 2 ) .
this basic validation of the underlying dynamic model is essential to the hd+nei approach , and strengthens the confidence on the conclusions drawn about the sn explosion . of @xmath13ni ,
interacting with an am of @xmath20 .
the contribution of the nonthermal emission from the blast wave is indicated by the dotted plot , the rest is thermal emission from the shocked ejecta ( figure from @xcite ) . ]
the stratification of the ejecta in the tycho snr is required by the fact that the @xmath7 of fe in the x - ray spectrum is lower than that of si , which can be explained naturally if most of the si is shocked earlier and at a higher density than most of the fe ( see also @xcite ) .
this argument is completely independent from the constraints on the ejecta structure obtained with optical spectroscopy of type ia sne , and constitutes a clear sign that a supersonic burning front ( i.e. , a detonation ) must have been involved in the physics of the explosion at some stage .
the most succesful sn model found by @xcite is a one - dimensional delayed detonation with @xmath21 erg that synthesizes @xmath22 of @xmath13ni ( figure 3 ) - in other words , a type ia sn of normal brightness , consistent with the historical records @xcite . 67.5 :
the le is well matched only by spectra of bright type ia sne like sn 1998es , sn 1999aa , and sn 1999dq ( figure from @xcite ) . ]
one largely unexpected , but very important , development in this field has been the possibility to calibrate the results obtained from the analysis of the ejecta emission in snrs using the light echoes from ancient sne .
these light echoes were originally discovered in the lmc @xcite , and they are important for two reasons . first , they can provide a reliable and independent age estimate for objects that previously had none , which is a crucial ingredient to build hd models ( see e.g. @xcite ) .
second , and most important , the spectroscopy of the light echoes can be used to determine the type ( cc vs. ia ) and subtype ( e.g. , bright ia or dim ia ) of ancient sne .
this technique was demonstrated by @xcite , who showed that snr [email protected] in the lmc was originated by a bright , @xmath13ni - rich type ia sn . at the same time
, @xcite revisited the archival observations of this snr and found that an hd+nei model based on a bright sn ia explosion reproduces both the x - ray emission from the shocked ejecta and the dynamics of the snr ( see figures 4 and 5 ) .
this agreement between two completely independent techniques applied to the same object has now been extended to historical sne in our own galaxy , with the discovery of light echoes from tycho and cas a @xcite .
spectroscopic analysis of the light echo from tycho @xcite confirmed that sn1572 was indeed a type ia sn of normal brightness , a spectacular validation of the previous result obtained with hd+nei models of the snr @xcite . 67.5 : the x - ray spectrum of the snr can only be reproduced with a bright type ia sn models that synthesizes @xmath23 of @xmath13ni ( figure adapted from @xcite ) . ] the emergence of the x - ray observations of snrs as well - established probes for type ia sne has important implications for sn research .
because snrs are nearby objects , the context of the exploded star can be studied in great detail , and relationships between specific explosion properties and progenitor scenarios can be tested _ in situ_. one powerful constraint on the single degenerate scenario for sn ia progenitors , for example , is the presence of the donor star , which should survive the explosion @xcite .
so far , the search for this star in the tycho snr has produced inconclusive results @xcite , but further work in this and other snrs should eventually clarify the identity of the progenitor of at least one sn ia explosion with known properties .
another recently opened line of research is the possibility to measure the metallicity of sn ia progenitors _ directly _ , using mn and cr lines in the x - ray spectrum of their snrs .
this technique yields a supersolar metallicity for the progenitor of tycho @xcite , a measurement that will soon be extended to other objects like kepler .
even a small census of metallicities obtained in this way will allow us to validate the correlations between the metallicity of the host galaxy and the sn ia brightness found in sn surveys @xcite . finally , the snr population in the magellanic clouds has great potential to constrain type ia sn progenitor models , because the resolved stellar population in these nearby galaxies can be studied in great detail .
snr [email protected] , for instance , which was originated by a bright type ia sn , is embedded in a relatively old , metal - poor stellar population , while snr n103b , which has signs of a csm interaction similar to kepler @xcite , is associated with a much younger , metal - rich population @xcite .
this suggests that the progenitor of snr n103b might have been relatively young and massive , and might have modified its surroundings through some kind of mass - loss process .
cc sne mark the final events in the lives of massive ( @xmath24 ) stars . after nuclear fuel
is exhausted in the inner regions , the stellar core collapses to form a neutron star or a black hole .
the layers of the star that do not end up in the central compact object bounce and start propagating outwards , but at present it is unclear exactly how the gravitational collapse of the core becomes an explosion @xcite . at the writing of this review , the most sophisticated multi - d simulations of cc sne still fail to explode . although the identity of the progenitors of cc sne is well - established , with several identifications in pre - explosion images @xcite , the details of which specific kinds of massive stars lead to which specific subtypes of cc sne ( e.g. type ib / c , type iil , etc . )
are still under debate .
most of these uncertainties stem from our imperfect knowledge of key processes in stellar evolution , like mass - loss mechanisms or the role played by binarity .
just like type ia snrs , the basic morphology of cc snrs reflects the fundamental characteristics of their birth events .
_ images of young cc snrs like cas a @xcite or snr [email protected] @xcite reveal a complex and turbulent structure , with marked asymmetries related both to the sn ejecta and the csm .
in contrast to type ia snrs , many cc snrs show prominent optical emission from radiatively cooled ejecta , the hallmark of an interaction with a relatively dense medium .
the poster child for young , ejecta - dominated cc snrs is cas a , an object that has played a central role in the history of _ chandra_. the first scientific result from the mission was the discovery of the central compact object @xcite , now thought to be a neutron star with a c atmosphere @xcite . the first refereed publication based on _ chandra _ data was also devoted to cas a @xcite .
this paper discusses the overturn of nucleosynthetic products in a large region in the se of the snr , where fe - rich material can be seen clearly ahead of si - rich material , indicating that some of the layers deep into the onion - skin structure of the exploding star have overtaken regions dominated by lower - z elements ( figure 1 , however , see @xcite for an alternative interpretation ) . over the years , more and more has been learned about the cas a sn and its progenitor . in 2003
, @xcite used the positions of the fluid discontinuities , the presence of high - velocity h , and the extent of the clumpy photoionized pre - sn wind to infer that the cas a explosion must have been of type iin or iib sn .
this ` prediction of the past ' , similar to the one made for the tycho snr , was also confirmed by the light echo , which has a spectrum very similar to that of the type iib sn1993j @xcite .
the morphology of the ejecta emission from cas a is very rich ( @xcite , figures 1 and 6 ) , but perhaps the most striking feature in the x - ray images is the jet / counterjet structure that runs from the sw to the ne @xcite . in order to understand the energetics of this feature and the role that it played in the explosion
, @xcite used hd+nei simulations to interpret the fits to small x - ray bright knots in the sn ejecta ( see figure 7 ) .
while it is clear that there was a significant local deposition of energy around the jet region @xcite , this was probably not enought to drive the entire explosion @xcite ( see also @xcite ) .
it turns out that the proper motion of the compact object is not aligned with the jet axis , as one would expect in a jet - driven explosion , but perpendicular to it , along a mysterious gap in the high - velocity ejecta revealed by _
hst _ images of optical knots @xcite .
the expansion of these knots is consistent with an explosion date close to 1680 @xcite , but the sn was not nearly as bright as other historical events like tycho or kepler . given the reddening towards the line of sight of the snr , this suggests a low ejected mass of @xmath13ni , @xmath26 @xcite .
cas a is also one of the few cc snrs that is young enough to detect the decay products of @xmath27ti and infer an ejected mass for this isotope ( @xmath28 , @xcite ) .
and @xmath7 values in small x - ray knots to hd+nei models .
the lines represent models appropriate for cas a , with @xmath29 , @xmath30 erg and different power - law indexes .
figure from @xcite . ]
several authors have used this wealth of observational information to build detailed models for the progenitor of the cas a sn . according to @xcite ,
all the observational constraints from the snr can be matched by a progenitor scenario where a 15 - 25 @xmath31 star loses its h envelope to a binary interaction and undergoes an energetic explosion in a ` stripped ' state .
this view has been challenged by later works , who favor a slightly more massive progenitor @xcite .
a particularly controversial issue is whether the progenitor went through a wolf - rayet phase before exploding , excavating a small cavity inside the dense @xmath32 csm that is being overrun by the blast wave today @xcite .
this brief ( and incomplete ) summary of what we have learned about the cas a sn from its snr exemplifies the great value of deep observations of nearby snrs .
although the explosion itself was never observed with the conventional tools used to study distant sne , the birth event of the cas a snr might be the best - studied demise of a massive star after that of sn 1987a .
this brief review on the x - ray observations of snrs has been written with two main goals .
the first is to illustrate the power of snr studies as probes of the sn phenomenon .
the unique capabilities of modern x - ray satellites like _ chandra _ offer a radically different view of the explosions and their progenitors , a vision that is independent of , and complementary to , the conventional studies of extragalactic sne at optical wavelengths .
the x - ray data sets assembled for well - known objects like tycho or cas a will be a lasting legacy of the _ chandra _ mission , and they represent the most detailed picture of the structure of sn ejecta currently available at any wavelength . the recent discovery that the statistical properties of the ejecta emission resolved by _ chandra _ can be used to distinguish cc from type ia snrs @xcite is a powerful example of the potential of these data .
it is therefore crucial that the campaign of x - ray observations of snrs with _
chandra _ and other satellites continue , and that deep exposures be completed for all the objects that do not have them .
the discovery of new young snrs , while difficult , is definitely possible , as illustrated by the @xmath33 yr old galactic snr [email protected] @xcite , and should be pursued in parallel to the study of well - known ones .
the second goal of this review is to emphasize the importance of careful modeling for the analysis of the x - ray spectra of snrs .
because of the nei character of the shocked plasma in snrs , it is extremely difficult to interpret their x - ray spectra in terms of magnitudes that can be used effectively to constrain explosion physics and progenitor scenarios . a robust , quantitative analysis usually requires putting together a dynamical model for the snr - in other words , the x - ray emission of a snr can not be interpreted without understanding the object as a whole , at least to some degree . in this context ,
one - dimensional hd+nei models have been successful in recovering the fundamental properties of sn explosions from the spatially integrated x - ray spectra of type ia snrs , and validation of the technique through the spectroscopy of sn light echoes has been possible in a few cases . these results are certainly encouraging , but much remains to be done before the full potential of the x - ray observations of snrs is realized . in order to take advantage of the spatially resolved spectroscopy capabilities of _ chandra
_ , it is necessary to build multi - dimensional hd+nei snr models .
this line of research has great promise as a benchmark for multi - dimensional sn explosion models , which hold the key to fundamental processes like the physical mechanism responsible for cc sne @xcite or the origin of the diversity within type ia sne @xcite .
present and future x - ray observations of snrs present many opportunities for sn research .
these opportunities are not devoid of challenges , but the excellent quality of the data obtained with _ chandra _ and other satellites fully warrant the effort required to meet them .
because of the unique view of the sn phenomenon that they offer , snrs should play a central role in shaping our understanding of sn explosions .
i want to thank kimberly arcand and peter edmonds for assistance with the _ chandra _ images .
i also want to express my gratitude to all the mebers of the sn and snr community with whom i have shared the excitement about the results of _ chandra _ and other x - ray satellites .
b , langer n , vink j , garca - segura g , van marle aj ( 2009 ) the hydrodynamics of the supernova remnant cassiopeia a. the influence of the progenitor evolution on the velocity structure and clumping .
_ a&a _ 503:495503 . | the unprecedented spatial and spectral resolutions of _ chandra _ have revolutionized our view of the x - ray emission from supernova remnants .
the excellent data sets accumulated on young , ejecta dominated objects like cas a or tycho present a unique opportunity to study at the same time the chemical and physical structure of the explosion debris and the characteristics of the circumstellar medium sculpted by the progenitor before the explosion .
supernova remnants can thus put strong constraints on fundamental aspects of both supernova explosion physics and stellar evolution scenarios for supernova progenitors .
this view of the supernova phenomenon is completely independent of , and complementary to , the study of distant extragalactic supernovae at optical wavelenghts .
the calibration of these two techniques has recently become possible thanks to the detection and spectroscopic follow - up of supernova light echoes . in this paper , i will review the most relevant results on supernova remnants obtained during the first decade of _ chandra _ , and the impact that these results have had on open issues in supernova research . he vast majority of the hundreds of sne discovered each year at optical wavelenghts explode very far away .
high quality light curves and spectra of some of these sne are obtained routinely from ground - based observatories , but the large distances involved restrict the view of the sn ejecta to a single line of sight through the receding photosphere , and usually preclude a detailed study of the immediate surroundings of the exploding star . only the occasional nearby objects like sn 1987a allow us to probe the explosion and its surroundings with better resolution , advancing our knowledge more effectively than many dozens of distant sne .
the high quality x - ray observations of young ejecta - dominated snrs performed by _
chandra _ and _ xmm - newton _ over the last decade have presented us with a different view of the sn phenomenon , a view that to a large extent complements the shortcomings of the optical studies of sne .
after the light from the sn fades away , the ejecta expand and cool down until their density becomes comparable to that of the surrounding material , either the ism or a more or less extended csm modified by the sn progenitor . at this point ,
a double shock structure is formed and the snr phase itself begins .
the forward shock or blast wave moves into the ambient medium , while the reverse shock moves into the sn ejecta . because the ambient densities are low ( @xmath0 or @xmath1 ) , the shocks are fast enough ( @xmath2 ) to heat the material to x - ray emitting temperatures .
these thermal x - rays from the high - z elements synthesized in the sn explosion are usually the dominant component in the x - ray spectra of young snrs .
the ability of _ chandra _ to perform spatially resolved spectroscopy on arcsecond scales reveals the structure of this shock - heated sn ejecta in galactic snrs with an unprecedented level of detail . because the morphology and spectral properties of the snrs also provide important clues about the structure of the ism or csm that is interacting with the ejecta , young snrs are the ideal setting to study simultaneously the aftermath of a sn explosion and the events leading up to it .
understandably , the _ chandra _ observations of snrs have drawn a great deal of attention from the sn community , becoming a prime benchmark for different aspects of sn theory , in particular the state - of - the - art three - dimensional models of cc and type ia sn explosions . however , the interpretation of these exceptional data sets is far from being a trivial task . in this review ,
i summarize the main results relevant to sn research obtained from snr observations by _ chandra _ and other x - ray satellites , point out the current challenges in the field , and discuss future prospects to meet these challenges .
the unique capabilities of _ chandra _ have obviously led to substantial progress in many other aspects of snr research , most notably the physics of cosmic ray acceleration in snr shocks @xcite .
these issues are outside the scope of the present work , which focuses on snrs as a tool to study sn explosions . for brevity
, i will discuss almost exclusively the so - called ` historical ' snrs , young objects whose ages are known well enough to model their dynamics with a certain degree of confidence .
i will also concentrate on results from non - dispersive x - ray spectroscopy , leaving out snr observations conducted with the high resolution gratings onboard _ chandra _ and _ xmm - newton_. |
quasars and active galactic nuclei ( agn ) are most probably powered by accretion of matter onto a supermassive black hole .
the amount of accreted matter required to explain the the observed energy output of seyfert galaxies and quasars is 0.001 - 10 solar masses per year , while the central mass required to maintain steady accretion for the observed luminosity ( the eddington limit ) is of the order of @xmath0 . comparing these two numbers gives a timescale @xmath1 years , which indicates a timescale for the evolution of the mass and related properties , such as the emergent spectrum .
as the mass grows due to the accreted material , the observed luminosity and spectrum change with time . since the continuum radiation ionizes the line - emitting material and drives the broad agn s , the change in the properties of the radiation from the central source changes also the properties of the emission lines .
in particular the spectral shape of the ionizing radiation would determine the ( the ratio between the energy flux in the line to the continuum energy flux at the line frequency ) of the s. the accretion disk postulated in agn provides a convenient ( although model - dependent ) framework for which the emergent continuum spectrum may be calculated ( e.g. sakura and sunyaev , 1973 ) , depending on the mass and the . alternatively , an empirical average agn uv spectrum may be related to the characteristic temperature of the .
baldwin ( 1977 ) has shown that the of the civ decreases with increasing continuum luminosity , a correlation known as the baldwin effect. a similar relation has been found also for other broad s. the origin of this effect is not understood ( for a recent review see shields and osmer , 1998 ) . in the first paper in this series ( wandel 1999a , hereafter paper i ) we have argued that the is related to the mass .
we have shown that the line decreases with increasing mass , so that the expected correlation between mass and luminosity in luminous agn implies an inverse correlation between and luminosity . in this work
we suggest that at least part of the can be attributed to evolution .
we estimate the relation between the and the continuum spectrum , and calculating the evolution of the continuum emission from the for changing mass and we show that for plausible evolutionary tracks the is anti - correlated with the continuum luminosity .
section 2 gives an empirical relation between the model parameters ( @xmath2 and @xmath3 ) and the implied evolutionary time scales , as well as the _ observed _ spectrum and luminosity . in section 3
we suggest some plausible evolutionary scenarios for the , and section in 4 we compose all these elements to produce the .
in order to maintain steady spherical accretion the luminosity must be less than the luminosity , @xmath4 or @xmath5 where @xmath6 , @xmath7 is the ratio , and @xmath8 .
we may define the time @xmath9 as the time required for an accretion - powered object radiating at the luminosity to double its mass , due to accreted matter , @xmath10 the observed luminosity implies an accretion rate of @xmath11 where @xmath12 is the efficiency .
combining these two expressions gives the mass , accumulated if the accretion rate is maintained during a time @xmath9 : @xmath13 the timescale for evolution ( for example , significant change in the mass ) is given by @xmath14 for @xmath15 this gives @xmath16 yr for a and @xmath17yr for a kerr .
the continuum spectrum depends on the mass .
the spectrum of the ( and modified ) disk is given by integrating over the entire disk , @xmath18 dr\ ] ] where @xmath19 is the planck function and @xmath20 is the transition radius from the optically thick to optically thin inner disk .
in paper i we have demonstrated that the basic property of the spectrum - thermal emission in the uv spectrum - leads to a peak in the sed , at an energy that is weakly decreasing with increasing mass ; from the approximate shape of the optically thick spectrum @xmath21 ( where @xmath22 is a normalization constant and @xmath23 is the cutoff frequency ) it is possible to derive an expression for the peak ( or cutoff ) energy of the uv spectrum : @xmath24 where @xmath2517ev for an around a and @xmath2620ev for a disk around a maximally rotating kerr , and @xmath27 is related to the maximal temperature in the optically thick part of the disk by @xmath28 for a and @xmath29 for a kerr one ( malkan 1990 ) .
plotting the cutoff frequency versus the luminosity it is possible to infer the mass and the ( or ratio ; fig . 1 in paper i ) .
the @xmath30 ratio found in paper i is 0.01 - 1 for quasars , 0.001 - 0.1 for seyferts , while nls1s are all in the @xmath31 range . a similar analyses for the has been performed by wandel and petrosian ( 1988 ) , who calculated for each pair of parameters @xmath2 and @xmath3 the actual observables , e.g. the uv luminosity and spectral index rather than fitting the blue - bump .
wandel and petrosian have calculated the flux and spectral index at 1450 , for a grid of parameters @xmath32 .
inverting the grid they have obtained contours of constant mass and constant ratio ( @xmath33 ) in the @xmath34 plane ( fig 2 ) . plotting in this plane samples of agn it is possible to read off the contours the corresponding mass and . comparing several groups of agn
a systematic trend appears : higher redshift and more luminous objects tend to have larger masses and luminosities closer to the limit ( see table 1 ) .
similar results were obtained by sun and malkan ( 1989 ) .
lccc & & & + agn group & log@xmath35 & log @xmath36 & @xmath37 + & & & + & & & + seyfert galaxies & 28 - 29.5 & 7.5 - 8.5 & 0.01 - 0.5 + low z quasars & 29 - 30.5 & 8 - 9 & 0.02 - 0.1 + medium z quasars & 30 - 31.5 & 8 - 9.5 & 0.1 - 0.5 + high z quasars & 31 - 32 & 9 - 9.5 & 0.03 - 2 + & & & + 1 be2f1.ps 1 .
accretion - disk parameters and spectral properties in the @xmath38 plane ( wandel and petrosian 1988 ) .
crosses : seyfert galaxies , triangles , open circles and filled circles : low , medium and high redshift quasars , respectively .
continuous lines : constant mass ; dashed lines : constant @xmath39 ( note that the labeling of the curves in the figure follows the notation @xmath40 ) .
in paper i we have shown that the line luminosity for some lines - in particular civ and ovi - is approximately proportional to the ionizing luminosity , as the line emissivity is weakly dependent on the ionization parameter .
the luminosity in the line is given by @xmath41 where @xmath42 is the effective covering factor , and @xmath43 is the peak emissivity .
the ionizing luminosity is determined by the maximal temperature of the optically thick part of the , which has been shown to depend on the mass approximately as @xmath44 for a variety of models .
combining these two results with eq .
[ equ : ecocas ] above gives the mass dependence of the - @xmath45\ ] ] and similar dependencies for the luminosity ( @xmath46 ) .
for the empirical power law agn spectrum in the euv ( @xmath47 ) we find @xmath48 with @xmath49 .
the evolution of the mass over cosmological times depends on the , which may of course change over time . from the theoretical efficiency and observed luminosity we can estimate the present . in order to determine theevolution
, we need to specify the variation of the over time .
we consider three representative scenarios : * constant * limited accretion ( constant ratio ) * spherical accretion from a homogeneous medium the tracks in the @xmath50 vs. @xmath51 plane of these three accretion scenarios are shown in fig .
2 . see also the discussion in wandel ( 1999b ) .
this would correspond to external feeding of the , which is regulated by an external potential , e.g. by stellar encounters , where the stellar motions are governed by the gravitational potential in the bulge of the host galaxy , far from the influence of the . as the mass of the grows , @xmath3 does not change , and hence the accretion parameter , @xmath52 .
eventually will become very sub- , and the will be starved , or at least on a diet . since @xmath53 , in this scenario the ( bolometric ) luminosity also is also constant .
note that the spectral energy distribution does change ( though slowly ) as @xmath2 increases .
this would be the case in a that is over - fed .
the can not become very super- , and the source will regulate the to @xmath54 or @xmath55 . since @xmath56 , @xmath57 constant implies @xmath58 and hence @xmath59 .
in this case we assume the matter supply comes from a homogeneous distribution ( gas or stars ) , due to the gravitational potential of the . on large scales , the accretion will be spherical , and the accretion radius at which the gravitational potential becomes significant is given by @xmath60 where @xmath61km / s is the stellar velocity dispersion .
the is given by the bondi formula @xmath62 where @xmath63 is the stellar number density .
in this case @xmath64 and @xmath65 , that is , the ratio increases with time .
eventually the ratio will approach unity and the accretion will become limited . a similar expression can be derived for a non - homogeneous medium , with a density profile @xmath66 . assuming the velocity dispersion is independent of @xmath67 , eqs .
[ equ : rac ] and [ equ : sar ] give for the in that case @xmath68 if also the velocity dispersion has a radial dependence , @xmath69 then we get a more complicated dependence , @xmath70 in this case , any functional dependence is possible . if for example @xmath71 and @xmath72 , @xmath73 , and if @xmath74 ( point mass ) and @xmath72 , @xmath75const . 1 be2f2.eps 2 .
black hole evolution for the three scenarios described in the text in the cutoff energy vs. the monochromatic luminosity at 5100 .
diagonal dashed lines indicate constant masses , and diagonal dotted lines indicate the ratio .
vertical arrows describe the direction of evolution for a constant , horizontal arrows show the evolution for a spherically accreting from a homogeneous medium , and diagonal dashed arrow are for accretion at a fixed ratio .
the superimposed data is pg quasars ( diamonds ; from boroson and green , 1994 ) , seyfert 1 galaxies ( triangles and squares , the latter indicate agn with blr reverberation data from wandel , peterson and malkan 1999 , with luminosities modified for @xmath76 ) and narrow line seyfert 1s ( boller , brandt and fink , 1996 )
in the previous section we have found the relation between the accretion parameters @xmath2 and @xmath3 .
it is easy to find the functional dependence of the continuum luminosity on the mass for several accretion scenarios .
this cab be shown as evolutionary curves in the @xmath77 and uv - spectrum vs. luminosity plane ( figs . 1 and 2 ) .
for example , an bound will evolve along the constant @xmath33 curves ( dashed curves in fig .
1 ) . a spherically accreting will evolve along curves of @xmath65 , which correspond to nearly horizontal ( from left to right ) lines in fig . 1 , and a constant will yield @xmath78 which are almost vertical ( upward ) lines in fig .
similar evolutionary tracks are shown in fig .
2 ( note however that the vertical axes of figs . 1 and 2 have an opposite sense ) . combining this with the dependence of the effective radiation - temperature on the accretion parameters , and using the result that the luminosity is positively correlated with the radiation temperature , it is possible to find the relation between the equivalent width and the continuum luminosity for each pair of accretion scenario and temperature model .
llll & & & + accretion scenario : & constant @xmath3 & constant @xmath30 & spherical + @xmath79&const&@xmath80&@xmath81 + + @xmath82 model : & & & + & & & + @xmath83&@xmath84 & @xmath44&const + & & & + @xmath85&@xmath44 & @xmath44&@xmath44 + & & & + @xmath86 or @xmath87(bb / rad)&const & @xmath44&@xmath84 + & & & + & & & + @xmath87(bb - var)@xmath88&const & @xmath44&@xmath84 + & & & + the dependence of the effective temperature on the mass is summarized in table 2 .
the header shows the three accretion scenarios and the corresponding dependence of the continuum luminosity on the mass .
the next row gives the dependence of the luminosity on the mass for each scenario .
the left column shows the temperature models . for each pair
the table gives the functional dependence of @xmath82 on m. since the -radiation boundary and the photospheric models , have a very similar dependence on the accretion parameters , we combine them in one row in tables 2 and 3 .
table 2 has nine ( @xmath89 ) pairs , for models , and an additional row for the model independent - variability estimate of @xmath82 ( marked `` @xmath87(bb - var ) '' ) . for all pairs except two ( constant @xmath3 with @xmath90 and homogeneous medium accretion with @xmath83 ) .
the effective disk temperature is decreasing with increasing mass .
we note that this is true also for the general temperature - luminosity relation ( eqs .
21,22 in paper i ) , so the model is not essential . in that case
the anti - correlation between temperature and luminosity is built into the the expression for the temperature .
for all accretion scenarios the optical and uv continuum luminosity is increasing with @xmath2 .
this is certainly the case for bounded accretion and for accretion from a homogeneous medium ; for a constant the bolometric luminosity is constant , but as the mass increases the peak of the spectrum moves to lower frequencies .
since the optical or near uv continuum are on the rayleigh - jeans part of the spectrum , the luminosity in these bands increases as the mass increases even if the total luminosity remains constant .
however , since for a radially extended the low energy slope in the near uv is much flatter than for a planck spectrum , this is a very small effect .
as the @xmath3=const is a limiting case , and in reality the is expected to grow , even if slower than linearly , as argued above , we replace the @xmath3=const .
column in table 3 by a more general case , @xmath91 , @xmath92 , which allows for an evolution in luminosity . applying the dependence of l on m it is possible to deduce the dependence of t on l. relating the ew to the spectral uv peak energy ( paper i )
we can quantify the expected for each of the models .
* decreasing .
* in the three cases considered above , the increases with time or remains constant .
what if the accretion rate decreases with time ?
in the two latter -temperature models in paper i ( 5.2.c and d ) we find @xmath93 , so @xmath82 increases , and so does the line luminosity .
if we take @xmath94 the continuum luminosity decreases , and there is still a negative correlation .
this is the case also in the general l - t relation . in the model ,
the temperature is independent of the , and since the mass increases , the for the and radiation - pressure dominated models the situation is less clear , because @xmath82 decreases with decreasing @xmath3 ( as @xmath95 for the model and being independent of the in the radiation - pressure dominated model , it still decreases as @xmath44 ) .
since @xmath53 decreases , the question whether @xmath82 is correlated or anti correlated with @xmath51 would depend on the details of the behavior of the . how does the vary ?
the is defined as @xmath96 , where @xmath97 is the flux per unit wavelength at the wavelength of the line . in paper
i we have shown that @xmath98 depends on the ionizing continuum , which correlates strongly with @xmath82 .
the luminosity in the line is also roughly linearly dependent on the continuum luminosity , @xmath99 , so that @xmath100 @xmath101 being a monotonic function with an exponential increase near @xmath102ryd .
this gives for the @xmath103.\ ] ] applying the observed uv spectral slope , @xmath104 ( @xmath105 ) , we can derive from the temperature mass dependence shown in table 2 a relation between and luminosity . from paper
i we have @xmath106
so we can derive the expected for each model quantitatively .
table 3 assumes @xmath107 . for a @xmath108 spectrum
the exponent of l should be multiplied by @xmath109 .
lllll & & & & + accretion scenario : & & @xmath110 & @xmath30 & spherical + @xmath111&&@xmath112&const&@xmath80 + + @xmath82 model:&@xmath113(ev ) & & & + & & & + @xmath83&@xmath114&@xmath115 & @xmath116&const + & & & & + @xmath85&@xmath117 & @xmath118&@xmath116&@xmath119 + & & & & + @xmath86 & @xmath120&@xmath116 & @xmath116&@xmath121 + & & & & + & & & & + @xmath87(bb - var)&@xmath122&@xmath116 & @xmath116&@xmath116 + & & & & + we see from table 3 that for most combinations of effective temperature model and accretion scenario there is an evolutionary , and the predicted slope ( @xmath123 ) is @xmath124 , consistent with the observed of 0.15 - 0.2 .
we conclude that for many accretion - scenarios and temperature - models the line luminosity decreases with increasing mass , except the two for which the effective temperature is constant , implying an evolutionary : as the evolves over cosmological times due to the mass accreted , the continuum luminosity ( optical or near uv ) increases , while the decreases .
also the two pairs with a constant effective temperature will eventually show a . to see this
, note that for the ( @xmath125(spherical ) ) pair , @xmath33 is increasing as @xmath2 , so it will eventually approach the , and move from the @xmath83 model to one of the other @xmath82 models , which ( for @xmath3(spherical ) ) do have a . in the other pair , ( constant @xmath3 with @xmath90 )
, the situation is opposite , but with the same result : since there @xmath78 , eventually the will become enough sub - so that the @xmath83 model will apply , which for the constant scenario does have a .
in paper i we have made the case for a due to the dependence of the on the mass .
we have shown that the line decreases with increasing mass , so that the expected correlation between mass and luminosity in luminous agn implies a . in this work
we show that an intrinsic evolution of the mass and can create an evolutionary .
in addition , the relation between the emission - line luminosity and the ionizing continuum shape depends on the unknown density profile of the line emitting gas , and on the unknown differential covering factor , which could show intrinsic evolution ( for an individual object along it s lifetime ) as well as cosmological evolution .
how can we infer whether evolution is the _
main cause _ of the baldwin effect ?
this work shows that evolution is a possible cause .
but there are several other possible causes as well .
for example , the ranges in actual black hole masses and accretion rates in individual agn might deviate greatly ( and systematically ) from the average relations assumed .
we suggest that the growth of the central mass due to the accreted matter causes the luminosity and the spectrum to evolve over over the agn life time . using general model - independent relations , or the thin spectrum
, we estimate the evolution of the and the continuum spectrum , and show that for plausible evolutionary tracks as well as for the model - independent temperature estimate and for most variants of the thin model the decreases with increasing continuum luminosity implying an evolutionary origin to the baldwin effect .
stimulating discussions with jack baldwin during the meeting `` quasars as standard candles in cosmology '' at la serena , chile , triggered this work .
i also acknowledge valuable discussions with gary ferland , matt malkan and vah ' e petrosian , the contribution of an anonymous referee , and the hospitality of the astronomy department at ucla and stanford university .
edelson , r. and nandra , k. 1999 , apj malkan , m.a .
1990 , in iau colloquium no . 129 , `` structure and emission properties of accretion disks '' , eds . c. bertout et.al . , editions frontiers : paris , p. 165 .
shields , j. and osmer , p. 1998 in `` quasars as standard candles for cosmology '' , la serena , chile , ed .
g. ferland , asp .
sun , w.h . and malkan , m.a .
1989 , , 346 , 68 .
wandel , a. 1997 , 430 , 131 .
wandel , a. 1999a ( paper i ) , , in press wandel , a. 1999b , , 509 , l39 wandel , a. and mushotzky , r.f .
1986 , , 306 , l61 .
wandel , a. , peterson , b.m . and malkan , m.a . 1999 ,
apj , 526 , wandel , a. and petrosian , v. 1988 , , 329 , l11 . | we suggest that the growth of the central black hole in active galactic nuclei ( agn ) due to the matter accreted over the agn lifetime causes an evolution of the luminosity and spectrum . in a previous paper
we have shown that the effective temperature of the uv continuum spectrum and the are anti - correlated with the black - hole mass . here
we estimate the change in the equivalent width of the emission lines due to the growth of the black hole mass caused by the accreted material and show that for several plausible accretion scenarios and effective - temperature models the evolving equivalent width is anti - correlated with continuum luminosity , implying that intrinsic evolution could contribute to the baldwin effect . #
1#2#3#4 = # 1 cm # 2 cm .3 cm |
this paper is concerned with schrdinger operators @xmath8 in @xmath9 with dirichlet condition at @xmath10 . in particular , we are interested in the dependence of various spectral properties of @xmath0 on the parameter @xmath5 .
note that potentials of the type @xmath11 have a special role , since the resulting operator @xmath0 is scaling invariant .
moreover , it is known that the potentials which satisfy @xmath12 as @xmath13 represent a borderline case for certain important spectral inequalities such as dispersive or strichartz estimates , see @xcite .
it is therefore not surprising that schrdinger operators with inverse square potentials have recently attracted certain attention ; we might mention for example the heat kernel bounds obtained in @xcite , or strichartz estimates in dimension three studied in @xcite .
inverse square potentials appear naturally also in connection with two - dimensional schrdinger operators with aharonov - bohm - type magnetic field , see @xcite .
main objects of our interest here are the spectral density @xmath14 of @xmath0 , and the unitary group @xmath4 .
in particular , we are going to study the asymptotic behavior of @xmath1 for @xmath2 and the @xmath3 dispersive estimates associated to the evolution operator @xmath4 .
it is very well - known that the asymptotic behavior of @xmath1 for small @xmath15 is closely related to the asymptotic behavior of @xmath4 for large @xmath16 .
there is a huge amount of literature on this subject , see e.g. @xcite and references therein .
we are not going to discuss this connection any further since it will not be used in our proofs . for general one - dimensional schrdinger operators of the type @xmath17 the behavior of both @xmath18 and @xmath19
is known provided the potential @xmath6 decays fast enough at infinity . in particular ,
if zero is a regular point of @xmath20 , ( which is the generic case ) , then @xmath21 in a suitable operator topology , see @xcite . accordingly , for such short range potentials , under certain regularity conditions , murata @xcite proved @xmath22 where @xmath23 is a weight function with a sufficient growth at infinity .
the corresponding @xmath3 was established by schlag @xmath24 with @xmath25 , see @xcite .
it is important to mention that the decay conditions on @xmath6 , under which all the above results were obtained , imply that @xmath26 as @xmath27 .
the goal of the present note is to show that if @xmath6 is of type @xmath28 with @xmath29 , then the asymptotic relation is no longer valid and has to replaced by a new one , and , on the other hand , the estimates and can be improved . in particular
@xmath1 decays _
faster _ to zero than in , see theorem [ thm-1 ] .
accordingly the decay in the dispersive estimate can be improved provided the weight function @xmath30 grows fast enough at infinity , see theorem [ thm-2 ] .
although our results regard a family of schrdinger operators with explicit potentials , it can be expected that similar results should hold also if @xmath0 is perturbed by a sufficiently short - range perturbation .
it should be finally mentioned that our main results , i.e. theorems [ thm-1 ] and [ thm-2 ] , fail in the case of schrdinger operators on the whole line due to the presence of the zero resonance .
we set @xmath31 on @xmath32 . for any @xmath33 we denote @xmath34 let @xmath35 be the space of bounded linear operators from @xmath36 to @xmath37 and let @xmath38 denote the corresponding operator norm .
finally , we put @xmath39 we have [ thm-1 ] let @xmath40 .
then for any @xmath41 and any @xmath42 it holds @xmath43 in @xmath44 , where @xmath45 is the integral operator in @xmath9 with the kernel @xmath46 equation shows that for positive values of @xmath5 the density @xmath1 is of lesser order than in the case of a short - range potential , see equation .
for a throughout discussion of threshold expansion of resolvents of one - dimensional operators with short - range potentials we refer to @xcite .
asymptotic behaviour of schrdinger groups generated by operators with inverse square decay on conical manifolds was studied in @xcite in the setting of weighted @xmath47spaces .
[ thm-2 ] let @xmath48
. then for any @xmath49 $ ] there exists a constant @xmath50 such that @xmath51 for @xmath52 the dispersive estimate can be derived from ( * ? ? ?
* thm.1.11 ) by considering the restriction of inequality ( * ? ? ?
* eq.(1.29 ) ) to radial functions . on the other hand ,
the result for @xmath53 , namely the faster decay of @xmath54 in @xmath16 is new . the maximal decay rate @xmath55 , achieved by the choice @xmath56 ,
should be compared with the @xmath57 decay rate in the estimate .
note also that in the border - line case @xmath58 , which means @xmath59 , theorem [ thm-2 ] with the choice @xmath60 gives the decay rate @xmath61 , which is the decay rate of the free evolution @xmath62 in dimension two .
this is not surprising since the operator @xmath63 in @xmath9 with dirichlet boundary condition at zero is unitarily equivalent , by means of the unitary mapping @xmath64 , to the laplacian @xmath65 in @xmath66 restricted to radial functions .
an immediate consequence of theorem [ thm-2 ] is the following let @xmath48 .
then for any @xmath49 $ ] and any @xmath67 there exists a constant @xmath68 , depending only on @xmath69 and @xmath70 , such that @xmath71 let @xmath72 and let @xmath73 . then by the cauchy - schwarz inequality @xmath74 and @xmath75 where we have used the fact that @xmath76 .
hence from theorem [ thm-2 ] and we obtain @xmath77 this proves . inequality should be compared with the estimate valid for short - range potentials .
for simplicity we shall drop the index @xmath5 in the sequel and write @xmath78 instead of @xmath79 .
we also use the notation @xmath80 we first study the solutions @xmath81 of the generalized eigenvalue equation @xmath82 after setting @xmath83 , equation writes @xmath84 with @xmath85 . the latter is a bessel equation of the first kind , see ( * ? ? ?
* sec.9.1 ) .
we now find two solutions @xmath86 of which satisfy @xmath87 and @xmath88 for im@xmath89 . since @xmath90 see ( * ? ?
* eqs.9.1.3 , 9.2.3 ) , and @xmath91 by ( * ? ? ?
* eq.9.1.7 ) , the sought solutions @xmath92 and @xmath93 take the form @xmath94 hence by the theory of sturm - liouville problems we obtain the resolvent kernel @xmath95 the stone formula then implies that @xmath96 from we now easily verify that @xmath97 let us define the rest term @xmath98 as the integral operator in @xmath9 with the kernel given by @xmath99 to prove theorem [ thm-1 ] we need the following [ first ] for any @xmath41 and any @xmath100 we have @xmath101 we will use the fact that @xmath102 from ( [ densi ] ) we get that @xmath103 note that the operator @xmath45 is hilbert - schmidt in @xmath44 .
this follows from the identity applied to @xmath45 .
hence by applying the taylor formula to the operator @xmath104 at @xmath105 we find that the claim of the lemma will follow if we show that @xmath106 where @xmath107 denotes the hilbert - schmidt norm in @xmath9 . using the recurrence relations for the derivatives of @xmath108 : @xmath109 see ( * ? ? ?
* eq.9.1.27 ) , we get from ( [ dens ] ) @xmath110 \ . \label{deriv}\end{aligned}\ ] ] hence by the cauchy - schwarz inequality @xmath111 where @xmath112 and @xmath113 is a constant independent of @xmath15 . to estimate the last two integrals we will need a point - wise estimate on the bessel function @xmath108 . from the integral representation @xmath114 see ( * ?
* eq.9.1.20 ) , it follows that @xmath115 for all @xmath116 and @xmath117 . on the other hand , by (
* eq.9.1.20 ) we have @xmath118 for all @xmath116 and @xmath119 .
a combination of these two upper bounds then implies that for any @xmath120 there exists a constant @xmath121 such that @xmath122 in @xmath123 and with @xmath124 in @xmath125 together with the fact that @xmath126 , we find @xmath127 in view of this implies and therefore completes the proof .
theorem [ thm-1 ] now follows from and
. we will prove theorem [ thm-2 ] by estimating the integral kernel of the operator @xmath128 . to provide a formula for the integral kernel
, we will follow ( * ? ? ?
* sec.5 ) , where the formula for the integral of the heat semi - group @xmath129 was established , see also @xcite . equation in combination with the weyl - titchmarsh - kodaira theorem , cf .
* chap.13 ) , shows that the operator @xmath0 is unitarily equivalent to a multiplication operator , namely we have @xmath130 where @xmath131 denotes the operator domain of @xmath0 and the mappings @xmath132 are given by @xmath133 the mapping @xmath134 and @xmath135 define unitary operators on @xmath136 .
let @xmath137 . by ( * ?
* thm.3.1 ) @xmath138 in view of we thus get @xmath139 where we have used ( * ? ? ?
* eq.4.14(39 ) ) to calculate the integral with respect to @xmath140 .
moreover , from ( * ? ? ?
* eq.9.6.18 ) it follows that the function @xmath141 is bounded on every compact interval uniformly with respect to @xmath41 .
since the support of @xmath142 is compact , we can use the dominated theorem and interchange the limit and integration in . taking the limit @xmath143 and using the identity @xmath144 ,
see ( * ? ? ?
* eq.9.6.3 ) , we obtain @xmath145
now we apply the upper bound with @xmath146 $ ] and @xmath147 .
this yields @xmath148 the last equations now imply that @xmath149 for all @xmath74 .
this proves inequality .
h.k . would like to thank the institut fourier in grenoble for the warm hospitality extended to him during his stay .
the work of h.k . has been partially supported by the miur - prin2010 - 11 grant for the project
calcolo della variation .
both authors thank the referee whose remarks and comments helped them improve the original version of the text .
w. schlag : dispersive estimates for schrdinger operators : a survey
. mathematical aspects of nonlinear dispersive equations , 255 - 285 , ann . of math . stud .
, 163 , princeton univ . press , princeton , nj , 2007 | we consider schrdinger operators @xmath0 given by equation below .
we study the asymptotic behavior of the spectral density @xmath1 for @xmath2 and the @xmath3 dispersive estimates associated to the evolution operator @xmath4 .
in particular we prove that for positive values of @xmath5 , the spectral density @xmath1 tends to zero as @xmath2 with higher speed compared to the spectral density of schrdinger operators with a short - range potential @xmath6 .
we then show how the long time behavior of @xmath4 depends on @xmath5 .
more precisely we show that the decay rate of @xmath4 for @xmath7 can be made arbitrarily large provided we choose @xmath5 large enough and consider a suitable operator norm .
= 1 |
quasar absorption line systems , as traced by their , provide a unique way to study our universe .
current classifications schemes for systems with detected file them into three main categories : those that are characterized by their strong absorption , those that reveal weak , narrow , single cloud profiles , and those that exhibit weak , multiple cloud absorption . in this paper
we analyze in detail the absorption profiles from a particular multiple cloud weak absorber at @xmath10 along the line of sight toward pg @xmath1 .
the longer term goal of a collection of similar studies will be to understand the relationships between the different classes of absorption systems , and the connections with the different types of galaxies and structures at various redshifts .
although there are no distinct divisions between the three categories of absorption systems , by their properties they appear to be related to three different types of gaseous structures at @xmath11 .
strong absorbers [ those with @xmath12 ] show lyman limit breaks and contain multiple clouds spread over tens to hundreds of kilometers per second @xcite .
a large majority of these absorbers are known to be associated with luminous galaxies ( @xmath13 , where @xmath14 is the schechter luminosity ) , within an impact parameter of @xmath15 kpc of the quasar @xcite .
all of the strong absorbers ( @xmath16 at @xmath17 @xcite ) can be accounted for by regions of this size around the known population of luminous galaxies .
the strong absorber spectral profiles , in chemical transitions of low and high ionization states , generally require multiple phases of gas , i.e. regions of differing densities that are spatially distinct ( e.g. @xcite ) .
the kinematics of the low ionization gas is generally consistent with what one would expect from the combined disks and halos of galaxies of a variety of morphological types @xcite .
in contrast to the strong absorbers , the single component systems with weaker ( rest frame equivalent widths @xmath18 ) are typically not known to be directly associated with luminous galaxies ( i.e. they are not within @xmath19@xmath20 kpc of @xmath21 galaxies ) @xcite . these single component weak systems have a significant absorption cross - section , with @xmath22 for @xmath23 at @xmath24 @xcite , similar to the value for all strong absorbers . the distribution of the number of voigt profile components used to fit absorption systems is roughly gaussian with a median of seven components .
however , there is a strong excess of single cloud components relative to the rest of the distribution ( see fig . 2 in @xcite and table 7 of @xcite ) .
this implies that the single component weak absorbers are primarily produced by a different class of object than the strong absorbers .
consistent with this interpretation , most of the single component weak absorbers do not produce lyman limit breaks @xcite .
their profiles are generally quite narrow , with doppler parameters of @xmath25
@xmath26 .
their profiles are broader and require a separate , higher ionization , phase of gas .
their low ionization phases are inferred to have metallicities greater than a tenth solar and to have sizes less than @xmath27@xmath20 pc @xcite .
the general properties of the single cloud weak absorbers suggest an origin in some type of faded early extragalactic star cluster or in metal rich fragments in cold dark matter mini
halos failed galaxies @xcite
. the third category of systems is characterized by multiple , weak components spread over tens of kilometers per second .
about one third of the weak absorbers are in this category @xcite . by definition , these are `` weak absorbers '' , since the total rest frame equivalent width of 2796 is @xmath28 . unlike the the strong absorbers , little is directly known about the environments of the multiple component weak systems .
however , the multiple component weak category appears consistent with an extension of the distribution of the numbers of components fit to strong absorbers ( see fig . 2 in @xcite ) .
the weakness and kinematics of the multiple component weak profiles might suggest a line of sight through the outer regions of a galaxy where the gas could be more diffuse and less dense .
it is also possible that a dwarf galaxy , or even a low metallicity giant galaxy would give rise to such weak absorption . in this paper
, we study a multiple component , weak [ rest frame equivalent width , @xmath29 ] absorber at @xmath11 .
this absorber was modeled previously by @xcite , using low
resolution fos / hst data along with the high resolution profiles from hires / keck .
they determined that the clouds have a relatively low metallicity of @xmath30 solar , based on a comparison to a partial lyman limit break .
the , detected in the fos spectrum , was clearly offset from the low ionization clouds .
this required an additional high ionization cloud producing absorption @xmath31 to the blue of the detected clouds .
@xcite predicted that this cloud should produce observable , , and in the high resolution stis spectrum .
however , with just the low resolution uv spectrum , they were unable to determine if the detected at the same velocity as could arise in the same phase of gas with the , or if a separate phase would be needed .
also , the ionization conditions and metallicities of the high ionization clouds could not be well constrained .
stis / hst spectra of pg @xmath1 are now available , with high resolution coverage of and the lyman series , , , , , and . with these new data ,
we aim to test the models of @xcite , and to reach more detailed conclusions about the phase structure , metallicity , ionization conditions and kinematic properties of the @xmath10 system toward pg @xmath1 . in [ sec :
data ] we describe the spectra of quasar pg @xmath1 used in our analysis , obtained with stis / hst , fos / hst and hires / keck .
we then present the data for the @xmath10 system along this line of sight in
[ sec : system ] and make a qualitative comparison of the individual chemical transitions .
our analysis method , considering photoionization and collisional ionization models , is outlined in
[ sec : method ] . in [ sec : results ] we present results from modeling of this system and we summarize the conclusions in
[ sec : summary ] . in
[ sec : discussion ] , we discuss possible scenarios for the origin of the @xmath10 system and the general implications for the nature of multiple component , weak absorbers .
spectra of the @xmath32 quasar pg @xmath1 , obtained with the three instruments hires / keck , stis / hst , and fos / hst , were used to study the various transitions from an intervening absorber at @xmath33 .
other systems along this line of sight are published elsewhere .
there are three weak , single cloud absorbers at @xmath34 , @xmath35 and @xmath36 @xcite and a strong absorber at @xmath37 @xcite .
although a wfpc2/hst image of the quasar field exists , it is not deep , so it provides only a limited constraint on the luminosity of the host galaxy of this @xmath10 absorber .
after subtraction of the point spread function of the @xmath38 quasar from the image , @xcite did not detect any galaxies within the @xmath39 planetary camera field of view to a magnitude of @xmath40 ( at the @xmath41 level ) .
this implies that there are no galaxies brighter than @xmath42 with impact parameter less than @xmath43 kpc at the redshift of our absorber ( @xcite used @xmath44 ) . due to uncertainties in the psf
the limit is not as restrictive within @xmath45 kpc .
since even strong absorbers are associated with galaxies with luminosities as small as @xmath46 , this image does not allow us to usefully constrain the host galaxy properties for this or any of the other systems along the line of sight .
hires / keck @xcite spectra at @xmath47 ( fwhm@xmath48 ) were obtained on 1995 july 4 and 5 .
the wavelength coverage is @xmath49 to @xmath50 .
the typical signal - to - noise ratio per resolution element is @xmath19 .
the transitions we study here are mg0.1emii
@xmath51 and @xmath52 .
@xcite describe the data reduction , continuum fitting , and voigt profile fitting .
we adopt the column densities and doppler parameters presented in @xcite .
archival stis / hst @xcite spectra were used to study , , , , ly0.1em@xmath53 , @xmath54 , @xmath55 , @xmath56 , si0.1emiv @xmath57 , , and o0.1emvi @xmath58 .
two data sets were obtained with different tilts of the e230 m grating with an aperture of @xmath59 x @xmath59 ( @xmath60 , fwhm@xmath61 ) .
the first was obtained by jannuzi et al .
( proposal identification 8312 , wavelength range of 2303 to 3111 ) and the second by burles et al .
( proposal identification 7292 , wavelength range from @xmath62 to @xmath63 ) .
a third data set was obtained by burles et al .
( wavelength coverage @xmath64 to @xmath65 ) with the @xmath66 x @xmath67 aperture of the g230 m grating ( @xmath68 , fwhm@xmath69 ) .
we used this spectrum to study the higher order transitions of the lyman series and the lyman limit break .
the reduction and calibration of all stis data were performed using the standard stis pipeline @xcite .
continuum fitting was performed using the techniques described in @xcite .
we co - added the e230 m spectra in cases of duplicate wavelength coverage to obtain higher signal - to - noise ratios ( per resolution element ) ranging from @xmath53 to @xmath5 .
we have verified that wavelength calibration is consistent between the stis and hires datasets by comparing mg0.1emii @xmath51 , 1260 and 1335 in the three single component , low ionization absorption systems found along this same line of sight @xcite .
fos / hst data , with lower resolution ( @xmath70 ) and wavelength coverage @xmath71 to @xmath72 ( g270h grating ) , were obtained as part of the quasar absorption line key project @xcite .
these data were only used to study the c0.1emiv @xmath73 absorption , since all other transitions had higher resolution coverage from stis .
figure [ fig : data ] presents the absorption profiles of selected transitions for the @xmath10 absorber .
these transitions are , mg0.1emii @xmath51 , @xmath74 , @xmath55 , @xmath75 si0.1emiv @xmath57 , and o0.1emvi @xmath58 . also , c0.1emiv
@xmath73 , covered in the lower resolution fos spectrum , is shown .
the region of the spectrum that provides a limit on @xmath76 and @xmath52 is also shown .
the rest - frame velocities are aligned with the @xmath77 transition , where @xmath78 is defined at the apparent optical depth median of the profile .
table @xmath79 gives the rest frame equivalent widths or @xmath41 equivalent width limits for the illustrated transitions . in the stis spectra , two low ionization `` kinematic subsystems '' , separated by @xmath80 ,
are detected in and .
we refer to these as subsystem a ( at @xmath81 ) and subsystem b ( at @xmath82 ) .
in addition , the doublet is detected in two broad components , one @xmath83 to the red of subsystem a ( red broad component ) and the other @xmath83 to the blue of subsystem b ( blue broad component ) .
table @xmath79 lists separately the contributions to the equivalent width from subsystems a and b. in subsystem a , the profile was fitted with four voigt profile components over a @xmath19 interval @xcite .
the shows the same kinematic structure , yet there may be a very slight kinematic offset , with each of the four components redward with velocity offsets ranging from @xmath84@xmath85 . to illustrate this ,
the 2796 and 1394 profiles are superimposed in figure [ fig : offset ] . as mentioned in section [ sec : stis ] , the alignment of similar transitions in other absorbers along this same line of sight verifies that this is not likely to be a result of wavelength calibration @xcite .
in contrast , the profile is smooth . in subsystem b ( @xmath86 )
an echelle order break prevented observation of @xmath77 .
however , the weaker @xmath87 transition is not detected .
the kinematics of subsystem b is characterized by an asymmetric blend with a velocity width of @xmath88 .
figure [ fig : data ] also presents the c0.1emiv @xmath73 doublet observed with fos / hst .
the profiles are resolved , indicating a kinematic velocity spread slightly greater than the instrumental resolution of @xmath89 ( fwhm ) .
the equivalent widths in table 1 were derived from this lower resolution fos spectrum .
the upper panel of figure [ fig : lya ] shows the partial lyman limit break observed with the g230 m grating of stis / hst .
the optical depth is @xmath90 , using the method of @xcite , which implies @xmath91 .
we derive constraints on the kinematic , chemical , and ionization conditions of the absorbing gas .
our method is to compare the absorption profiles of the observed chemical transitions ( presented in figures @xmath79 and @xmath92 ) with those predicted by photoionization and collisional ionization models .
we summarize the method here , and it is also described and applied in our earlier papers @xcite . for photoionization , we use the code cloudy @xcite and adopt the extragalactic background spectrum of @xcite for the ultraviolet ionizing flux .
we use the spectrum normalized at @xmath93 .
we have also considered the influence of stellar ionizing flux , which we describe below in [ sec : stellar ] . for collisional ionization ,
we adopt the equilibrium models of @xcite , which provide the column densities of various transitions for assumed temperatures and metallicities . for photoionization
, each cloud is modeled as a constant density plane - parallel slab with the ionizing flux incident on one face .
we assume a solar abundance pattern , however , deviations from this pattern are considered when suggested by the data .
the ionization conditions are defined by the ionization parameter , @xmath94 , where @xmath95 , the ratio of the number density of hydrogen ionizing photons to the number density of hydrogen .
the quantity @xmath96 is fixed by the normalization of the extragalactic background spectrum , so that the relationship between @xmath94 and @xmath97 is @xmath98 .
since the abundance pattern and ionization parameter are interdependent , the models are not unique .
we have found that the absorption lines can not be modeled as a single ionization `` phase . '' in this context , phases denote regions or `` clouds '' with similar ionization parameter , density , metallicity , and temperature . for simplicity , clouds in the same phase are assumed to have the same metallicity .
moreover , we consider physical scenarios that require a minimum number of phases to explain the data .
the lower ionization transitions in the @xmath10 absorber reveal clear kinematic structure and we therefore use them as a template to place constraints on the phase structure of the system . in order to obtain the best initial estimates for the cloud properties , we begin with the lowest ionization transition that is most clearly detected and resolved in each of the subsystems .
we obtain the column densities and doppler @xmath99 parameters for this transition using voigt profile ( vp ) fits . in subsystem
a , we adopt the column densities and doppler parameters for the doublet from @xcite as stated above . in subsystem b ,
@xmath100 is not covered by keck and so we fitted the @xmath75 profile .
vp profile fits are dependent on spectral resolution .
although they do not provide a unique description of the physical conditions of the gas , we can still use them to identify a plausible range of parameter space .
the data are of lower resolution than the data . compared to the
, the fits to the grouping will in general likely yield fewer clouds , which will have larger @xmath99 parameters and smaller column densities , and this is likely to be a bias rather than a real , physical effect .
we use the optimization mode of cloudy . for each cloud , we `` optimize '' on the vp column density of a selected transition , for an assumed ionization parameter and metallicity .
cloudy then calculates the column densities for all other ionization species and determines the kinetic temperature .
this temperature is used to calculate the thermal component of the @xmath99 parameter for the optimized transition .
this thermal @xmath99 is then compared to the vp @xmath99 parameter to calculate the turbulent component assuming gaussian turbulence .
doppler parameters for all other transitions are then computed using this turbulent component and scaling for ion mass . to compare the cloudy model predictions to the observations , we generate synthetic spectra based upon the cloudy column densities and derived @xmath99 parameters .
the synthetic spectra are convolved with the appropriate instrumental profile , and are assigned the same pixel sampling as the instrument with which a given transition was observed . for hires and fos , we used gaussians with fwhm@xmath101 and @xmath89 , respectively .
for stis , the instrumental spread function for the e230 m grating was taken from the stis performance web pages at _
www.stsci.edu_. the ionization parameters and metallicities of cloudy models are then iteratively adjusted in order to minimize deviations between the observed and synthetic spectra .
although we considered a strict @xmath102 criterion for optimization , we found this to be unrealistic given the tendency for one or two pixels to have a dominant contribution to that statistic .
visual inspection , comparing the model profiles to the data , generally yield ionization parameters and metallicities accurate to @xmath103@xmath59 dex . the ionization conditions are determined by examining the relative absorption strengths of the different metal transitions . for the optically thin regime ,
the entire cloud is subject to the same ionizing spectrum ( there is no ionization structure ) and thus @xmath94 can be determined from two metal transitions independent of metallicity .
this point is illustrated in fig .
[ fig : cloudyill ] , which shows two grids of cloudy models with different metallicities , @xmath104 solar and solar . to construct the grids ,
the neutral hydrogen column density was fixed at @xmath105 , and the ionization parameter was varied from @xmath106 to @xmath107 .
the figure shows that the ratio of the column densities of various transitions , at a given @xmath94 , does not depend on the metallicity . for the allowed range of @xmath94 , metallicity
is then constrained by the partial lyman limit break and the transition . in this paper
, we have assumed a solar abundance for the optimized transition .
however , it is straightforward to consider how the metallicity derived would differ if we assumed a different abundance .
if the optimized transition has an abundance of @xmath108 dex relative to solar , then a model with a metallicity of @xmath109 would be correct .
of course , the abundances of other elements would have to be adjusted relative to the optimized transition based on whatever abundance pattern was assumed .
we present constraints on the gross properties of the @xmath33 absorber along the line of sight to pg @xmath1 .
we do not attempt to find a unique model to describe the data , however we consider what general conclusions can be drawn about the kinematics and multi - phase ionization structure of the absorbing gas .
we present examples of plausible models of the column densities , doppler parameters , ionization parameters , and metallicities in tables @xmath92 and @xmath3 .
we constrained the range of ionization parameters for subsystems a and b using detected metal transitions and limits . for subsystem
a , we consider two models : one in which the and arise in the same phase ( model 1 ) , and the other in which they arise in different phases ( model 2 ) .
the vp fits yield four clouds with column densities ranging from @xmath110 to @xmath111 and @xmath112 parameters from @xmath3 to @xmath113 ( see table @xmath92 ) . in model 1 we assume that the and arise in the same phase . to fully account for the absorption the four clouds would have @xmath94 values of @xmath114 , @xmath115 , @xmath114 , and @xmath114 , respectively .
also , the abundances of silicon and carbon must be decreased to make model 1 consistent with the data . with a solar abundance pattern , and these @xmath94 values , , , , , and would be overproduced relative to .
a @xmath116 dex decrease of silicon and carbon , relative to magnesium , resolves this discrepancy .
figures [ fig : model1a ] and [ fig : model1b ] present this model ( model 1 ) , with and without the abundance pattern adjustment , superimposed on the data . the small kinematic offset between and ( mentioned in
[ sec : system ] ) is apparent in the @xmath117 panel as well as in the expanded data in figure [ fig : offset ] .
this leads us to consider an alternative model , in which the and are produced by separate , kinematically offset clouds . in model 2
we assume that the absorption for subsystem a is produced entirely in a phase separate from the .
we again choose as the optimized transition for the lower ionization `` clouds '' , and use the vp fits to the as the starting point for our model .
however , now we also perform vp fits to the , obtaining four additional `` clouds '' , and use as the optimized transition for this separate phase . for the clouds , insignificant
is produced at the same velocity if @xmath118 . the ionization conditions for the clouds
can also be constrained by @xmath52 .
@xmath52 is not detected , providing an upper limit of @xmath119 .
@xmath120 limit of the equivalent width assuming a doppler parameter of @xmath4 .
the curve of growth of @xmath52 for this equivalent width is linear and therefore independent of doppler parameter .
the @xmath121 limit is derived at the position of each component , assuming an unresolved line at that velocity . ] in principle , this would provide a lower limit on @xmath94 .
however , below @xmath122 , @xmath123 depends only weakly on @xmath94 .
significant absorption is not produced , even in models with @xmath94 as low as @xmath124 .
as with model 1 , these clouds overproduce and over the full range of @xmath94 . a similar abundance pattern adjustment ( @xmath125 dex ) of carbon and silicon is required .
the properties for a sample version of model 2 for these clouds , with @xmath126 , near its maximum , are given in table 2 .
the contribution of these clouds to the absorption in other transitions ( without the @xmath127 dex abundance pattern adjustment ) is show as the dotted curve in figures [ fig : model2a ] and [ fig : model2b ] . to produce the absorption in model 2
, we assumed each individual component was shifted by a slightly different velocity ( given in table 2 ) .
these additional clouds , centered on the ( `` clouds '' ) , would have @xmath128 .
a lower @xmath94 would produce in these clouds as well ( thus overproducing in the combined and clouds ) and a higher @xmath94 produces , which is not detected at this velocity .
this model including the and clouds is superimposed as a long dashed line on the data in figures [ fig : model2a ] and [ fig : model2b ] , respectively .
the agreement with the observed @xmath117 profile is better than in model 1 ( see figure [ fig : model1b ] ) .
however , in both cases , the model profile , which is produced mainly by the clouds centered on , has a definite cloud structure which is not apparent in the data .
this indicates that neither model is fully consistent with the observed absorption and suggests that an additional phase is needed to fit the .
also , the strongest cloud overproduces at @xmath129 .
the ionization parameter can not be adjusted to fit both the and without overproducing higher ionization transitions , like .
if the ionization parameter is increased to fit these two transitions , oxygen would have to be deficient with respect to silicon .
we also considered a model in which both the clouds and clouds would contribute to the profile .
since the and clouds are `` staggered '' in velocity , they could then , in principle , blend out the structure within the and produce the `` flat - bottom '' profile seen in the data .
however , the combined contribution to is too large and is overproduced .
the ionization parameters can not be adjusted in such a way as to decrease the total contribution to without overproducing .
this alternative model may be excluded .
vp fits to the profile ( at @xmath130 ) yield three clouds ( `` clouds '' ) with a range of column densities from @xmath131 to @xmath132 and @xmath99 parameters from @xmath133 to @xmath134 ( see table @xmath92 ) .
@xmath87 and were not detected in this velocity range .
the limits on these transitions were used to constrain the ionization conditions of the clouds , at @xmath135 and @xmath136 .
the properties of the cloud at @xmath137 , which is required to fit the , are not well constrained because it is weak and not definitively detected in other transitions .
for this reason , we assume it has the same ionization parameter as the other clouds in subsystem b. a lower limit of @xmath138 is derived from the lack of @xmath87 absorption . an upper limit of @xmath139 applies in order that is not overproduced . a model with @xmath140 for the subsystem b clouds
is superimposed on the data in figures [ fig : model1a ] , [ fig : model1b ] , [ fig : model2a ] and [ fig : model2b ] . for the range of @xmath94 derived in sections [ sec : suba ] and [ sec : subb ]
, we consider the constraints on metallicity for subsystems a and b. since the column densities of the and transitions remain approximately constant from @xmath141 ( within @xmath127 dex : see figure [ fig : cloudyill ] ) , the metallicity constraint is only weakly dependent on the @xmath94 .
the metallicity of the subsystems is constrained by the partial lyman limit break . for simplicity , we assume that all the clouds in both subsystems , a and b , have the same metallicity . figure [ fig : lya]a shows synthesized spectra ( using model 2 for subsystem a ) , for four metallicities , expressed in solar units , ( @xmath142 ) superimposed on the observed partial lyman limit break .
we find that @xmath143 is clearly between @xmath144 and @xmath145 .
for @xmath146 , @xmath147 values would be too small to account for the observed partial lyman limit break .
although the break could arise from a separate phase , the metal transitions do not suggest such a phase .
a lower metallicity is not possible because a larger @xmath147 would leave too little flux beyond the break .
models 1 and 2 for subsystem a yield the same final result for metallicity , with a best fit value of @xmath148 . for the allowed range of metallicities
, the break would have significant contributions from several different clouds with @xmath149 to @xmath150 from both subsystem a and subsystem b. alternatively , one of these clouds could have a lower metallicity so that it would fully account for the partial lyman limit break . to compensate , the other clouds would then have higher metallicities .
figure [ fig : lya]b shows that the absorption produced by our models for the two subsystems , with @xmath148 does not fully account for the absorption in the wings of the seen in the data .
this implies that additional phases are needed , as will be described in
[ sec : broad ] . knowing the inferred ionization parameters and metallicities , the cloud sizes can be estimated .
the size of a cloud scales with the total hydrogen column density : @xmath151 , where @xmath152 , assuming a haardt - madau background ionizing spectrum .
therefore , the cloud size scales inversely with metallicity .
it is important to note that the `` size '' corresponds to the thickness of the slab assumed in the photoionization models .
realistically , a variety of geometries are possible . for subsystem
a , the two models that we have presented produce very different conclusions about the cloud sizes . in model 1 , a large ionization parameter ( @xmath153 )
is required in order that the be produced in the same phase with the .
large cloud sizes , ranging from @xmath133@xmath154 kpc , result for these four clouds .
in contrast , model 2 separates the total column density into two different phases , the and clouds , which have ionization parameters of @xmath155 and @xmath156 , respectively .
the clouds would be relatively small , with upper limits on their sizes of @xmath157@xmath158 pc , while the clouds would be larger , @xmath92@xmath26 kpc . for subsystem b
, the three clouds would have sizes of @xmath159@xmath160 pc .
the cloudy photoionization modeling process takes into account collisional ionization based on the temperature derived for a system in equilibrium .
however , a separate layer not in equilibrium , e.g. due to a shock , could exist within a system providing an additional heating effect .
in this additional layer , collisional ionization could dominate , giving rise to different absorption properties .
the combination of the subsystem a clouds ( either the four in model 1 or the eight in model 2 ) produces distinct cloud structure in the and does not adequately fit its smooth profile .
this discrepancy can be seen from the dotted curves in figures [ fig : model1b ] and [ fig : model2b ] .
photoionization models also do not account for the `` wings '' of the observed absorption .
a collisionally ionized phase , however , would naturally produce a broader , smoother profile of .
absorption peaks at a temperature of @xmath161 @xcite .
however , at this temperature , @xmath52 is overproduced .
a temperature of @xmath162 could produce without producing detectable and .
the metallicity of the layer would have to be @xmath163 in order not to overproduce .
the absorption contribution from a model of a collisionally ionized layer with these properties is given as a dashed line in figures [ fig : model1a ] , [ fig : model1b ] , [ fig : model2a ] and [ fig : model2b ] .
the need for a broad component or components is apparent from the underproduction of absorption by the narrow subsystem clouds ( see figure [ fig : lya]b and [ sec : metals ] ) . also ,
although two features are apparent , their velocity offsets ( @xmath19 for both ) from the subsystems makes it impossible that they are produced in the same phase of gas .
although is only covered at low resolution , it is clear that the blue part of the profile is not produced by the low
ionization components from the subsystems discussed in
[ sec : narrow ] .
the red part of the profile is fully produced by the subsystems , both for model 1 and model 2 of subsystem a ( see the solid curves in figures [ fig : model1b ] and [ fig : model2b ] ) , with model 1 producing only slightly more .
therefore , the considerations for the physical conditions of the broad components are similar for both models .
the discussion that follows pertains specifically to model 2 .
two higher ionization phase components can be added to simultaneously produce the needed and absorption .
that is , if these components are centered on the two features , it is also possible to fit the two extremes of the profile .
the fit to the absorption is uncertain , especially in the blue broad component .
there are apparent inconsistencies between the @xmath164 and @xmath165 transitions indicating that blends are present .
the data may also suggest that there is substructure and these phases may actually contain multiple clouds . in particular , there may be additional narrower components at @xmath166 and @xmath167 .
however , for simplicity , we model each broad component as one cloud and optimize on . from our fit
to , we determine that @xmath168 and @xmath169 for the blue broad component and @xmath170 and @xmath171 for the red broad component .
we consider collisional ionization and photoionization for these two broad components .
the measured @xmath99 parameter for hydrogen from the blue broad component ( @xmath172 ) implies a maximum temperature , @xmath173 , of @xmath174 .
this is too cool for this broad component to give rise to significant absorption through collisional ionization .
the measured @xmath99 for ( @xmath175 ) yields a larger maximum temperature of @xmath176 .
although this value does imply that could be produced by collisional ionization , there would then have to be another phase that accounts for the blue broad component of .
similarly , the @xmath99 parameter of hydrogen for the red broad component ( @xmath177 ) implies a maximum temperature of @xmath178 , which , again , is too low for to be produced through collisional ionization .
the @xmath99 parameter of ( @xmath179 ) allows for a maximum temperature of @xmath180 . at this temperature
, most of the oxygen would be in the form of .
if some of the broadening was due to turbulent motion , the actual temperature would be lower and significant could be produced through collisional ionization .
however , like the blue broad component , the red broad component of would require an additional phase if the were to be produced by this mechanism .
photoionization appears to be a simpler solution .
photoionized broad components can be added in such a way as to produce while simultaneously `` filling out '' the profile . for a given metallicity ,
if it is high enough , a @xmath94 can be chosen to simultaneously produce the observed and .
for this @xmath94 we can then consider whether the model is also consistent with the observed .
this requires substantial production in the blue broad component , but negligible production in the red broad component .
first , we consider whether the broad components could have a metallicity of @xmath148 , the same as the low ionization subsystems .
for the blue broad component a model with @xmath181 is consistent with the observed profiles , though there is some uncertainty because of the apparent inconsistency between @xmath182 and @xmath183 .
however , the at this velocity is underproduced , and in fact can not be produced at this metallicity for any ionization parameter .
although the could be produced in a different phase from the , for simplicity we favor a higher metallicity model which has the minimum number of phases . for the red broad component , with @xmath148 , the maximum @xmath184
is produced for @xmath185 , however this is not enough to match the observed profiles .
therefore , for both the blue and the red broad components , a larger metallicity is favored .
next , we consider @xmath186
. models with this metallicity and with various ionization parameters are superimposed on the high ionization transition profiles in the left panel of figure [ fig : broad ] . for the blue broad component ,
the model is consistent with the observed profiles for @xmath187 , depending on the contribution of blends to the two profiles .
however , to produce the observed absorption , a somewhat lower @xmath188 model is needed .
we conclude that either a model with this metallicity is not consistent with the observed blue broad component or that an abundance pattern adjustment of carbon relative to oxygen applies .
the is also strongly overproduced by this model , so it requires an abundance pattern adjustment of @xmath189 dex .
a model with @xmath190 and @xmath191 , that achieves the best agreement with and , produces a cloud with size @xmath154 kpc . for the red broad component
, a @xmath192 model produces an excellent fit to the profiles ( except for the apparent blend in the 1038 profile at @xmath193 ) .
as required , it does not contribute significantly to the absorption .
the cloud size for this model is quite large , @xmath194 kpc .
models with @xmath195 are also presented in figure [ fig : broad ] . with a @xmath195 model ,
a single @xmath196 value produces both the and in the blue broad component , without adjustment of the abundance pattern .
the observed and for the blue broad component can be produced by a model with @xmath195 and @xmath197 .
such a cloud has a size of @xmath198 kpc .
again , is overproduced , and a reduction of the nitrogen abundance by @xmath199 dex is required .
however , the red broad component is overproduced at this metallicity and an ionization parameter can not be chosen in order to fit both the and simultaneously . for @xmath107 , which provides the best fit to , but overproduces ,
the cloud would have a size of @xmath200 kpc . because it is only captured in the lower resolution fos data , the fit to the is quite uncertain . for some models
, it appears that the @xmath201 transition is overproduced , yet the @xmath202 transition is consistent . for models with @xmath203 , both the @xmath201 and the @xmath202 absorption
are overproduced in the red broad component .
one possible resolution to this problem is an abundance pattern adjustment in the red broad component , such that carbon is deficient with respect to oxygen . alternatively , since the clouds from subsystem a also contribute significantly to the at that velocity , a model in which carbon is deficient with respect to silicon in those clouds could also make a red broad component with @xmath195 consistent with the observed data .
another possibility is a scenario in which the broad components have different metallicities .
such a model is listed in table 3 , since it has the fewer abundance pattern adjustments than the other consistent models . in this model
, the blue broad component has a metallicity of @xmath195 , while the red broad component has a metallicity of @xmath186 and a @xmath204 . for all the modeling in the previous sections
, we assumed a pure haardt and madau extragalactic background photoionizing spectrum .
this is appropriate for absorption systems in regions where the extragalactic background radiation dominates over the ionizing flux from high mass stars .
unfortunately , the environment of the @xmath10 absorption system is unknown ( see [ sec : data ] ) . at @xmath11 ,
the extragalactic background radiation is a factor of @xmath205 times larger than at @xmath206 @xcite .
however , the ionizing flux close to an extreme starburst galaxy could dominate over the haardt and madau spectrum .
such a galaxy would have a photon flux of @xmath207 , of which @xmath208 escape @xcite .
the stellar flux would strongly dominate haardt and madau ( with a flux @xmath4 times larger at @xmath79 rydberg ) only within @xmath133 kpc of the center of an extreme starburst . since the absorption is relatively weak in the @xmath10 system
, we consider this unlikely . however , as an example , we explore the effects of starburst models on the ionization parameter and metallicity we infer for the clouds in model 2 for subsystem a. instantaneous starburst models with ages of @xmath104 gyr and @xmath209 gyr were combined with the extragalactic background .
starburst model spectra , which assume solar metallicity and a salpeter imf were taken from @xcite .
we consider extreme models with a strength of @xmath4 times that of haardt and madau at @xmath79 rydberg .
the @xmath104 gyr model spectral shape is characterized by the sharp edges of , and .
these edges influence the balance between the different ionization states of various metal line transitions . specifically , we find more and less and than in the pure haardt and madau case .
therefore , the inferred upper limit on the ionization parameter would be slightly higher and a more extreme abundance pattern adjustment would be needed for carbon .
also , the ionizing spectrum in the @xmath104 gyr model , has a relatively large number of photons above the edge . since these photons have the largest cross section for ionizing hydrogen , the clouds will be more ionized than in the pure haardt madau case .
therefore , the inferred metallicity is @xmath116 dex lower .
a sharp lyman edge characterizes the @xmath209 gyr model .
however , unlike the @xmath104 gyr model , the spectrum is rather flat over higher energies .
this flat spectrum is quite similar to the haardt madau spectrum and therefore there is little change in the ionization balance between the metal line transitions .
the inferred metallicity for this model would be slightly higher ( @xmath210 dex ) in this case .
this is due to a slightly softer ionizing spectrum for energies just above the lyman edge . from this investigation of the effect of spectral shape
we conclude that our general results based upon a pure haardt madau spectrum are likely to apply .
in the previous section , we gave a detailed description of how we constrained physical conditions for the multi - cloud , weak absorber at @xmath10 along the line of sight to pg @xmath1 .
here , we summarize those results , present an overview of a consistent model for the absorber , and compare to the results of @xcite , which were based on lower resolution spectra : 1 . two subsystems , a and b , detected in low ionization transitions , are separated by @xmath80 .
assuming , for simplicity , that each cloud in the subsystem has the same metallicity , then the observed partial lyman limit break is consistent with a metallicity of @xmath211 .
models with higher metallicities and dust depletion are not viable because carbon and silicon are depleted less than magnesium , and they are already overproduced by models with a solar abundance pattern .
subsystem a has clouds with an ionization parameter of @xmath212 , however , the could either be produced in those clouds ( model 1 ) or in additional lower ionization clouds , @xmath213 ( model 2 ) .
the ionization parameter of the clouds in subsystem b are slightly lower , @xmath214 .
2 . the issue of whether subsystem a , centered at @xmath78 , has such an additional low ionization phase hinges partially on the question of whether the and profiles are kinematically aligned .
figure [ fig : offset ] shows that , although the shape of the @xmath215 absorption profile is similar to the profile in subsystem a , it appears to be offset to the red by a few kilometers per second .
an offset of the same magnitude was observed for one of the clouds in the @xmath216 absorber along the line of sight to the hdf - south quasar , where , , and are offset from and by @xmath217 @xcite . @xcite
interpret this offset as the consequence of an region flow .
a similar offset between versus and may exist in the @xmath218 absorber toward 3c 273 @xcite .
the interpretation of the @xmath219 system is more complex because several clouds would be offset , but a similar explanation is possible .
3 . the cloud sizes may also provide insight into whether a one ( model 1 ) or two ( model 2 ) phase model of subsystem a is more plausible . if the and arise in the same phase , the four components of subsystem a have sizes ranging from @xmath133@xmath154 kpc .
it is difficult to conceive of a physical picture in which four such large structures would lie along the line of sight , within @xmath19 .
for a two phase model ( with the offset components ) , the cloud sizes of the components responsible for the are reduced to several kpcs , with smaller ( @xmath20@xmath158 pc ) clouds also contributing to the absorption .
this does seem more plausible in relationship to basic features of galactic interstellar gas .
although both the offset clouds and the clouds centered on contribute to the absorption profile , they do not account for its smooth , broad shape .
this may indicate that much of the is produced by collisional ionization at a temperature of @xmath220 .
two broad absorption profiles are observed , one @xmath19 to the red of subsystem a and the other @xmath19 to the blue of subsystem b. assuming the and arise in the same phase , the measured doppler parameter of associated with the places an upper limit on the temperature such that the could not be produced by collisional ionization .
both the and the , which could not be fully produced by the subsystems , can arise in the same photoionized phase . the low resolution profile , observed by fos / hst , also requires a contribution from the blue broad component .
these highly ionized clouds ( @xmath221 ) have metallicities larger than the low ionization subsystems .
the model cloud sizes range from 10 s to 100 s of kpc , depending on the metallicity , with larger metallicities leading to smaller sizes .
@xcite studied the @xmath10 system before the stis / hst spectrum was released , using the low resolution fos / hst spectra in conjunction with the same hires / keck spectra as were analyzed here .
now , using the higher resolution stis / hst uv data , we confirm most of the basic results of that paper , and we are able to resolve some ambiguities that were presented there . the metallicity constraint is confirmed , along with the presence of an additional high ionization cloud @xmath31 to the blue ( corresponding to the blue broad component in the present models ) . the metallicity agreement is not surprising , since in both cases the constraint came from hires / keck spectra covering in conjunction with the lyman limit break whose appearance is not sensitive to resolution .
the presence of an additional high ionization component is convincingly confirmed by the presence of absorption at a consistent velocity .
it seems likely that the seen in the fos spectrum and the seen in the stis spectrum would arise in the same phase . because of the limitations of the low resolution data , the @xcite work was unable to distinguish between two scenarios for the origin of the in the redward part of the profile .
based on those data , it could have arisen either in the same clouds with the or in a separate phase .
now , because the profile has narrow components resolved in the high resolution stis data , we can see that the arises either in the clouds ( model 1 ) or in the separate clouds ( model 2 . )
it does not arise in the red broad component which is responsible for producing and the red wing of .
the metallicities and ionization conditions of the blue broad component were poorly constrained by @xcite , because only and were available , and at low resolution .
they found that a model with @xmath222 and @xmath223 could be consistent with the data , but higher metallicities were also compatible , depending on the @xmath99 parameter assumed for the unresolved broad component .
they indicated that low ionization transitions might be observed at this same velocity .
now , we find that the detection of strong indicates a higher ionization parameter and a higher metallicity . also , although low ionization transitions are detected in the stis spectrum , their absorption profiles are offset by @xmath83 from the blue broad component .
the previous work therefore had the general idea that there was absorption blueward of the main low ionization system , but was able to discern little about the details .
from the derived physical constraints of the @xmath10 system along the line of sight to pg @xmath1 , we determined certain basic properties that this absorber must have .
there are two separate subsystems , implying that there are two distinct structures .
these subsystems have lower metallicity , lower ionization regions , and kinematically distinct , higher metallicity and higher ionization `` halos '' .
although these conclusions do not indicate a unique solution , they suggest scenarios and types of structures that might be responsible . _
the subsystems of low - ionization material in the @xmath10 system have a low metallicity ( @xmath224) _ although giant galaxies would be expected to have lower metallicities at higher redshifts than they do at present , a value as low as @xmath224 would be more extreme than our expectations .
almost all of the stars in the disk of the milky way formed well before redshift one ( coincident with the peak of the cosmic star formation ) , and nearly all of them have metallicities greater than @xmath209 solar , with values as large as solar being common @xcite .
it is consistent that the metallicities of the low ionization clouds composing strong absorbers at @xmath11 appear to have higher metallicities , from 1/10th solar up to the solar value @xcite .
low metallicities at @xmath11 are seen in damped absorbers @xcite , and this is attributed to their hosts having a mix of morphologies , commonly including dwarfs and low surface brightness galaxies @xcite . in local group dwarf galaxies at present , metallicities range from @xmath225 to @xmath226 @xcite .
the outer disks of spiral galaxies could also have low metallicities consistent with our derived value for the @xmath10 system .
the iron radial metallicity gradient in the milky way disk is @xmath227 dex kpc@xmath228 , out to a radius of @xmath229 kpc @xcite .
we would expect weak absorption ( as opposed to strong ) to arise only at galactocentric radii greater than @xmath5@xmath6 kpc .
an extrapolation of the observed metallicity gradient would therefore produce a low metallicity for radii at which weak absorption might be expected . _
the broad , higher ionization components in the @xmath10 system have a higher metallicity than the low ionization subsystems _ the subsystem absorption is produced in higher density regions than the broad higher ionization components . also , the individual subsystem clouds are narrower and smaller ( sizes of @xmath230 kpc ) than the high ionization clouds ( sizes of tens to hundreds of kpcs ) .
this suggests that the subsystems arise in a smaller region internal to a galaxy or structure , more likely to be close to the stellar sources of metals .
if so , at face value , it is puzzling why the more diffuse , highly ionized gas would have a higher metallicity than the concentrated regions .
one possibility would be a scenario in which relatively undiluted supernova ejecta are ejected from the central regions of a galaxy into a diffuse medium .
this is seen to some extent in the ejecta of nearby galactic winds , observed with chandra , although the metallicities of the wind material in these local systems are only somewhat larger than those of the interstellar material from which they originate @xcite .
another clue may result from the abundance pattern of the blue broad component .
if we tune its ionization parameter and metallicity so as to fit the and absorption ( along with ) , the absorption is severely overproduced .
a reduction of the abundance of nitrogen relative to carbon and oxygen would bring the model in accord with the data .
such a nitrogen deficiency ( about an order of magnitude ) is characteristic of the dwarf galaxies in the local group @xcite .
this pattern is consistent with primary production of nitrogen in a low metallicity environment @xcite . _
each of the two subsystems of the @xmath10 absorber has several narrow low ionization components spread over tens of , which may be related to a broad , high ionization component offset by @xmath19 _ the kinematic spread of a subsystem is reminiscent of that we would expect for a line of sight through a spiral disk with some increase due to the effects of interactions ( eg .
contribution from warps , tidal debris , or fountain material ) .
strong absorbers often show subsystems with kinematic spreads of tens of , though they have larger equivalent width and they also often have `` satellite clouds '' at larger velocities which contribute to a larger kinematic spread for the overall system @xcite . in the case of the strong absorbers , the kinematics of the strongest subsystem
have been demonstrated to be consistent with the expectations of thick disk kinematics @xcite .
a kinematic spread of tens of is also similar to the rotation velocities of the larger local group dwarf irregulars and to the velocity dispersions of local group dwarf spheroidals @xcite .
significant offsets between the low and high ionization gas are not uncommon in the nearby universe . for example
, offsets are typically @xmath231 between the absorption detected in the large magellanic cloud and its low ionization absorption , which is associated with its interstellar medium @xcite .
the lmc is consistent with an origin in coronal gas , being similar in strength and velocity dispersion to that associated with the milky way @xcite .
the observed kinematics for each of the @xmath10 subsystems could be due to a low ionization contribution from clumps in the disk or internal region of a dwarf , which would be surrounded by a broad phase corona or halo .
the measured doppler parameter of @xmath232@xmath5 is consistent with a dwarf galaxy halo dispersion @xcite .
the offset of @xmath83 of the broad component from the low
ionization gas could be due to a kinematic offset between central / disk and halo components of the dwarfs .
another possibility for the kinematic difference between the low and high ionization components in the @xmath10 absorber is a layer structure of the cone surrounding a breakthrough superbubble or wind . in the case of a starbursting dwarf ,
hot , x ray emitting gas is ejected at high velocity from the central region of the galaxy ( see , eg .
the edges of the cone surrounding the x - ray gas are cooler and are likely to have condensations that would give rise to lower ionization absorption .
in fact , strong absorption is detected in absorption against the majority of nearby starbursts @xcite .
we would expect `` transition layers '' in between the hot and cooler regions . in these regions ,
moderate to higher ionization absorption could occur .
the broad absorption observed in the @xmath10 system could trace such a transition layer in the interior . as the line of sight passes through the outer layers
, one might expect to see a group of clouds producing lower - ionization absorption that would be moving at a speed lower than that of the interior of the wind .
_ the low ionization absorption in the two @xmath10 absorber subsystems is quite weak .
each of them would be classified as a multiple cloud , weak absorber _ because of the weak , low ionization absorption , statistically , we would infer that neither of the two subsystems would be produced by a line of sight through the inner @xmath5@xmath6 kpc of a giant galaxy .
unfortunately , in the case of pg @xmath1 we have no direct information from imaging of the field ( as described in [ sec : data ] ) .
we can gain insight by considering the nature of the host galaxies or structures for nearby weak absorbers , for which imaging can reach down reasonably far on the dwarf galaxy luminosity function . @xcite
have proposed that the weak low ionization absorber at @xmath218 toward 3c 273 is produced in a thin shell around a @xmath233 `` post starburst '' galaxy that has been found at an impact parameter of @xmath234 kpc .
this is consistent with @xcite s speculations about that same system , and with the theoretical ideas of @xcite concerning the formation of condensations in dwarf galaxy winds .
@xcite have proposed that this is likely to be a common origin of weak , low ionization absorption systems .
there is another weak , low ionization absorber found at @xmath235 along the rxj @xmath236 sightline , which is adjacent to the 3c 273 sightline , at a separation of @xmath237 kpc @xcite .
however , in that case , the galaxy responsible has not yet been identified .
there is a candidate host dwarf galaxy known for the multiple cloud , low ionization absorber at @xmath238 toward pg @xmath239 , but its star formation history is unknown ( 2003 , j. stocke , private communication ) offsets between low to intermediate ionization transitions have been seen in other absorption systems , as was noted in
[ sec : summary ] .
the interpretation could be a layered structure , with different kinematics for the layers responsible for the different ionization stages .
in the case of subsystem a , the offsets in multiple components would require a series of such structures with a common orientation .
the presence of a collisionally ionized layer at an intermediate temperature ( @xmath240 k ) has also been suggested for other @xmath11 absorption systems @xcite .
this may also be related to shocks and is consistent with a complex , layered structure .
@xcite have shown that most of the baryons in the universe are trapped in a warm / hot phase traced by .
this has been interpreted as intragroup gas , however , the kinematic spread of the in some of the absorbers is not as large as would be expected for a poor galaxy group @xcite .
it is also possible that such arises in the gaseous halo of a single galaxy @xcite .
some of the stronger systems have associated weak lower ionization absorption @xcite .
it follows that there is some overlap between the stronger absorbers seen by @xcite and multiple cloud weak absorbers such as the one studied here .
more generally , the kinematic spreads of the broad components in the @xmath10 absorber are consistent with the velocity dispersions of the larger local group dwarf galaxies @xcite .
they are also consistent with the kinematics of the coronal gas around the milky way and the lmc @xcite , as mentioned above .
the would also arise naturally in a wind model , somewhere between the central x ray emitting gas and the cone in which the low ionization absorption would arise .
so far , we have discussed possible origins of the individual subsystems in the @xmath10 absorber ( and their possibly associated high ionization broad components ) , but it is important to consider how two such systems could arise .
the two subsystems are kinematically distinct , i.e. they are well separated in velocity space .
there are many scenarios that could give rise to absorption components separated by @xmath241 , but most can be categorized into one of two themes . in the first theme , there are two separate structures along the line of sight that exist within the same group or halo . in the second theme ,
a superbubble or wind is responsible for producing two kinematic structures that arise in the same object .
to evaluate the idea that two separate structures are responsible for the two subsystems , we must consider the likelihood of intercepting two structures along the same line of sight .
the velocity separation of @xmath242 between subsystems a and b is characteristic of galaxy group velocity dispersions @xcite .
the responsible structures might be outer disks of galaxies or dwarf galaxies , as discussed above .
there are many ways of considering this problem . for example
, if there are typically ten satellite galaxies clustered within @xmath243 kpc around a giant galaxy at @xmath244 , and if each satellite has a cross section for weak low ionization absorption of @xmath245 kpc@xmath246 , there would be a @xmath247 chance of a line of sight intercepting one if it intercepted another . in this scheme
, we would expect that most multiple cloud weak absorbers would not have two separate subsystems , i.e. we would expect that typically only one satellite would be intercepted .
this seems consistent with the data since , in a sample of @xmath27 multiple cloud , weak absorbers , only in this one case are there two separate subsystems @xcite .
if a superwind or bubble is responsible for the @xmath10 absorber , we would expect that each of the two low - ionization subsystems and its associated phase would represent an opposite side of a wind .
the two sides could plausibly be separated by @xmath248 @xcite .
there would have to be cold condensations with a fairly large covering factor on both sides of the wind in order to see two low ionization subsystems .
one could easily imagine other cases with asymmetric blowouts or stochastic structure such that only one side was detected or such that one side was only detected in high ionization transitions .
the very strongest absorbers ( which are common only at @xmath249 ) , could also have an origin in superwinds @xcite .
these absorbers have a characteristic saturated `` double trough '' absorption profile in , which breaks up into multiple clouds in other weaker low - ionization transitions . as with our two subsystems ,
the saturated troughs are typically separated by @xmath157@xmath243 .
strong low - ionization absorption could arise from clouds close to the point of origin of the wind or from early evolutionary stages of the superwind phenomenon .
as we reach larger distances or later stages when low ionization superwind clouds may have fragmented or dispersed , we might expect weaker absorption profiles like the ones we are seeing in the @xmath10 absorber toward pg @xmath1 .
if more multiple cloud , weak absorbers with two subsystems are observed it should be possible to test between an origin in two structures and an origin in a wind or outflow . if winds are responsible then we would expect to see broad , high ionization absorption to the blue of the blueward subsystem and to the red of the redward subsystem .
if two separate structures were responsible , the offset low and high ionization components would be due to kinematic differences between two components in such structures .
in such a case , the broad , high ionization absorption should arise at random to the red or to the blue of each of the subsystems , and sometimes would even be superimposed .
_ there are about two thirds as many multiple cloud weak absorbers as there are strong absorbers _ there must be a set of structures responsible for multiple cloud weak absorption with a cross section comparable to the set of @xmath69 kpc regions around all @xmath14 galaxies .
if these are outer disks around all @xmath250 galaxies , an annulus with thickness of @xmath27 kpc would be required .
@xcite found that it was plausible for the expected number of superwinds at redshift @xmath251 to account for the observed number of saturated `` double trough '' strong profiles .
similarly , we speculate whether the outer winds of a plausible number of bursting dwarfs could account for the observed number of multiple - cloud weak absorbers .
there are about 2/3 as many multiple cloud weak absorbers are there are strong absorbers @xcite .
strong absorption arises from regions around @xmath13 galaxies with sizes of @xmath252 kpc @xcite .
if the regions around starbursting dwarfs that give rise to multi - cloud weak absorption have sizes of @xmath253 kpc , then a number density @xmath133 times that of @xmath254 galaxies could account for all these multi - cloud weak absorbers .
realistically , it seems likely that there would be contributions to multiple cloud weak absorption from a variety of types of structures and processes .
conversely , all types of structures that contain gas must all make some contribution to the absorber population , contributing either strong or weak absorption and/or contributing to or absorption .
it is fairly well understood that @xmath14 galaxies produce most all of the observed strong absorption at @xmath11 .
but it is not well understood what type of absorption systems quiescent dwarf galaxies or dwarf galaxies with winds would produce .
above , we have considered the various properties of the @xmath10 absorber and discussed what types of structures and processes might be consistent .
clearly , there is no unique interpretation .
however , a model in which this absorber is related to dwarf galaxy winds is consistent with most of the listed properties .
it can explain the phase structure of the absorber , the relative metallicities of the phases , and the kinematics of the two subsystems relative to the broad , higher ionization components .
this model is also appealing in that it accounts for some of the absorption cross section presented by starbursting dwarfs , which were common at redshift one .
many more multiple cloud , weak absorbers must be studied in order to evaluate the relative contributions of dwarf galaxy winds , of quiescent dwarf galaxies , of outer galaxy disks , and of other phenomena . in conjunction with such a statistical study
, it will be essential to find analogs in the local universe so that the host galaxies and responsible processes can be directly identified .
@xmath256 & @xmath257 & @xmath257 & @xmath258 + @xmath259 & @xmath257 & @xmath257 & @xmath260 + @xmath261 & @xmath257 & @xmath257 & @xmath262 + @xmath263 & @xmath257 & @xmath257 & @xmath264 + @xmath265 & @xmath257 & @xmath257 & @xmath266 + @xmath77 & @xmath267 & @xmath257 & @xmath267 + @xmath87 & @xmath268 & @xmath269 & @xmath270 + @xmath52 & @xmath271 & @xmath271 & @xmath272 + @xmath55 & @xmath273 & @xmath274 & @xmath275 + @xmath74 & @xmath276 & @xmath277 & @xmath278 + @xmath75 & @xmath279 & @xmath280 & @xmath281 + @xmath215 & @xmath282 & @xmath283 & @xmath284 + @xmath285 & @xmath286 & @xmath273 & @xmath278 + @xmath76 & @xmath287 & @xmath287 & @xmath288 + @xmath164 & @xmath289 & @xmath290 & @xmath291 + @xmath165 & @xmath290 & @xmath292 & @xmath291 + @xmath293 & @xmath257 & @xmath257 & @xmath294 + @xmath295 & @xmath257 & @xmath257 & @xmath296 + -0.05 in + @xmath297 & @xmath298 & @xmath299 & @xmath300 & @xmath114 & @xmath301 & @xmath53 & @xmath302 & @xmath303 & @xmath304 & @xmath304 & @xmath305 & @xmath306 & @xmath307 + @xmath308 & @xmath309 & @xmath310 & @xmath311 & @xmath115 & @xmath301 & @xmath133 & @xmath312 & @xmath313 & @xmath314 & @xmath314 & @xmath315 & @xmath316 & @xmath317 + @xmath318 & @xmath319 & @xmath320 & @xmath321 & @xmath114 & @xmath301 & @xmath154 & @xmath322 & @xmath111 & @xmath323 & @xmath323 & @xmath324 & @xmath325 & @xmath326 + @xmath327 & @xmath328 & @xmath329 & @xmath150 & @xmath114 & @xmath301 & @xmath330 & @xmath331 & @xmath332 & @xmath132 & @xmath132 & @xmath333 & @xmath334 & @xmath335 + + @xmath297 & @xmath298 & @xmath328 & @xmath336 & @xmath337 & @xmath301 & @xmath338 & @xmath302 & @xmath339 & @xmath340 & @xmath341 & @xmath342 & @xmath343 & @xmath307 + @xmath308 & @xmath309 & @xmath344 & @xmath150 & @xmath337 & @xmath301 & @xmath59 & @xmath312 & @xmath303 & @xmath304 & @xmath305 & @xmath343 & @xmath313 & @xmath317 + @xmath318 & @xmath319 & @xmath345 & @xmath311 & @xmath337 & @xmath301 & @xmath346 & @xmath322 & @xmath347 & @xmath132 & @xmath315 & @xmath303 & @xmath341 & @xmath326 + @xmath327 & @xmath328 & @xmath348 & @xmath349 & @xmath337 & @xmath301 & @xmath209 & @xmath331 & @xmath350 & @xmath314 & @xmath351 & @xmath352 & @xmath351 & @xmath335 + @xmath297 & @xmath353 & @xmath354 & @xmath355 & @xmath114 & @xmath301 & @xmath92 & @xmath342 & @xmath343 & @xmath351 & @xmath333 & @xmath340 & @xmath314 & @xmath356 + @xmath308 & @xmath357 & @xmath345 & @xmath355 & @xmath114 & @xmath301 & @xmath92 & @xmath358 & @xmath343 & @xmath111 & @xmath333 & @xmath340 & @xmath132 & @xmath359 + @xmath318 & @xmath129 & @xmath360 & @xmath336 & @xmath114 & @xmath301 & @xmath26 & @xmath352 & @xmath332 & @xmath315 & @xmath132 & @xmath333 & @xmath361 & @xmath362 + @xmath327 & @xmath363 & @xmath364 & @xmath365 & @xmath114 & @xmath301 & @xmath3 & @xmath339 & @xmath347 & @xmath305 & @xmath324 & @xmath315 & @xmath323 & @xmath366 + + @xmath297 & @xmath367 & @xmath368 & @xmath369 & @xmath370 & @xmath301 & @xmath371 & @xmath372 & @xmath358 & @xmath341 & @xmath313 & @xmath352 & @xmath111 & @xmath373 + @xmath308 & @xmath374 & @xmath345 & @xmath375 & @xmath370 & @xmath301 & @xmath199 & @xmath303 & @xmath313 & @xmath314 & @xmath132 & @xmath340 & @xmath376 & @xmath377 + @xmath318 & @xmath378 & @xmath379 & @xmath355 & @xmath370 & @xmath301 & @xmath346 & @xmath339 & @xmath343 & @xmath333 & @xmath315 & @xmath313 & @xmath304 & @xmath380 + -0.05 in & @xmath381 & @xmath345 & @xmath382 & @xmath383 & @xmath384 & @xmath198 & @xmath385 & @xmath376 & @xmath386 & @xmath172 + & @xmath387 & @xmath388 & @xmath389 & @xmath384 & @xmath390 & @xmath194 & @xmath333 & @xmath132 & @xmath391 & @xmath177 + -0.05 in | we analyze high resolution spectra of a multi cloud weak [ defined as @xmath0 ] absorbing system along the line of sight to pg @xmath1 .
this system gives rise to a partial lyman limit break and absorption in , , , , , , and .
the lower ionization transitions arise in two kinematic subsystems with a separation of @xmath2 .
each subsystem is resolved into several narrow components , having doppler widths of @xmath3@xmath4 . for both subsystems ,
the absorption arises in a separate higher ionization phase , in regions dominated by bulk motions in the range of @xmath5@xmath6 .
the two absorption profiles are kinematically offset by @xmath7 with respect to each of the two lower ionization subsystem . in the stronger subsystem ,
the absorption is strong with a distinctive , smooth profile shape and may partially arise in shock heated gas .
moreover , the kinematic substructure of traces that of the lower ionization , but may be offset by @xmath8 . based upon photoionization models , constrained by the partial lyman limit break , we infer a low metallicity of @xmath9 solar for the low ionization gas in both subsystems .
the broader phases have a somewhat higher metallicity , and they are consistent with photoionization ; the profiles are not broad enough to imply production of through collisional ionization .
various models , including outer disks , dwarf galaxies , and superwinds , are discussed to account for the phase structure , metallicity , and kinematics of this absorption system .
we favor an interpretation in which the two subsystems are produced by condensed clouds far out in the opposite extremes of a multi layer dwarf galaxy superwind . |
the ultraviolet ( uv ) continuum of galaxies traces young massive stars , and hence serves as one of the most commonly used star formation rate ( sfr ) indicators ( e.g. , kennicutt 1998 , hereafter k98 ) .
however , the severe dust attenuation in the uv significantly weakens its power for measuring the sfr . in nearby star - forming galaxies , the dust attenuation in
the uv can vary from zero to several magnitudes ( e.g. , buat et al . 2005 ; burgarella et al .
2005 ) . for near - uv ( nuv )
selected and far - infrared ( fir ) selected samples , buat et al .
( 2005 ) found that the median dust attenuations in the far - uv ( fuv ) are 1.1 mag and 2.9 mag respectively ( see also burgarella et al . 2005 ) , corresponding to 64% and 93% obscuration of the fuv light respectively .
corrections for dust attenuation therefore are essential before the uv emission is used to probe the sfr .
commonly two approaches are used to estimate the dust attenuation in the uv .
the power - law slope of the uv continuum @xmath5 ( @xmath6 , at wavelengths 1300 - 2600@xmath7 ; calzetti et al .
( 1994 ) ) was the first proposed uv dust attenuation indicator ( e.g. , kinney et al .
1993 ; calzetti et al .
1994 ; meurer et al .
1999 ) , and is based on the assumption that the intrinsic slope @xmath5 is only sensitive to the recent star formation history and any deviation from the intrinsic value is caused by dust attenuation ( e.g. , calzetti et al .
1994 ; meurer et al .
however , once similar analysis is extended from starburst galaxies to normal star - forming galaxies , @xmath5 will depend on the mean age of the dust - heating population , the dust / star geometry , and the dust properties ( e.g. , witt , thronson & capuano 1992 ; witt & gordon 2000 ; granato et al .
2000 ; burgarella et al .
2005 ) .
the other uv dust attenuation estimator is the ratio of ir to uv emission .
this method is built on an energetic budget consideration ( e.g. , meurer et al .
1999 ; buat et al .
according to the energy balance argument , all the starlight absorbed at uv and optical wavebands by the interstellar dust is re - emitted in the ir , so the combination of the ir luminosity and the observed uv luminosity should be able to probe the dust - free uv luminosity ( e.g. , wang & heckman 1996 ; heckman et al .
1998 ) , which then can be used in turn to estimate the dust attenuation by comparing to the observed uv luminosity .
compared to @xmath5 , the ir / uv ratio is a more reliable dust attenuation indicator because it is almost independent of the dust properties and the relative distribution of dust and stars ( e.g. , buat & xu 1996 ; gordon et al . 2000 ; witt & gordon 2000 ; cortese et al . 2008 ) .
however , the ir / uv ratio does depend on the age of the dust heating populations ( e.g. , cortese et al .
2008 ; b. d. johnson et al .
2011 , in preparation ) .
the non - ionizing uv emission is mainly from stars younger than a few hundred million years , whereas by contrast the ir emission can be from dust heated by stars over a wide age range , including not only young stars that emit a bulk of their bolometric luminosities in the uv but also evolved stars that radiate only a small fraction of their bolometric energies in the uv or even do not contribute to the uv emission at all . since the introduction of these two methods many efforts have been made to calibrate the relations between uv attenuation and the ir / uv ratio or @xmath5 ( or equivalently fuv - nuv color ) or the ir / uv versus @xmath5 relation ( e.g. meurer et al .
1999 ; buat et al .
1999 ; gordon et al .
2000 ; bell 2002 ; kong et al . 2004
; calzetti et al . 2005 ; seibert et al . 2005 ; cortese et al . 2006 ; boissier et al . 2007 ; gil de paz et al . 2007 ;
johnson et al . 2007a,2007b ; panuzzo et al . 2007 ; salim et al . 2007,2009 ; treyer et al . 2007 ; boquien et al . 2009 ; buat et al .
2010 ; takeuchi et al .
most of the calibrations are either purely based on theoretical models or based on comparisons of multi - wavelength photometry with stellar population synthesis models .
for example , meurer et al . ( 1999 ) obtained a relationship between fir / fuv and the attenuation at fuv ( @xmath8 ) using the stellar population synthesis models of leitherer & heckman ( 1995 ) and then they derived an @xmath8@xmath5 relation using this fir / fuv based @xmath8 for starburst galaxies .
buat et al .
( 2005 ) derived @xmath9ir / uv relations for star - forming galaxies at fuv and nuv wavebands based purely on stellar population synthesis modelling .
salim et al . ( 2007 ) estimated the uv attenuation by comparing the observed spectral energy distributions ( seds ) to modelled seds and then based on this uv attenuation they obtained @xmath8@xmath5 relations for star - forming galaxies selected spectroscopically or using nuv - r color .
other studies that use different samples or types of galaxies to derive prescriptions for estimating uv attenuation either from ir / uv or from @xmath5 ( or equivalently fuv - nuv color ) include burgarella et al .
2005 , seibert et al . 2005 and cortese et al . 2008 .
as discussed earlier , the ir / uv ratio and the fuv - nuv color are dependent on the age of the dust heating populations , so calibrations based on different samples , regardless of the methods adopted , should account for this age effect . the considerable inconsistencies between the above prescriptions ( see section [ subsec : comparisons ] ) imply that the age difference of the stellar populations of the samples used was largely taken into account , and these prescriptions have been applied widely in the literature to correct for uv attenuations ( e.g. , bell & kennicutt 2001 ; iglesias - pramo et al .
however , the reliability and accuracy of these corrections need to be evaluated using an independent set of attenuation estimates . in the literature
an independent attenuation estimate , the attenuation in the h@xmath0 line ( a@xmath10 ) as derived from the balmer decrement ratio , has also been used to estimate the attenuation in the uv . to achieve this ,
two methods have been employed .
one is based on the idea that the sfr derived from attenuation - corrected h@xmath0 luminosity matches that from attenuation - corrected uv luminosity ( e.g. , buat et al .
2002 ; treyer et al .
2007 ) . using this method ,
buat et al . ( 2002 ) explored the possibility of using the observed h@xmath0/uv ratio to estimate the uv attenuation .
treyer et al . (
2007 ) employed this method to estimate the uv attenuation and then built an a@xmath11 fuv - nuv relation .
the alternative approach is to obtain a scale factor between the attenuations in the h@xmath0 line and in the uv by assuming a dust attenuation law ( e.g. , buat et al . 2002 ) .
both methods are strongly model - dependent , however , which adds to systematic uncertainties in the results .
the uv attenuation derived using the first method , i.e. , by matching the uv and h@xmath0 derived sfrs , depends not only on the balmer - based attenuation estimates in the h@xmath0 line but also on the conversions of h@xmath0 and uv luminosities to sfrs . on the other hand , the second method rests heavily on the form of the assumed dust attenuation law . in this paper
we take advantage of the large sets of multi - wavelength observations of nearby galaxies , including uv , ir and optical spectra derived balmer decrement ratios , to calibrate ir / uv and fuv - nuv as uv attenuation estimators empirically in a self - consistent way .
the calibrations derived this way are purely empirical , and depend only weakly on the balmer - derived a@xmath12 .
this paper is the second of a series , in which we aim to use multi - wavelength measurements of nearby galaxies to derive attenuation corrections . in kennicutt
2009 ; hereafter paper i ) , we used the spitzer infrared nearby galaxies survey ( sings ; kennicutt et al .
2003 ) and an integrated spectral survey sample from moustakas & kennicutt ( 2006 ; hereafter mk06 ) to calibrate the combinations of the optical emission lines ( h@xmath0 and [ oii]@xmath133727 ) and the ir or radio continuum emission as attenuation - corrected sfr indicators .
we found that the dust attenuation corrected h@xmath0 luminosities , as derived from the balmer decrement ratios , can be closely traced by the combinations of the observed h@xmath0 emission - line luminosities with 24@xmath4 , total ir ( tir ; the bolometric luminosity over the wavelength range 3 - 1100@xmath4 ) , and even 8@xmath4 ir luminosities , with rms scatters @xmath14 0.1 dex . in this paper
we derive new calibrations for uv attenuation corrections as derived from the ir / uv ratio and the fuv - nuv color as described above , and test the consistency of the sfrs computed from ir / uv or fuv - nuv corrected uv luminosities to those from balmer - corrected h@xmath0 luminosities .
compared to fuv , nuv can be contributed by more evolved stars ( @xmath15 gyr ) .
therefore we mainly focus on fuv in this paper .
but we also explore the reliability of using ir / nuv as a dust attenuation estimator as nuv is sometimes used as a sfr indicator .
the remainder of this paper is organized as follows . in section 2
we describe the sample , and describe how the uv flux densities were obtained .
the calibrations of the relationships between the attenuation in the fuv and the tir to fuv ratio , the uv color are shown in section 3 . in section 4
we apply the calibrations derived in section 3 to two samples of nearby galaxies to test the reliability of using these attenuation - corrected fuv luminosities as sfr indicators .
second order effects of these attenuation - corrected fuv luminosities relative to the balmer - corrected h@xmath0 luminosities are also investigated in section 4 . at last
, we explore the possibility of using the combinations of fuv with monochromatic mir luminosities 25@xmath4 ( or 24@xmath4 ) and radio continuum to estimate dust attenuations in section 4 . in section 5 , the reliability of using nuv luminosity as a sfr indicator is discussed . in section 6
the limitations and the range of applicability of the use of the calibrations are discussed .
comparisons of our calibrations to those in the literature , to the attenuation curve of calzetti et al .
2000 and to results based on sfr matching method are made in section 6 as well .
finally we summarize our results in section 7 . for distance estimates ,
we assume @xmath16 with local flow corrections .
our galaxies are drawn from the integrated spectrophotometric survey by mk06 and the sings survey .
the survey information and the selection criteria were described in paper i. we briefly list them here for completeness .
the mk06 survey observed 417 galaxies , which were selected to cover the full range of optical spectral characteristics found in the local galaxies , using a long - slit drift - scanning technique .
the reduced spectrum was integrated over a rectangular aperture that includes most or all of the emission of the galaxy .
the sings survey was a multi - wavelength survey from the fuv to the fir and it was composed of 75 nearby galaxies that cover wide ranges in morphological type , luminosity , star formation rate and dust opacity . in this paper series
we aim to use attenuation - corrected h@xmath0 luminosities , as derived from balmer decrement ratios , to calibrate combinations of optical emission lines or uv continuum and the ir or radio continuum as dust attenuation corrected sfr indicators .
therefore several criteria were employed to ensure the quality of the data ( 1 ) galaxies without detectable h@xmath0 emission were excluded , ( 2 ) in order to obtain reliable measurements of the balmer decrement ratios , galaxies with low signal to noise ratio spectra at h@xmath5 ( s / n @xmath17 15 ) were removed , ( 3 ) galaxies classified as active galactic nuclei ( agns ) or agn - starburst composites ( c.f .
moustakas et al .
2006 ; kewley et al . 2001 ; kauffmann et al .
2003 ) were excluded , and ( 4 ) galaxies with lower than 3@xmath18 detections at 25@xmath4 were removed . given that we will calibrate the attenuation estimates in the uv from the tir / uv ratio and the fuv - nuv color , fuv and nuv photometric data are needed .
some of the galaxies used in paper i have not been observed by the galaxy evolution explorer ( galex ; martin et al .
2005 ) so they are not included in the following analyses .
similar to paper i , ir data and the attenuation corrected h@xmath0 luminosities from h@xmath0/h@xmath5 ratios are required .
the compilation of h@xmath0 fluxes , [ nii]/h@xmath0 , h@xmath0/h@xmath5 and ir data has been described in detail in that paper . as stated in paper i , most of the galaxies in the mk06 sample have not been observed with spitzer . to maximize the uniformity of the data , only the infrared astronomical satellite ( iras )
data were used to derive the tir luminosities on average . ] , which was estimated using a weighted sum of iras 25@xmath4 , 60@xmath4 and 100@xmath4 emission following dale & helou ( 2002 ; equation ( 5 ) of their paper ) .
we will discuss possible systematics introduced by using iras - based tir luminosities in section [ subsec : sys ] .
we also provide calibrations for the combinations of the monochromatic mir ( iras 25@xmath4 or spitzer 24@xmath4 ) luminosities and the observed uv luminosities in section [ sec : other ] .
this is useful when fir data are not available .
as shown in paper i , the iras 25@xmath4 luminosity ( @xmath19 ) and the spitzer / multiband imaging photometer for spitzer ( mips ) 24@xmath4 luminosity can be used interchangeably .
given that spitzer / mips 24@xmath4 photometry is more sensitive than iras we use those measurements in preference to iras for the sings galaxies in section [ sec : other ] .
spitzer / mips 24@xmath4 measurements were also performed for the central 20 @xmath20 20 regions of the sings galaxies in paper i , and they are also used in this paper ( section [ sec : other ] ) . for galaxies selected from the mk06 sample , we identified their counterparts in the uv from galex gr4 .
galex performs imaging observations at fuv and nuv bands , centered at 1528@xmath7 and 2271@xmath7 respectively , with a circular field of view of @xmath21 diameter ( martin et al . 2005 ) .
gr4 includes images taken from the all - sky imaging survey ( ais ) , medium imaging survey ( mis ) , nearby galaxies survey ( ngs ) , deep imaging survey ( dis ) and guest investigators survey ( gi ) . since artifacts appear
much more frequently at the edge of the field of view , only galaxies within a radius of @xmath22 from the field center were selected ( morrissey et al . 2007 ) .
because of the overlap in coverage of these surveys , sometimes there were more than one image available . in such cases , we used the one with the longest exposure time in the fuv .
after this process we were left with 98 galaxies ( 54 in ais , 4 in mis , 22 in ngs , 2 in dis and 16 in gi ) out of the 113 star - forming galaxies selected from the mk06 sample . but ngc1003 , which is in ais , was excluded later because of artifact contamination in its fuv image .
so the final sample consists of 97 objects , with 53 in the ais survey .
intensity ( -int ) and sky background ( -skybg ) images at fuv and nuv bands were retrieved from the multimission archive at space telescope science institute ( mast ) website ( morrissey et al . 2007 ) .
sky background was subtracted before performing photometry . in order to obtain the uv fluxes from the same region as h@xmath0
, the rectangular apertures used in the integrated spectra observations by mk06 were adopted in the uv photometry images for galaxies in both mk06 and sings samples to get a rough estimate of this fraction . by comparing the rectangular photometry to the global photometry which was adopted from dale et al .
( 2007 ) , we found that at most 10% of 24@xmath4 emission is from outside of the rectangular apertures .
this small fraction will not affect our results . ] .
foreground stars within the apertures , identified by comparing the nuv and the digitized sky survey ( dss ) optical images , were masked during the photometry .
the fuv and nuv magnitude zero points listed in morrissey et al .
( 2007 ) were used to convert intensities into magnitudes .
the high - resolution relative response images ( -rrhr ) from mast were used to estimate the uncertainties in the photometry .
the fuv and nuv flux densities were corrected for galactic extinction using the schlegel et al .
( 1998 ) dust map and the galactic extinction curve derived by cardelli et al .
( 1989 ) for a total - to - selective extinction of @xmath23 ( c.f .
, gil de paz et al . 2007 ) . specifically , @xmath24 and @xmath25 .
the galactic extinction corrected fuv and nuv flux densities of the mk06 sample are listed in table 1 .
the absolute calibration uncertainties in fuv ( 0.05 mag ) and nuv ( 0.03 mag ) are added in quadrature . since more than half of the mk06 sample galaxies ( 53 out of 97 ) were extracted from the shallow ais survey , with exposure times of 100s , 15 times shorter than the mis , we tested whether some of the extended uv emission was missed in ais images compared to observations with deeper exposures .
there are 43 objects in common between the ais and the other deeper surveys .
we compared their uv magnitudes measured from the ais images and those from observations with longer exposure time in figure [ comparelongexpaisflux.eps ] .
it is clear that the uv measurements based on the ais images are in excellent agreement , within @xmath1420% , with those from deeper exposures , which means that the uv fluxes measured from the ais images are not systematically under - estimated within the adopted rectangular apertures .
consequently they can be used along with the measures from deeper exposures without introducing any systematic errors .
uv flux densities for the sings sample were obtained from dale et al .
those authors calculated the fuv and nuv flux densities from the surface brightness profiles derived for the galex atlas of nearby galaxies ( gil de paz et al .
2007 ) . since dale et al .
( 2007 ) used li & draine ( 2001 ) reddening curve to correct for the galactic extinction , we redid the galactic extinction using the same recipes as those used for the mk06 sample for consistency . applying
the requirement of having both iras data and uv flux measurements left us with 36 objects for the sings sample measurements are used in preference to iras 25@xmath4 for the combination of fuv and 25@xmath4 ( or 24@xmath4 ) for the sings galaxies in section [ sec : other ] , all star - forming galaxies in the sings sample ( see paper i ) with galex observations are included .
47 objects meet this requirement . ] . to avoid small number statistics and the less reliable measurements of h@xmath0/h@xmath5 ratios for some of the sings objects
( see paper i ) , we do not use these objects to calibrate the relations in the following analyses .
however we include them in the comparison to see how they behave .
we also measure the fuv and nuv flux densities for the central 20 @xmath20 20 regions of the sings galaxies .
it should be noted that the large absolute calibration uncertainties ( 0.15 mag ) in gil de paz et al .
are included in the quoted errors in fuv and nuv for sings sample .
as addressed in section [ sec : intro ] , prescriptions of using tir / fuv as a dust attenuation indicator are usually derived under various model assumptions . in this section
, we employ a new method to derive such prescriptions empirically and avoid most of the systematic model dependencies .
our method is based on the statistical correlation between the fuv attenuation estimated from the tir / fuv ratio and that as derived from the fuv - nuv color . by deriving the prescriptions based on a sample of nearby star - forming galaxies , we virtually assume an average stellar population and dust attenuation curve of a certain type of galaxies .
therefore our calibrations are probably not applicable for other types of galaxies ( see section [ sec : uncertainties ] ) .
we adopt the convention of infrared excess irx from meurer et al .
( 1995 ) @xmath26,\ ] ] where the observed fuv luminosity takes the common definition of monochromatic luminosity @xmath27 .
according to an energy balance argument , the dust attenuation in the fuv , @xmath28 , can be estimated from the irx using the following form @xmath29 where @xmath30 is a scale parameter , whose meaning is explained below .
this approach is similar to that presented in paper i for the combination of the observed h@xmath0 luminosity and the tir luminosity .
as stated in paper i , such an expression is a simple approximation to a much more complicated radiative transfer process . to better understand the underlying physical foundations of eq .
( [ eq : uvextirx ] ) , we adopt the relevant equations from paper i and adapt them to the case of fuv .
we first define a scale parameter @xmath31 , which is the fraction of the bolometric luminosity that is emitted in the fuv , i.e. , the inverse of the bolometric correction @xmath32 where @xmath33 represents the intrinsic ( extinction - free ) fuv luminosity and @xmath34 denotes the bolometric luminosity . under the assumption of a foreground dust screen approximation
, the observed luminosity can be expressed as @xmath35 where @xmath36 is an effective optical depth in the fuv , as defined by equation ( [ eq : tau ] ) , and the corresponding attenuation in magnitudes @xmath37 .
the dust attenuation corrected fuv luminosity is a sum of the observed luminosity and the dust - attenuated luminosity @xmath38 considering the energy conservation , all the starlight reprocessed by dust is re - emitted in the ir .
so the tir luminosity can be expressed as @xmath39 where @xmath40 is the effective opacity for all of the dust - heating starlight , and is defined as @xmath41 this effective mean opacity depends on the star formation history of the galaxy , the dust extinction curve and the relative distribution of stars and dust .
we substitute eqs .
( [ eq : eta ] ) and ( [ eq : bolometric ] ) into eq .
( [ eq : attenfuv ] ) to eliminate @xmath34 and express @xmath33 using the observable luminosities @xmath42 and l(tir ) and the opacities . after this exercise ,
the dust attenuation estimate in the fuv becomes @xmath43 .
\label{eq : uvextbol}\ ] ] by comparing eq .
( [ eq : uvextirx ] ) with eq .
( [ eq : uvextbol ] ) , it can be seen that the coefficient @xmath30 in eq .
( [ eq : uvextirx ] ) is the approximation of the product of the inverse of the bolometric correction @xmath44 and the factor @xmath45 . when the fuv - nuv color is used to estimate the fuv attenuation , a linear relation between @xmath46 and the fuv - nuv color
is usually taken @xmath47 where the slope @xmath48 and the intercept @xmath49 are usually determined by a linear regression to the observations .
the physical foundations of this relation can be understood as follows .
the color excess in fuv - nuv , e(fuv - nuv ) is defined as @xmath50 where @xmath51 and @xmath52 denote the observed and the attenuation - corrected fuv - nuv colors .
e(fuv - nuv ) relates to the color excess e(b - v ) by the relation @xmath53 where @xmath54 represents an attenuation curve .
so eq . ( [ eq : tmp ] ) has the modified form @xmath55 .
\label{eq : modtmp}\ ] ] by comparing this modified equation to the original eq .
( [ eq : tmp ] ) , we can see that @xmath56 , which is the slope of the uv part of the attenuation curve , and @xmath57 is equal to the dust - free uv color , @xmath52 .
since the estimated @xmath46 from irx and fuv - nuv should agree , the predicted relation between irx and fuv - nuv color can be obtained by equating eq .
( [ eq : uvextirx ] ) and eq .
( [ eq : modtmp ] ) . given that @xmath58 and @xmath59 can not be derived separately from the fitting , @xmath60 is still used in the irx fuv - nuv relation @xmath61}}-1]-\log(a_{\rm fuv } ) .
\label{eq : irxbeta}\ ] ] before we proceed to derive the parameters @xmath30 , @xmath62 and @xmath48 in eq .
( [ eq : irxbeta ] ) from the observations , we discuss the implicit assumptions in our method .
the comparison between eq .
( [ eq : uvextirx ] ) and eq .
( [ eq : uvextbol ] ) shows that the coefficient @xmath30 is the approximation of the product of the inverse of the bolometric correction @xmath44 and the factor @xmath63 . by deriving single values for @xmath30 , @xmath52 and @xmath48
, we virtually study the average properties of this sample , without taking into account variations in star formation histories and dust attenuation curves from galaxy to galaxy , which contribute to the scatters in the observed correlations . in order to minimize the number of free parameters in the fitting
, we first estimate the intrinsic ( i.e. , dust - free ) fuv - nuv color empirically , which is achieved by the correlation between the observed fuv - nuv color and the attenuation in the h@xmath0 line .
this method is similar to calzetti et al .
( 1994 ) in which the unattenuated power - law slope of the uv continuum @xmath64 was estimated using the correlation between @xmath5 and the balmer decrement ratio .
this correlation between the dust reddening in the stellar continuum and in the gas content ( see also calzetti 1997 and figure [ uvcolor.aha.eps ] below ) implies that the fuv attenuation in the stellar continuum is proportional to the attenuation in the h@xmath0 line .
the proportional factor is dependent on the extinction curve and the relative distribution of stars , gas and dust , which may vary from galaxy to galaxy .
unfortunately , we can not study this property on a galaxy - to - galaxy basis based on current data , so only an average relation is obtained ( see section [ subsec : comdustlaw ] ) .
the fuv - nuv color versus @xmath65 relation is shown in figure [ uvcolor.aha.eps ] . throughout this paper
the mk06 and sings objects are represented by solid and open circles respectively . from this figure
, we can see that the mk06 and the sings samples tell us different stories .
the mk06 sample shows that @xmath65 is correlated with fuv - nuv color , whereas the sings sample shows almost no correlation between these two quantities after the two reddest objects are excluded .
this is not unexpected given the large calibration uncertainties of uv photometry and the way we compiled the balmer decrement ratios for the sings galaxies , as described in section [ sec : sample ] and paper i. therefore only the mk06 sample is used to derive the dust - free fuv - nuv color , as indicated in section [ sec : sample ] . since the scatter is probably intrinsic , which means that the measurement errors are not the main contributors to the scatter about the regression line , an ordinary least square bisector fit method is adopted ( isobe et al .
1990 ; feigelson & babu 1992 ) .
the fitting to the mk06 sample that is plotted as a solid line in figure [ uvcolor.aha.eps ] gives @xmath66 this indicates an intrinsic fuv - nuv color for zero attenuation of @xmath67 mag .
the errors quoted here in the slope and intercept were yielded by the fitting routine .
the dust - free fuv - nuv color derived here is in excellent agreement with the value @xmath68 mag derived by gil de paz et al .
those authors used the prescriptions given by buat et al .
( 2005 ) to correct for the dust attenuation in the fuv and nuv using the tir / fuv and tir / nuv ratios , and found that the intrinsic fuv - nuv color for their sample is @xmath68 mag on average . to understand the average stellar population of our sample galaxies , we compare the intrinsic fuv - nuv color we derived here with the stellar population synthesis predictions . by comparing to models constructed using galaxev by bruzual & charlot ( 2003 )
, we find that our galaxies are @xmath14 0.02 , 0.03 and 0.04 mag redder in fuv - nuv color than a 10 gyr old stellar population with an e - folding timescale of 5 gyr , 10 gyr and a constant star formation history , respectively .
for a constant star formation history starburst99 ( leitherer et al . 1999 ; vazques & leitherer 2005 ) produces consistent result ( see section [ subsec : sys ] ) . after fixing the term @xmath69 in eq .
( [ eq : irxbeta ] ) , the coefficients @xmath48 and @xmath30 can be obtained by fitting eq .
( [ eq : irxbeta ] ) to the data .
figure [ irxfuv.uvcolor.eps ] shows the irx versus fuv - nuv color relation , with the corresponding @xmath5 labelled on the top axis . the relation between fuv - nuv and @xmath5
was derived using the definition given by kong et al .
( 2004 ) for our adopted central wavelengths for fuv and nuv ( see section [ sec : sample ] ) , which has the form @xmath70 . in order to have a clearer view of the average properties of the data , we overplot the irx median , lower ( 25% ) and upper ( 75% ) quartiles in bins of width 0.2 mag in fuv - nuv color as open squares with error bars in figure [ irxfuv.uvcolor.eps ] .
the solid line represents the chi - square minimization fit to the mk06 sample with the deviation in the irx minimized , which has the form defined by eq .
( [ eq : irxbeta ] ) with @xmath48 and @xmath30 equal to @xmath71 and @xmath72 respectively . in order to test the effect of the adopted @xmath52 on the fitting results , we varied @xmath52 by @xmath73 about the best - fitted value , i.e. , assumed @xmath74 and @xmath75 .
it turned out that the changes in the best fitted @xmath48 and @xmath30 caused by these new assumptions of the @xmath62 are within @xmath73 , and the resulting irx versus fuv - nuv color relations can not be distinguished from that using @xmath76 over the ranges of fuv - nuv color and irx covered by our sample galaxies ( see section [ subsec : limitations ] ) . substituting the best fitted coefficients into eqs .
( [ eq : uvextirx ] ) and ( [ eq : modtmp ] ) , the attenuation estimates based on the irx and the fuv - nuv color become @xmath77 , \label{eq : irxextfinal}\ ] ] and @xmath78 .
\label{eq : fncextfinal}\ ] ]
in order to evaluate the reliability of the dust attenuation corrected fuv luminosities derived using the above calibrations as sfr indicators , we compare the irx and fuv - nuv color corrected fuv luminosities ( according to eq .
( [ eq : irxextfinal ] ) and ( [ eq : fncextfinal ] ) ) with the balmer decrement ratio corrected h@xmath0 luminosities and investigate possible second - order trends in the residuals as functions of various properties of the galaxies in this section
. combinations of fuv with monochromatic 25@xmath4 ( or 24@xmath4 ) infrared and radio continuum as dust attenuation measures are also explored .
we note that the attenuation calibration methods we employed in section [ sec : calibration ] make only minimal use of the balmer decrement data .
so when we compare the dust - corrected fuv luminosities based on our calibrations with the balmer - corrected h@xmath0 luminosities , we are not making a circular argument .
before we proceed to compare the dust attenuation corrected fuv luminosities with the h@xmath0 luminosities , it is interesting to see the behavior of the uncorrected fuv luminosities as a function of the balmer decrement corrected h@xmath0 luminosities in figure [ lfuvobs.lhacor.eps ] , which can help us understand the importance of the attenuation corrections for this sample .
the straight lines in figure [ lfuvobs.lhacor.eps ] are model predictions for matched sfrs as described below .
figure [ lfuvobs.lhacor.eps ] shows that the uncorrected fuv luminosities can underestimate the sfrs by more than an order of magnitude , which varies with the intensity of the star formation activity .
galaxies with different sfrs are not equally attenuated , with more active galaxies suffering from higher attenuations .
this is consistent with studies in the literature ( e.g. , wang & heckman 1996 ; treyer et al .
2007 ) . in the following part of this section
, the reader will see how much will be improved with our dust attenuation corrections .
we compare the irx - corrected fuv luminosities with the balmer decrement ratio corrected h@xmath0 luminosities in the left panel of figure [ comlumiirx.his.eps ] .
it can be clearly seen that the irx - corrected fuv luminosities correlate tightly and linearly with the balmer - corrected h@xmath0 luminosities with a rms scatter of @xmath79 dex . in order to quantify the degree of consistency between the sfrs derived from the attenuation - corrected fuv and h@xmath0 luminosities , in the left panel of figure [ comlumiirx.his.eps ] we overplot the predicted relations from different sfr prescriptions .
the red short - long dashed line represents the prediction from the widely used k98 s sfr prescriptions .
the calibrations in k98 assume a salpeter initial mass function ( imf ; salpeter 1955 ) with mass limits 0.1 - 100@xmath80 , and 1990 s generation stellar evolution models ( see kennicutt et al .
1994 ; madau et al .
1998 ) . to compare with the k98 s prediction ,
we first construct a model under the same assumptions as k98 did , i.e. , salpeter imf , constant star formation history lasting for 100 myr and solar metallicity ( the magenta short dashed line ) , using version 5.1 of the stellar population synthesis model starburst99 ( leitherer et al .
1999 ; vazquez & leitherer 2005 ) , which uses the state - of - the - art stellar evolutionary models .
star - forming galaxies in our sample probably have experienced continuous star formation for at least 1 gyr . in order to examine age effects ,
we build models with a constant star formation history lasting for 1 gyr ( represented by long dashed lines ) .
it is worth noting that for a constant star formation history the intrinsic fuv to h@xmath0 luminosity ratio stays constant after 1 gyr .
therefore , the model predictions we derive by assuming a constant sfr for 1 gyr are the same as those under the assumption of a constant sfr for longer time .
although we assume constant star formation histories here , models with exponentially declining star formation histories do not give significantly different results , as long as the sfr does not drop too rapidly .
for example , for an e - folding timescale of 5 gyr , the fuv to h@xmath0 ratio differs from the constant star formation history by @xmath81% at any given age , according to our experiments based on stellar population synthesis models constructed using galaxev by bruzual & charlot ( 2003 ) .
the imf also affects the fuv / h@xmath0 ratio .
so apart from models with salpeter imf , models with a more realistic kroupa imf ( kroupa & weidner 2003 ) with mass limits of 0.1 - 100@xmath80 are also explored ( represented by lines in blue ) .
the fuv and h@xmath0 luminosities and @xmath82 , which connect luminosities and their corresponding sfrs by the relation @xmath83 . ] and their ratios are listed in table 2 . in the left panel of figure [ comlumiirx.his.eps ]
the different models ( represented by the colored lines ) are difficult to separate , because of the large dynamic range in luminosities .
for that reason we show in the right panel of figure [ comlumiirx.his.eps ] histograms of the attenuation - corrected fuv to h@xmath0 luminosity ratio , using a logarithmic scale .
the mk06 and sings samples are represented by the solid and dotted histograms respectively .
the straight lines in color are the same model predictions as those shown in the left panel of this figure .
the median and mean values of the irx - corrected fuv to h@xmath0 luminosity ratios for the mk06 sample are 123.88 and 127.94 , which are in excellent agreement with the model predictions for kroupa imf at 100 myr and salpeter imf at 100 myr respectively ( see table 2 ) .
however , the fuv / h@xmath0 ratio only increases by 8% and 9% respectively for the same imfs but older age ( @xmath84 gyr ) .
the correlation between the irx - corrected fuv luminosities and the balmer - corrected h@xmath0 luminosities is so tight and linear that it is useful to test for any second order trends in the residuals as functions of various galactic properties . in figure [ fuvirxresi.para1.eps ]
, we plot the ratios of the irx - corrected fuv luminosities to the balmer - corrected h@xmath0 luminosities against the balmer - corrected h@xmath0 luminosities ( i.e. , the dust corrected sfrs ; top - left panel ) , the balmer - corrected h@xmath0 luminosity densities ( top - right panel ) , the attenuations in the h@xmath0 line ( bottom - left ) and the axial ratios b / a ( i.e. , inclinations ; bottom - right ) . in figure [ fuvirxresi.para2.eps ] , the optical spectral features h@xmath0 equivalent width ew(h@xmath0 ) ( top - left panel ) , 4000 break @xmath85 ( top - right panel ) , gas - phase oxygen abundance 12+log(o / h ) ( bottom - left panel ) and the dust temperature indicator far - infrared 60@xmath4 to 100@xmath4 flux ratio @xmath86 ( bottom - right ) are explored .
given that h@xmath0 line and continuum emission are sfr and stellar mass tracers respectively , the integrated h@xmath0 equivalent width is a measure of specific sfr ( kennicutt et al .
the @xmath85 is often used as a reddening - insensitive rough star formation history indicator of the stellar populations ( e.g. , bruzual 1983 ; balogh et al .
1999 ; kauffmann et al .
2003 ) and it correlates with the specific sfr ( brinchmann et al .
2004 ) . in all panels
the dashed line indicates the predicted fuv to h@xmath0 luminosity ratio for a kroupa imf at age 100 myr , as shown in figure [ comlumiirx.his.eps ] .
as can be seen from figure [ fuvirxresi.para1.eps ] and [ fuvirxresi.para2.eps ] , the ratios of the irx - corrected fuv to the attenuation - corrected h@xmath0 luminosities show no or at most marginal trends with all the parameters investigated .
the rms scatters are about @xmath79 dex , which is even smaller than the average uncertainty in @xmath87}$ ] ratio , @xmath88 dex , with the error introduced by using eq .
( [ eq : irxextfinal ] ) to correct for fuv attenuation included .
this demonstrates the reliability of using the combination of the fuv and tir luminosities as a sfr indicator .
the absence of correlations implies that over the ranges of the galactic properties probed by our galaxies , either no trend exists in the dust - free fuv / h@xmath0 ratios and the irx method with the parameters tested above , or any trend in the dust - free fuv / h@xmath0 ratios is compensated by that in the irx method .
whatever it is , the sfr estimated from the irx - corrected fuv luminosity agrees well with that derived from the balmer - corrected h@xmath0 luminosity , and the ratio of these dust - corrected sfrs is not a function of the galactic properties .
therefore the combined fuv and tir luminosities can serve as a very good attenuation - corrected sfr indicator over the ranges of physical properties covered by our sample objects . from figure [ comlumiirx.his.eps ] to figure [ fuvirxresi.para2.eps ] , we note that the sings galaxies deviate from the mk06 galaxies systematically .
the systematic deviations can be caused either by overestimates of the tir luminosities or by underestimates of the attenuation in h@xmath0 lines .
as we showed in paper i ( figure 2 in that paper ; see also dale et al .
2009 ) , for cold galaxies ( @xmath89 ) , the tir luminosities measured by iras bands are systematically underestimated with respect to the more sensitive and accurate spitzer / mips bands . therefore using tir luminosities based on spitzer / mips measurements for the sings galaxies would make the sings galaxies deviated more .
furthermore , we did not see a correlation between the deviations and the @xmath86 colors in the bottom - right panel of figure [ fuvirxresi.para2.eps ] .
another piece of evidence is that the same deviations show up in the comparison of the combined fuv+25@xmath4 luminosities with the balmer - corrected h@xmath0 luminosities ( section [ sec : other ] ) , and the monochromatic 25@xmath4 luminosity measured by iras does not suffer from the same problem as the tir luminosity luminosities are in well agreement with the spitzer / mips 24@xmath4 luminosities . ] .
therefore , the systematic deviations of the sings galaxies can not result from overestimates of the tir luminosities by iras .
instead , they may be due to underestimates of the attenuation in h@xmath0 lines , caused by the noisier h@xmath5 detections .
similar to the analyses we did for the irx - corrected fuv luminosities in section [ subsec : uvtir ] , we compare the fuv - nuv color corrected fuv luminosities with the balmer decrement ratio corrected h@xmath0 luminosities in this subsection . figure [ comlumifnc.his.eps ] shows the fuv - nuv corrected fuv luminosities as a function of the balmer - corrected h@xmath0 luminosities ( left panel ) and the histogram of the fuv - nuv corrected fuv to the balmer - corrected h@xmath0 luminosity ratio ( right panel ) .
compared to the irx - corrected fuv luminosities versus the balmer - corrected h@xmath0 luminosities relation , the correlation between the fuv - nuv corrected fuv luminosities and the balmer - corrected h@xmath0 luminosities is non - linear and has a much larger rms scatter of @xmath90 dex .
figure [ comlumifnc.his.eps ] shows that the relation between the fuv - nuv color corrected fuv luminosities and the balmer - corrected h@xmath0 luminosities is not completely linear .
the slope of the relation is less than unity , which implies that for galaxies forming stars vigorously ( i.e. , with high sfrs ) , the fuv - nuv corrected fuv luminosities will under - estimate the sfrs . to better understand the applicability and limitation of the fuv - nuv corrected fuv luminosities as sfr measures ,
apart from the balmer - corrected h@xmath0 luminosities , we also examined the ratios of the uv color corrected fuv luminosities and the balmer - corrected h@xmath0 luminosities as functions of several other galactic properties , as we did for the irx - corrected fuv luminosities .
we found that the residuals of the uv color corrected fuv luminosities relative to the balmer - corrected h@xmath0 luminosities luminosities as references , we also examined the trends of the residuals of the uv color corrected fuv luminosities relative to the irx - corrected fuv luminosities as functions of the above parameters .
we found that these residuals behave similarly to the residuals of the uv color corrected fuv luminosities with respect to the balmer - corrected h@xmath0 luminosities .
] show significant negative correlations with several of the parameters , including the balmer - corrected h@xmath0 luminosity densities , the attenuations in the h@xmath0 lines , the fir color and the ew(h@xmath0 ) , as shown in figure [ fuvuvcolorresi.para1.eps ] and [ fuvuvcolorresi.para2.eps ] .
the negative trends with the fir color and the h@xmath0 equivalent width are similar to what kong et al .
( 2004 ) found .
those authors found correlations between the deviation from the average irx-@xmath5 relation defined by starburst galaxies and the above two parameters , and they explained these correlations as direct observational evidence for the dependence of the scatter of the irx-@xmath5 relation on star formation history .
galaxies that are forming stars more actively in the current epoch than in the past are intrinsically bluer than those with lower current to past sfr ratios .
so the attenuation in those galaxies with warmer fir color , larger ew(h@xmath0 ) and hence bluer intrinsic fuv - nuv color , would tend to be under - estimated by the prescription defined by the average population of the galaxy sample ( see eq .
( [ eq : fncextfinal ] ) ) .
since stars older than @xmath91 gyr do not contribute to uv emission , the above argument is only valid for galaxies with either continuous star formation history or continuous star formation history with superposed bursts younger than @xmath15 gyr . for a continuous star formation history with superposed bursts taking place at more than @xmath15 gyr ago
, the fuv - nuv color behaves as if no bursts occurred at all in the history . on the other hand ,
the correlations of the residuals of the fuv - nuv corrected fuv luminosities with the h@xmath0 luminosities , the h@xmath0 luminosity densities and the h@xmath0 attenuations are hard to be interpreted in terms of the dust - free fuv - nuv color because there is no evidence for a correlation between the absolute star formation activity and the intrinsic fuv - nuv color .
the possible reason for these correlations is the change in the effective attenuation curves ( see eq .
( [ eq : modtmp ] ) ) with the star formation intensities given that the intensity of the star formation activity may change the dust properties and the stars / dust geometry , and hence the effective attenuation curves .
these speculations agree with the argument by burgarella et al .
( 2005 ) and boquien et al .
( 2009 ) , who suggested that both star formation history and dust attenuation curve effects should be taken into account when the fuv - nuv color is used as a uv attenuation estimator .
in other words , a simple formalism of using fuv - nuv color as a fuv attenuation measure ( e.g. , eq .
( [ eq : tmp ] ) ) is far from enough .
additional constraints on star formation histories and dust attenuation curves are needed .
since fir photometry and hence the tir luminosity is sometimes not available , it is worthwhile calibrating the combinations of the fuv and the monochromatic ir waveband luminosities ( e.g. , 25@xmath4 luminosities data is small , no attempt was made to calibrate the combination of uv and 8@xmath4 luminosity as an attenuation or a sfr indicator .
] ) and radio emission at 1.4ghz as attenuation - corrected sfr measures . by matching the combination of the observed fuv and the 25@xmath4 infrared luminosities with the irx - corrected fuv luminosities ,
we obtain a coefficient of @xmath92 for fuv+25@xmath4 , which indicates an average l(25@xmath4 ) to l(tir ) ratio of @xmath14 0.12 .
this is well within the observed range for star - forming galaxies ( c.f .
, paper i ; calzetti et al .
the left panel of figure [ lumifuv25umradio.lha.eps ] shows a tight and linear correlation between the balmer - corrected h@xmath0 luminosities and the combined fuv and 25@xmath4 luminosities . since the relation between the balmer - corrected h@xmath0 luminosity and the irx - corrected fuv luminosity , which is the reference of the combined fuv and 25@xmath4 luminosity , is consistent with the predicted relation for a constant star formation history lasting for 100 myr and kroupa imf ( see section [ subsec : uvtir ] ) , we plot this model prediction here for reference ( the dashed line ) .
as expected , the data match the model well .
the dispersion is @xmath88 dex . in this panel , besides sings ( open circles ) and mk06 ( filled circles ) samples , we also include the central 20 @xmath20 20 regions of the sings galaxies , as denoted by open squares .
although the calibration was derived based on the integrated measurements of mk06 sample , it also compares well to the centers of the sings galaxies .
this perhaps is not surprising because of the large coverage of the centers ( typically several square kilo - parsecs ) . to our knowledge
, there exists only one such calibration in the literature .
zhu et al . ( 2008 ) calibrated a fuv and 24@xmath4 combination by comparing to a fuv - nuv corrected fuv luminosity .
they used the prescription published by treyer et al .
( 2007 ) to derive the fuv - nuv corrected fuv luminosity and obtained @xmath93 .
the coefficient of 6.31 is much higher than our value . in order to understand the discrepancy between zhu et al . and our calibration
, we compare our eq .
( [ eq : fncextfinal ] ) with the relation given by treyer et al .
we find that the relation provided by treyer et al .
gives systematically lower attenuation estimates than ours , which implies that a coefficient lower than 3.89 is required to match the fuv - nuv corrected fuv luminosity derived using treyer et al.s relation . on the contrary , zhu et al .
obtained a much higher value .
we can not figure out the reason that leads to a high coefficient obtained by zhu et al .
( 2008 ) . the combination of the observed fuv and 1.4ghz radio luminosities yields a similarly tight relation @xmath94 dex as shown in the right panel of figure [ lumifuv25umradio.lha.eps ] , but its correlation with the balmer - corrected h@xmath0 luminosities shows clear nonlinearity , with a slope of @xmath95 from a bisector fit . for the average h@xmath0 luminosity of our sample , @xmath96
, the radio - corrected fuv luminosity can deviate from a linear relation by 0.02 dex .
we conclude that when fir data are absent , the monochromatic 25@xmath4 luminosity and the 1.4ghz radio luminosity can be used as a slightly less accurate surrogate of the tir luminosity to correct for the dust attenuation in fuv .
in the previous sections , we have calibrated the fuv luminosity as a sfr indicator .
a relevant question is whether nuv can also serve as a reliable sfr tracer . in theory ,
fuv traces emission from stars younger than @xmath97 myr ( e.g. , meurer et al . 2009 ) , whereas nuv can be contributed by stars as old as @xmath15 gyr .
fuv and nuv emissions are both dominated by stars younger than @xmath98 myr though .
our experiment with starburst99 shows that for a galaxy with constant star formation history lasting for more than 1 gyr , its fuv emission increases 8% from 100 myr to 300 myr old and only 1% from 300 myr to 1 gyr old .
the equivalent numbers for nuv are 12% and 5% , respectively .
we note that models built from galaxev by bruzual & charlot ( 2003 ) produce consistent results .
in other words , the nuv emission of such a galaxy contributed by stars older than 100 myr is about 17% ( compared to 9% for fuv emission ) , which makes it a less reliable sfr tracer in galaxies with a significant fraction of stars being 100 myr
1 gyr old . in the past , however , when fuv observations were rarely available nuv measurements were the only uv sfr indicators available . even today
, nuv is sometimes used to probe the star formation activity ( e.g. , iglesias - pramo et al .
2006 ) ; recipes for using the ir / nuv ratio to estimate the attenuation in the nuv have been presented recently ( e.g. , buat et al . 2005 ; burgarella et al .
2005 ; cortese et al .
thus it is useful to assess the reliability of using nuv luminosity as a sfr indicator , using our new approach and datasets . given that nuv emission can be from more evolved stars than fuv , we did not fit the corresponding coefficient @xmath99 in eq .
( [ eq : irxbeta ] ) .
instead we estimated @xmath99 from @xmath48 according to @xmath100 ( i.e.,@xmath101 ) to obtain @xmath99 and @xmath102 , we got a smaller than @xmath103 value but this value is still consistent with @xmath103 within 1 @xmath18 error . ] .
then @xmath104 was estimated from fitting eq .
( [ eq : irxbeta ] ) with nuv substituted for fuv data and @xmath99 fixed . the resulting fitted value of @xmath104 was [email protected] .
following the analyses for fuv , we compare the tir / nuv corrected nuv luminosities with the balmer - corrected h@xmath0 luminosities in figure [ comluminuvirx.lhacor.rat.his.eps ] . to our pleasant surprise ,
the combined nuv and tir luminosities correlate with the balmer - corrected h@xmath0 luminosities nearly as well as fuv , and the scatter in nuv+tir vs. l@xmath106 is almost the same as that in fuv+tir vs. l@xmath106 ( 0.10 vs. 0.09 dex respectively ) .
the average ratio of the tir / nuv - corrected nuv luminosity to the balmer - corrected h@xmath0 luminosity is consistent with that expected for a constant star formation history over @xmath107 gyr for a kroupa imf .
we also examined the trends of the residuals of the combined nuv and tir luminosities relative to the balmer decrement ratio corrected h@xmath0 luminosities as functions of various parameters .
we find that all the behaviors shown in nuv are similar to those in fuv .
when 25@xmath4 infrared and radio continuum 1.4ghz luminosities are used as substitutes for tir luminosities , we obtain coefficients @xmath104 of [email protected] and @xmath108 respectively .
the scatters shown in nuv+25@xmath4 and nuv+1.4ghz versus l(h@xmath0)@xmath109 are identical to those shown in the same comparisons for fuv . from the above analyses we can see that the nuv luminosity can serve as a rather good sfr indicator for our sample galaxies , and star formation activities have been going on smoothly in these galaxies over the last 1 gyr . however , the use of nuv - based sfrs in galaxies with non - constant star formation histories during the last 1 gyr may be problematic .
our experiment with the stellar population synthesis modelling code galaxev ( bruzual & charlot 2003 ) shows that for a galaxy with a constant star formation history lasting for 100 myr , the nuv emission at 300 myr is 4.3% of its value at 100 myr ( compared to 2.6% for fuv ) . for a galaxy with high sfr ( @xmath110 m@xmath2yr@xmath111 ) lasting for 100 myr , the nuv emission at 300 myr can not be distinguished from a milky way like star - forming galaxy ( with sfr @xmath15 m@xmath2yr@xmath111 ) .
in previous sections we have derived prescriptions of using tir / uv and fuv - nuv as uv dust attenuation indicators by employing a new empirical method . specifically , these calibrations were obtained by fitting a tir / uv versus fuv - nuv relation , which was defined by assuming that the uv attenuation estimated from tir / uv ratio matches that from fuv - nuv color , to a nearby star - forming galaxy sample .
we have demonstrated that the combinations of the observed fuv and nuv luminosities with the tir luminosities ( i.e. , the tir / uv corrected uv luminosities ) are tightly correlated with the balmer - corrected h@xmath0 luminosities , and hence can serve as good sfr measures . by referencing to
the tir / uv corrected uv luminosities , we used the 25@xmath4 ( or 24@xmath4 ) infrared or 1.4 ghz radio continuum luminosities as surrogate of the tir luminosity to correct for the dust attenuation in the uv and found that they are slightly less accurate than the tir luminosity .
the combinations adopt the form of @xmath112 , and the corresponding attenuation in the uv has the form of @xmath113 $ ] , where @xmath114 is the approximation of the product of the inverse of the bolometric correction @xmath115 and the factor @xmath116 ( see section [ sec : calibration ] ) .
the coefficient @xmath114 and the dispersions of the combinations relative to the balmer - corrected h@xmath0 luminosities are listed in table 3 .
as we emphasized in paper i , these composite indices have been calibrated over limited ranges of galaxy properties and physical environments .
it is therefore important to understand the range of observations over which these calibrations are derived , and any systematics which were built into our calibrations .
it is also useful to compare our results with the calibrations presented in the literature by different groups .
we also reassess the consistency between the sfr derived from our dust attenuation corrected fuv luminosity and that based on balmer - corrected h@xmath0 luminosity .
in paper i we have already discussed the ranges of the galaxy properties ( e.g. , the attenuation in h@xmath0 , tir to h@xmath0 flux ratio , @xmath117 and tir luminosity ) covered by our samples . briefly our sample is composed of normal star - forming galaxies in the local universe .
neither early - type red galaxies nor dusty starburst galaxies are included in this sample , which is further demonstrated by the uv properties presented in this paper .
the ranges of irx and fuv - nuv color spanned by our sample galaxies are -0.19 to 2.39 dex and 0.07 to 1.06 mag respectively .
this coverage in irx and fuv - nuv color introduces uncertainties in the best fitted coefficients , as presented in section [ sec : calibration ] , due to the lack of constraints from objects redder than 1.06 mag and bluer than 0.07 mag . as a result caution
should be exercised when applying our calibrations to objects outside the above ranges ( see section [ subsec : sys ] ) . there
exist several calibrations for normal star - forming galaxies in the literature .
these calibrations were obtained under different assumptions about detailed star formation histories , dust attenuation curves , and the parameterization of the a@xmath11 irx relation ( see below ) .
it is thus necessary to test the consistency between them .
we compare our calibration for irx - based fuv attenuation estimates with those commonly used in the literature in figure [ comparebuatsbresifuv.eps ] . here the fuv attenuation ( top panel ) , the residual of the fuv attenuation estimated using the calibrations given by other groups relative to that using our calibration ( middle panel ) and the normalized ( percentage ) residual ( bottom panel ) are plotted as a function of irx .
the black solid line denotes our calibration .
the calibrations presented by other authors are shown with different line types as denoted in the top panel . specifically , the commonly used calibrations for star - forming galaxies by buat et al .
( 2005 ) , burgarella et al .
( 2005 ) and kong et al .
( 2004 ) are plotted .
we also show the well - known prescription for starburst galaxies by meurer et al .
the shaded region denotes the uncertainty in our calibration that is estimated using error propagation of eq .
( [ eq : irxextfinal ] ) , without considering the measurement errors in fuv and tir luminosities .
the range of this region in irx indicates the coverage by our sample objects , -0.19 to 2.39 dex ( see section [ subsec : limitations ] ) . outside this range
the relation shown by the black solid line is simply the extrapolation of that defined by our data . before comparing the calibrations in the literature to our a@xmath11 irx relation ,
it is interesting to compare the published relations to each other first . from figure [ comparebuatsbresifuv.eps ]
, it can be clearly seen that over the whole range of irx probed by our sample galaxies , the relations given by buat et al . and burgarella et al .
are almost indistinguishable from each other the difference is 0.1 mag at maximum . when compared to the kong et al .
relation , which is also derived for normal star - forming galaxies , at low irx ( l(tir)@xmath118 ) , the three relations for star - forming galaxies stay close to each other the difference is no more than 0.1 mag in the estimated attenuations .
but at higher irx ( l(tir)@xmath119 ) the kong et al .
relation shows significant difference from the other two , 0.3 - 0.5 mag .
interestingly , the difference between the starbursts relation and the buat et al . and burgarella et al .
relations is not dramatic ( no more than 0.5 mag ) , given that they are derived for different types of galaxies .
this comparison tells us for galaxies with @xmath120 irx @xmath121 , when the attenuation prescriptions were used blindly , regardless of galaxy types , the uncertainty in the uv attenuation would be less than 0.5 mag .
since our sample galaxies are mostly normal star - forming galaxies in the local universe ( see section [ subsec : limitations ] ) , it is expected that our calibration is more similar to those calibrated for star - forming galaxies than for starbursts .
figure [ comparebuatsbresifuv.eps ] confirms this expectation .
the starburst curve obtained by meurer et al . systematically overestimates the a@xmath11 compared to our calibration , although the difference is not dramatic , still within @xmath122 uncertainty of our calibration .
the difference between starbursts and star - forming galaxies is expected because the mean age of the dust heating population should be systematically lower in starburst galaxies than in normal star - forming galaxies .
overall our calibration is very close to those for star - forming galaxies derived by burgarella et al . and buat et al .
over the whole range of irx spanned by our sample .
it is worth noting that the prescriptions given by burgarella et al . and
buat et al .
are in polynomial forms , different from the form presented in eq .
( [ eq : uvextirx ] ) , which contributes to part of the differences . among the three star - forming relations that we explored here , that of kong et al . ( 2004 ) appears to deviate the most , as seen in figure [ comparebuatsbresifuv.eps ] . at @xmath123 the consistency between kong et al .
relation and ours is good , but at @xmath124 and @xmath125 the relation in kong et al .
over- and under - estimate @xmath28 at @xmath126 level , respectively .
we compare our calibration on a@xmath11 versus fuv - nuv relation with those in the literature in figure [ compareafuvuvcolorbisecresi.eps ] .
similar to figure [ comparebuatsbresifuv.eps ] , the a@xmath11 versus fuv - nuv relation is plotted in the top panel , the difference in a@xmath11 between the prescriptions in the literature and ours is shown in the middle panel , and the normalized ( percentage ) residual is shown in the bottom panel .
the shaded region denotes the uncertainty in our calibration over the range of fuv - nuv color covered by our sample galaxies . as before the relation obtained in section [ sec : calibration ] is plotted with a black solid line and the calibrations presented by other authors are represented by different types of lines as denoted in the top panel . in this case , the relations plotted include the starburst galaxy calibration by meurer et al . ( 1999 ) , the calibrations for spectroscopically and nuv - r color selected star - forming galaxies by salim et al .
( 2007 ) , the relation by treyer et al .
( 2007 ) for star - forming galaxies selected spectroscopically and the relation by seibert et al .
( 2005 ) for a wide assortment of galaxy types .
it is obvious that at a given fuv - nuv color , the starburst line derived by meurer et al .
( 1999 ) overestimates the a@xmath11 of normal star - forming galaxies dramatically compared to others .
this is consistent with the well - known deviation of star - forming galaxies from starbursts in irx-@xmath5 diagram ( e.g. , kong et al .
2004 ; cortese et al .
2006 ; dale et al .
2007 ; johnson et al .
the relations defined by star - forming galaxies stay close to each other but with different slopes and intercepts .
most of the differences shows up at fuv - nuv @xmath127 .
for fuv - nuv @xmath128 , the relations converge except for those by salim et al .
, which have constant a@xmath11 for galaxies redder than @xmath129 and @xmath130 in fuv - nuv color for nuv - r and spectroscopically selected galaxies respectively .
as discussed in section [ sec : other ] , the calibration derived by treyer et al . ( 2007 ) underestimates the attenuation by 0.4 mag at fuv - nuv @xmath131 mag with the discrepancy decreasing as the object gets redder in fuv - nuv color . at fuv - nuv=1.7 mag , the treyer et al .
calibration gives consistent result with ours .
the calibration by seibert et al .
( 2005 ) is the most consistent one with ours , and the estimated a@xmath11 from their relation agrees with that from ours within @xmath73 for the range of fuv - nuv 0.21.7 mag . for bluer color ( 0 0.2 mag ) , the difference is a bit larger but always smaller than 0.2 mag in absolute scale . in terms of relative difference ( bottom panel ) , both treyer et al . and seibert et al .
relations are consistent with ours within @xmath73 at fuv - nuv @xmath132 0.5 , with increasing discrepancy at bluer color . for salim et al .
calibrations , both the absolute and the relative differences are large , far beyond @xmath73 , for galaxies redder than @xmath133 in fuv - nuv because of the plateau in a@xmath11 defined in these relations , which is hard to be tested using our sample because of the limited range in fuv - nuv color covered by our galaxies ( as indicated by the shaded region ) . as shown in section [ sec : calibration ] , the scale parameter @xmath48 in equation ( [ eq : tmp ] ) represents the slope of the uv part of the attenuation curve .
we derive @xmath134 from @xmath48 and compare it with those derived from the commonly used dust attenuation laws published by calzetti et al .
( 2000 ; originally from calzetti et al . 1994 ; ) and charlot & fall ( 2000 ) . the attenuation curve by calzetti et al .
( 1994 ) was obtained directly from the fuv - to - near - ir spectra of a sample of uv - selected starburst galaxies .
charlot & fall ( 2000 ) used a similar but larger sample to derive their effective attenuation curve by tuning their model parameters to account for the distribution of the galaxies in the irx , h@xmath0/h@xmath5 ratio and ew(h@xmath0 ) versus @xmath5 diagrams .
the estimated values of @xmath135 from our calibration , calzetti et al .
( 2000 ) and charlot & fall ( 2000 ) are 1.35@xmath136 , 1.24 and 1.32 , respectively , which means that in the uv part of the dust attenuation curve , the slope of the attenuation curve defined by our sample is consistent with that derived using charlot & fall ( 2000 ) attenuation law within the uncertainties but marginally steeper ( at @xmath137 level ) than that estimated from calzetti et al .
( 1994;2000 ) .
the slope of our attenuation curve also suggests that the global fuv and nuv attenuations show no evidence for the effects of a strong 2175@xmath7 bump ( e.g. , odonnell 1994 ) .
for example , the odonnell milky way extinction curve has a @xmath134 ratio close to 0.9 , which is much smaller than 1.35@xmath136 defined by our galaxies .
based on the attenuation in the fuv estimated using our prescriptions , we can study the a@xmath11
a@xmath138 relation .
figure [ afuv.aha.cal.eps ] shows the irx - derived ( left panel ) and fuv - nuv derived ( right panel ) a@xmath11 as a function of h@xmath0/h@xmath5-derived a@xmath138 .
the solid line in this figure is derived from eqs .
( [ eq : ahafnc ] ) and ( [ eq : fncextfinal ] ) , which present the relations between fuv - nuv color and a@xmath10 and a@xmath11 respectively . for comparison , the relation derived from calzetti ( 2001 ) law after adapted to galex fuv waveband , which has the form @xmath139 , is plotted as the dotted line . as can be seen from the figure , at a given h@xmath0 attenuation , calzetti s law gives a lower fuv attenuation estimate than ours and the difference becomes more significant for galaxies that are more attenuated .
quantitatively speaking , however , the slope of the solid line , which determined from our data , is [email protected] , consistent with that given by calzetti s law within the error .
we performed similar comparisons for nuv with the calibrations published in the literature . when compared to the calibrations in the literature ,
similar to the results we saw for fuv , our calibration is in good agreement with the one given by burgarella et al .
( 2005 ) within 0.1 mag for the whole range of log[(tir)/l(nuv ) ] .
the a@xmath140/a@xmath10 ratio defined by our data is [email protected] , coincident with the predicted value from calzetti ( 2001 ) s law within the error .
it is now well - known that a significant fraction of ir emission comes from dust heating by old stellar populations ( e.g. , sauvage & thuan 1992 ; bell 2003 and references therein ) . in paper
i , by comparing our calibration for h@xmath0+tir as a sfr indicator with stellar population synthesis models , we found that for our sample objects up to 50% of the tir emission could be from stars older than 100 myr .
are the results presented here consistent with those estimates ? as mentioned in section [ sec : calibration ] , the coefficient @xmath30 in eq .
( [ eq : uvextirx ] ) is the product of the inverse of the bolometric correction @xmath44 and the factor @xmath141 . in paper
i , we have empirically derived @xmath40 by combining the tir luminosity and the multi - wavelength seds of a subset of mk06 galaxies with galex , sdss and 2mass observations .
briefly , eq . ( [ eq : bolometric ] ) was used to estimate @xmath40 from the tir and the bolometric luminosities , which were derived by adding the tir luminosity to the integrated luminosity from the fuv to k band . after substituting the values of @xmath36 and @xmath40 into @xmath142 , we find that the median of @xmath143 is around 1.5 . then the bolometric correction @xmath144 is estimated to be @xmath145 by comparing @xmath30 with @xmath142 . in order to estimate the effect of dust heating from old stars , we compare the bolometric correction derived from our data to those from stellar population synthesis models .
we found that the bolometric correction defined by our data is more than two times greater than the model prediction for a 100 myr old stellar population with constant star formation history .
this implies that stars older than 100 myr must contribute to the dust heating .
the spectral synthesis modelling using starburst99 indicates that the bolometric correction for a 10 gyr old stellar population with constant star formation history , solar metallicity and either a salpeter or a kroupa imf is only 15%20% lower than the estimated value above . in other words ,
the mean property of our sample objects approximates a stellar population that has been constantly forming stars for 10 gyr , which is exactly what we inferred from the analyses of h@xmath0+tir in paper i. it is worth noting that the consistent inference drawn from paper i and this paper ensure us the robustness of our results .
however , this estimate of the average stellar population of our sample galaxies conflicts with the dust - free fuv - nuv color derived in section [ sec : calibration ] , which is @xmath146 mag redder than a 10 gyr old galaxy with constant star formation history . all the estimates made here are very rough and they should only be used as a guide to our understanding of the underlying physics shown by the data qualitatively .
more careful modelling is needed for a quantitative understanding . as discussed in section [ subsec : limitations ] , our calibrations were derived over limited ranges of galaxy properties .
so it is important to understand possible systematics which were built into our prescriptions .
first of all , prescriptions calibrated for different types of galaxies are different .
this is not surprising because different types of galaxies show different irx-@xmath5 ( i.e. , fuv - nuv color ) relations . as shown in section [ subsec : comparisons ] , the irx - based and the fuv - nuv color - based attenuation correction prescriptions for star - forming galaxies differ from those for starburst galaxies
. another two extreme examples are dwarf irregulars and ( ultra)luminous infrared galaxies ( ( u)lirgs ) .
dale et al . (
2009 ) examined the location of the local volume legacy ( lvl ) sample galaxies in the irx-@xmath5 diagram .
nearly two - thirds of lvl galaxies are dwarf / irregular systems .
those authors showed that the majority of lvl galaxies fall below the irx-@xmath5 relation as defined by normal star - forming galaxies , and they cluster in an area with blue fuv - nuv color and low irx ( see also lee et al . 2009 ) .
the range in irx spanned by the lvl galaxies , for a given fuv - nuv color , is large more than an order of magnitude . for ( u)lirgs , howell et al .
( 2010 ; see also goldader et al .
2002 ; calzetti 2001 ) showed that ( u)lirgs in the great observatories all - sky lirg survey ( goals ) lie above the starburst relation , and the correlation between irx and fuv - nuv color is not strong either .
similar to dwarf / irregulars , at a given fuv - nuv color , the ( u)lirgs span more than an order of magnitude in irx .
one assumption we made in our calibration method is that the attenuation derived from irx matches that estimated from fuv - nuv color , which is reasonable for our normal star - forming galaxy sample given the relatively tight correlation between irx and fuv - nuv color , as shown in figure [ irxfuv.uvcolor.eps ] .
however , it breaks down for dwarf irregulars and ( ultra)luminous infrared galaxies given the large range of irx spanned by these galaxies at a given fuv - nuv color . therefore no attempt was made to calibrate an irx - based and fuv - nuv color - based attenuation correction prescriptions for these galaxies .
apart from different galaxy samples , the differences between different calibrations can be traced to the different methodologies .
previous calibrations have been based either on pure model assumptions , or on a comparison of observational data with models .
the difference in the form of the parameterization of the prescriptions also contributes to the differences . as we can see from the comparisons in section [ subsec : comparisons ] , the differences in methodology and parameterization ( buat et al . 2005 ; burgarella et al .
2005 and this work ) only account for a small portion of the differences , which can not even be distinguished within @xmath73 uncertainty , and the main contributors to the different relations are the star formation history and the dust attenuation curve that includes the effect of the dust / star distribution , as suggested by other authors ( e.g. , charlot & fall 2000 ; kong et al .
2004 ; burgarella et al .
2005 ) .
it is difficult to test the impact of dust attenuation curves on attenuation calibrations .
however , it is possible to estimate the effect of dust - heating stellar populations .
as discussed in section [ subsec : oldpops ] , for our sample galaxies , up to half of the tir emission could be from dust heated by stars older than 100 myr .
by contrast , in starburst galaxies , the dust is dominantly heated by stellar populations younger than 100 myr ( meurer et al .
this difference in dust - heating stellar populations of starbursts and our star - forming galaxies results in the difference in attenuation estimates , 0.3 mag at maximum over @xmath120 irx @xmath121 , as shown in figure [ comparebuatsbresifuv.eps ] .
the effects on the uv color - based corrections are much larger than on irx - based prescriptions .
the irx - based prescriptions are surprisingly consistent with each other .
however different fuv - nuv based attenuation schemes have by comparison enormous inconsistencies .
this underscores our concerns about color - based attenuation corrections based on scatter in our own data . in principle , by comparing modelled seds with multi - wavelength data , the differences in star formation histories and dust attenuation curves can be taken into account on a galaxy - to - galaxy basis .
however , the dust attenuations derived this way and the resulting relations of @xmath46 versus irx or @xmath46 versus fuv - nuv color are dependent on the adopted stellar population synthesis models and the dust attenuation curves . on the other hand ,
the requirement of the availability of large libraries of models and multi - wavelength data makes this method less practical in many cases .
by contrast , our approach is more straightforward and less model - dependent .
our calibrations rely almost entirely on the data themselves and hence can serve as independent checks of other model - dependent methods . although we assumed an average stellar population and dust attenuation curve ( as discussed in section [ sec : calibration ] ) , they do not introduce systematics into the calibrations and are valid on a statistical basis .
therefore , our calibrations should be robust .
they also tend to be consistent with most of the published calibrations for normal star - forming galaxies .
the last possible systematic error comes from the tir luminosities we adopted in our calibrations .
we used tir luminosities based on iras observations , which are systematically lower than those based on spitzer / mips measurements by @xmath147 on average ( section [ sec : sample ] and figure 1 in paper i ) . if we simply substitute the iras tir luminosities by the mips tir luminosities , the coefficient in front of irx in eq .
( [ eq : irxextfinal ] ) will be lower by 24% , i.e. , changes to 0.37 , which is within the @xmath122 uncertainty of the coefficient .
so if one uses mips tir luminosities to estimate the fuv attenuation using our irx - based prescription ( i.e. , eq .
( [ eq : irxextfinal ] ) ) , the uncertainty will be less than 0.3 mag . in summary ,
our irx - based attenuation correction has an uncertainty of @xmath148 mag , when it is applied to starburst galaxies or to galaxies with mips based tir luminosities .
by contrast , if one uses our fuv - nuv color prescription to estimate the fuv attenuation for starbursts , he can underestimate the fuv attenuation by more than 1 - 2 mag .
but for dwarf irregulars or ( u)lirgs , our prescriptions may suffer from severe problems . in the introduction
, we mentioned that treyer et al .
( 2007 ) derived an a@xmath149 fuv - nuv relation by matching the fuv - based and h@xmath0-based sfrs .
the attenuation derived this way is dependent on the intrinsic ratio of the fuv to h@xmath0 luminosity , which in turn is dependent on model assumptions .
here we use a similar method to re - calibrate the a@xmath11irx relation in order to evaluate the model dependence of the calibration obtained this way .
we converted the balmer - corrected h@xmath0 luminosity to a balmer - corrected fuv luminosity using a theoretical fuv / h@xmath0 ratio and then force the combined observed ( i.e. , without internal dust attenuation correction ) fuv and tir luminosity to match the balmer - corrected fuv luminosity so that the coefficient @xmath30 in eq .
( [ eq : uvextirx ] ) can be obtained .
we used the same set of models as those used in figure [ comlumiirx.his.eps ] and figure [ comlumifnc.his.eps ] .
the model predictions and the best - fit coefficients @xmath30 are listed in table 2 ( for reference the corresponding quantities for nuv are also listed . ) .
consistent with what we saw from figure [ comlumiirx.his.eps ] , table 2 shows that our calibration derived in section [ sec : calibration ] is in excellent agreement with that predicted by assuming a constant star formation history with kroupa imf and age 100 myr .
the fuv / h@xmath0 ratio changes at most by 10% when only the effect of a single parameter is taken into account , and
accordingly the resulting coefficient @xmath30 varies by 11% .
the change in the coefficient directly affects the estimated attenuation from irx .
this impact is shown in figure [ compareafuvirxdiffimfage.eps ] .
it is clear that assuming a salpeter or a kroupa imf does not affect the calibrations dramatically at maximum @xmath14 0.07 mag over the range of irx=0 3 .
the calibrations are also affected by assuming either a 100 myr old or a 1 gyr ( or older ) old stellar population at maximum @xmath14 0.11 mag over the range tested .
however the most widely used k98 s prescriptions give larger discrepancy -0.2 mag at maximum when irx=3 . this large discrepancy with k98 s prescriptions
is probably due to the use of the old stellar evolutionary models , which leads to an over - estimate of h@xmath0 luminosity by 11% compared to that from starburst99 under the same set of assumptions of imf , metallicity and star formation history . from these tests
, we can draw the following conclusions : ( 1 ) the calibration established using our new method is consistent with that based on sfr matching method for a constant star formation history with kroupa imf and age 100 myr . ( 2 ) compared to our prescription , the adoption of a salpeter imf and/or an older age both leads to an overestimate in a@xmath11 by up to 0.1 mag over the range of irx=0 3 . ( 3 ) relative to our calibration , the widely used k98 s sfr prescriptions can underestimate a@xmath11 by up to 0.2 mag ( at irx=3 ) .
we have calibrated @xmath150}$ ] ( i.e. , irx ) and the fuv - nuv color as fuv dust attenuation indicators using a nearby star - forming galaxy sample studied by moustakas & kennicutt ( 2006 ) . the combinations of 25@xmath4 and 1.4ghz radio continuum luminosities with the observed fuv luminosities were also empirically calibrated to probe the dust attenuation corrected fuv luminosities .
similar calibrations were derived for nuv band as well .
the coefficients in these calibrations are summarized in table 3 ( see section [ sec : uncertainties ] ) .
these prescriptions provided in table 3 can be used to derive attenuation - corrected star formation rates .
our main results can be summarized as follows .
the irx - corrected fuv luminosities based on our new calibration , i.e. , @xmath151 show tight and linear correlation with the balmer decrement corrected h@xmath0 luminosities , with a rms scatter of @xmath1050.09 dex ( @xmath152 mag ) . statistically speaking ,
their ratios are consistent with the model predictions from starburst99 by assuming a constant star formation rate over 100 myr and solar metallicity .
this consistency applies whether a salpeter or a kroupa imf with lower and upper mass limits of 0.1 and 100 m@xmath2 is adopted .
furthermore , these dust attenuation corrected fuv / h@xmath0 ratios do not show any trend with other galactic properties over the ranges covered by our sample objects .
these results suggest that linear combinations of tir luminosities and the observed fuv luminosities ( without internal attenuation corrections ) are excellent star formation rate tracers .
the fuv - nuv corrected fuv luminosities are broadly correlated with the balmer - corrected h@xmath0 luminosities .
but they do not trace each other linearly and the rms scatter is large @xmath1050.23 dex ( @xmath153 mag ) , which is @xmath3 times larger than the case for irx - corrected fuv luminosities .
in addition , the attenuation corrected fuv / h@xmath0 ratios show correlations with a few other galactic properties , which will introduce systematic errors into the attenuation estimates in different galaxy samples .
we confirm others findings that fuv - nuv color is not a good dust attenuation indicator for normal star forming galaxies though it can be used with caution when ir data are not available .
linear combinations of 25@xmath4 and 1.4ghz radio continuum luminosities with the observed fuv luminosities can be used as surrogates of irx - corrected fuv luminosities to trace the attenuation - corrected star formation rates .
their correlations with the balmer - corrected h@xmath0 luminosities are slightly less tight than the irx - corrected fuv luminosities
0.13 dex and 0.14 dex , and a non - linearity is shown in the correlation between fuv+1.4ghz and balmer - corrected h@xmath0 luminosities with a slope of @xmath95 .
our calibrations are tested on normal star - forming galaxies with balmer - based a@xmath138 = 0 2.5 mag , irx = @xmath154 2.39 and fuv - nuv = 0.07 1.06 mag .
the a@xmath11 irx relation shows overall consistency with prescriptions for star - forming galaxies by burgarella et al .
( 2005 ) and buat et al .
( 2005 ) within @xmath73 uncertainty . while our a@xmath11 fuv - nuv color relation is in good agreement with that presented by seibert et al .
when our irx - based attenuation correction is applied to starburst galaxies or to galaxies with mips based tir luminosities , it has an uncertainty of @xmath148 mag .
by contrast , if our fuv - nuv color prescription is used to estimate the fuv attenuation for starbursts , it can underestimate the fuv attenuation by more than 1 - 2 mag .
our analyses show that the combination of the nuv and the tir luminosities can serve as a rather good sfr indicator for our sample galaxies , but may become problematic in galaxies with non - constant star formation histories in the past @xmath15 gyr .
hao acknowledges the support of a royal society uk china fellowship .
hao also acknowledges the support from the nsfc ( grant no .
10833006 and 11003015 ) .
we thank an anonymous referee for helpful comments that improved the paper .
some / all of the data presented in this paper were obtained from the multimission archive at the space telescope science institute ( mast ) .
stsci is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 .
support for mast for non - hst data is provided by the nasa office of space science via grant nag5 - 7584 and by other grants and contracts .
lrrr arp256 & 113.9 & [email protected] & [email protected] + ngc0157 & 24.0 & [email protected] & [email protected] + ngc0245 & 56.5 & [email protected] & [email protected] + ic0051 & 24.8 & [email protected] & [email protected] + ngc0337 & 23.5 & [email protected] & [email protected] + ic1623 & 84.1 & [email protected] & [email protected] + mcg-03 - 04 - 014 & 139.9 & [email protected] & [email protected] + ngc0695 & 135.4 & [email protected] & [email protected] + ngc0877 & 54.6 & [email protected] & [email protected] + ngc0922 & 43.4 & [email protected] & [email protected] + ngc0958 & 79.5 & [email protected] & [email protected] + ngc0959 & 10.9 & [email protected] & [email protected] + ngc0976 & 59.8 & [email protected] & [email protected] + ngc1058 & 9.8 & [email protected] & [email protected] + ngc1084 & 20.0 & [email protected] & [email protected] + ngc1140 & 21.2 & [email protected] & [email protected] + ngc1156 & 7.8 & [email protected] & [email protected] + ngc1359 & 28.0 & [email protected] & [email protected] + ngc1385 & 21.2 & [email protected] & [email protected] + iras03359 + 1523 & 147.8 & [email protected] & [email protected] + ngc1421 & 29.7 & [email protected] & [email protected] + ugc02982 & 74.5 & [email protected] & [email protected] + ngc1569 & 2.0 & [email protected] & [email protected] + ngc2139 & 26.7 & [email protected] & [email protected] + ngc2337 & 7.9 & [email protected] & [email protected] + ngc2415 & 57.2 & [email protected] & [email protected] + ngc2500 & 11.0 & [email protected] & [email protected] + ngc2537 & 6.9 & [email protected] & [email protected] + ugc05028 & 56.7 & [email protected] & [email protected] + ngc2903 & 8.9 & [email protected] & [email protected] + ngc3239 & 10.5 & [email protected] & [email protected] + ngc3265 & 25.0 & [email protected] & [email protected] + ugc05720 & 26.9 & [email protected] & [email protected] + ngc3344 & 7.9 & [email protected] & [email protected] + ngc3353 & 19.1 & [email protected] & [email protected] + ngc3367 & 49.3 & [email protected] & [email protected] + arp270 & 30.3 & [email protected] & [email protected] + ngc3432 & 10.4 & [email protected] & [email protected] + ngc3442 & 31.6 & [email protected] & [email protected] + ngc3521 & 8.6 & [email protected] & [email protected] + ngc3600 & 13.3 & [email protected] & [email protected] + ic0691 & 23.7 & [email protected] & [email protected] + ngc3726 & 19.8 & [email protected] & [email protected] + ngc3738 & 4.9 & [email protected] & [email protected] + ngc3769 & 19.8 & [email protected] & [email protected] + ugc06665 & 87.4 & [email protected] & [email protected] + ngc3870 & 19.8 & [email protected] & [email protected] + ngc3893 & 15.5 & [email protected] & [email protected] + ngc3928 & 16.9 & [email protected] & [email protected] + ngc3949 & 19.8 & [email protected] & [email protected] + ngc4004 & 55.6 & [email protected] & [email protected] + arp244 & 25.0 & [email protected] & [email protected] + ngc4051 & 15.5 & [email protected] & [email protected] + ngc4062 & 11.4 & [email protected] & [email protected] + ngc4085 & 19.8 & [email protected] & [email protected] + ngc4088 & 19.8 & [email protected] & [email protected] + ngc4096 & 10.8 & [email protected] & [email protected] + ngc4100 & 19.8 & [email protected] & [email protected] + ngc4214 & 2.9 & [email protected] & [email protected] + ngc4218 & 19.8 & [email protected] & [email protected] + ngc4254 & 12.9 & [email protected] & [email protected] + ngc4303 & 10.6 & [email protected] & [email protected] + ngc4384 & 42.8 & [email protected] & [email protected] + ngc4414 & 17.7 & [email protected] & [email protected] + ngc4605 & 5.2 & [email protected] & [email protected] + ngc4618 & 8.9 & [email protected] & [email protected] + ngc4625 & 10.2 & [email protected] & [email protected] + ngc4651 & 20.6 & [email protected] & [email protected] + ngc4656 & 8.7 & [email protected] & [email protected] + ngc4666 & 27.6 & [email protected] & [email protected] + ngc4670 & 23.2 & [email protected] & [email protected] + ngc4713 & 13.7 & [email protected] & [email protected] + ngc4900 & 9.0 & [email protected] & [email protected] + ngc5014 & 23.8 & [email protected] & [email protected] + mrk0066 & 95.6 & [email protected] & [email protected] + ngc5194 & 8.0 & [email protected] & [email protected] + arp240 & 104.9 & [email protected] & [email protected] + ngc5253 & 3.1 & [email protected] & [email protected] + ngc5430 & 49.0 & [email protected] & [email protected] + ngc5607 & 112.2 & [email protected] & [email protected] + ngc5591 & 117.6 &
[email protected] & [email protected] + ngc5653 & 58.7 & [email protected] & [email protected] + ngc5676 & 37.7 & [email protected] & [email protected] + ugc09560 & 23.7 & [email protected] & [email protected] + ic1076 & 94.6 & [email protected] & [email protected] + ngc5996 & 54.6 & [email protected] & [email protected] + ngc6052 & 74.8 & [email protected] & [email protected] + ngc6090 & 130.2 & [email protected] & [email protected] + ngc6926 & 87.6 & [email protected] & [email protected] + ic5179 & 50.3 & [email protected] & [email protected] + ngc7316 & 79.0 & [email protected] & [email protected] + cgcg453 - 062 & 106.0 & [email protected] & [email protected] + ngc7624 & 61.1 & [email protected] & [email protected] + ngc7625 & 24.4 & [email protected] & [email protected] + ngc7673 & 48.9 & [email protected] & [email protected] + ngc7678 & 49.7 & [email protected] & [email protected] + ngc7798 & 34.6 & [email protected] & [email protected] + lccccccc k98 & -43.147 & -42.975 & -41.102 & 0.0090 & 0.0134 & 0.38 & 0.21 + salpeter imf , @xmath155 , 100 myr & -43.170 & -42.959 & -41.056 & 0.0077 & 0.0125 & 0.48 & 0.24 + kroupa imf , @xmath155 , 100 myr & -43.350 & -43.137 & -41.257 & 0.0081 & 0.0132 & 0.45 & 0.22 + salpeter imf , @xmath155 , 1 gyr & -43.207 & -43.023 & -41.056 & 0.0071 & 0.0108 & 0.53 & 0.30 + kroupa imf , @xmath155 , 1 gyr & -43.384 & -43.196 & -41.257 & 0.0075 & 0.0115 & 0.50 & 0.27 + lcc l(fuv)@xmath156 + a*l(tir ) & [email protected] & 0.09 + l(fuv)@xmath156 + a*l(25 @xmath157 m ) & [email protected] & 0.13 + l(fuv)@xmath156 + a*l@xmath158 & [email protected] @xmath20 1.e20 & 0.14 + l(nuv)@xmath156 + a*l(tir ) & [email protected] & 0.10 + l(nuv)@xmath156 + a*l(25 @xmath157
m ) & [email protected] & 0.13 + l(nuv)@xmath156 + a*l@xmath158 & [email protected] @xmath20 1.e20 & 0.14 + | we present new calibrations of far - ultraviolet ( fuv ) attenuation as derived from the total infrared to fuv luminosity ratio ( irx ) and the fuv - nuv color .
we find that the irx - corrected fuv luminosities are tightly and linearly correlated with the attenuation - corrected h@xmath0 luminosities ( as measured from the balmer decrement ) , with a rms scatter of @xmath1 dex .
the ratios of these attenuation - corrected fuv to h@xmath0 luminosities are consistent with evolutionary synthesis model predictions , assuming a constant star formation rate over 100 myr , solar metallicity and either a salpeter or a kroupa imf with lower and upper mass limits of 0.1 and 100m@xmath2 . the irx - corrected fuv to balmer - corrected h@xmath0 luminosity ratios
do not show any trend with other galactic properties over the ranges covered by our sample objects .
in contrast , fuv attenuation derived from the fuv - nuv color ( uv spectral slope ) show much larger random and systematic uncertainties . when compared to either balmer - corrected h@xmath0 luminosities or irx - corrected fuv luminosities the color - corrected fuv luminosities show @xmath3 times larger rms scatter , and systematic nonlinear deviations as functions of luminosity and other parameters .
linear combinations of 25@xmath4 and 1.4ghz radio continuum luminosities with the observed fuv luminosities are also well correlated with the balmer - corrected h@xmath0 luminosities .
these results provide useful prescriptions for deriving attenuation - corrected star formation rates of galaxies based on linear combinations of uv and ir or radio luminosities , which are presented in convenient tabular form .
comparisons of our calibrations with attenuation corrections in the literature and with dust attenuation laws are also made . |
the notion of energy in the radiation regime ( in our paper referred to as the trautman - bondi energy ) has been introduced in einstein s theory of gravity by trautman @xcite and independently by bondi @xcite .
it measures that part of the gravitational energy of an isolated system , which `` has not yet been radiated '' . in conformal spacetime compactification
, the trautman - bondi energy may be assigned to any space - like hypersurface having a regular intersection with the conformal boundary @xmath0 ( null infinity , or the _ scri _ ) @xcite . due to radiation , the trautman - bondi energy unlike the total ( adm ) energy is not conserved , but is _ decreasing _ because it may be partially radiated in a form of gravitational waves .
as shown in @xcite , the validity of the trautman - bondi energy goes far beyond the gravitational context and may be used in any hyperbolic field theory , also special - relativistic one . in particular , it has a beautiful hamiltonian interpretation .
the goal of the present paper is to apply this idea to the scalar field theory , where the cauchy data of the system are assigned to a hyperboloid
. field evolution on hyperboloids is proved to be a hamiltonian system , if we complete cauchy data by the _ radiation data _ at the _ scri _ , and supplement the trautman - bondi energy with the corresponding _
radiation energy_. the sum of the two defines the total hamiltonian function of the `` matter + radiation '' system . but
to `` tailor '' the two disjoint objects : 1 ) the field cauchy data on a hyperboloid and 2 ) the radiation data on the _ scri _ , into a single object , appropriate compatibly conditions ( often called `` corner conditions '' @xcite ) must be imposed .
we propose a universal approach which solves all these issues .
hamiltonian description of any field dynamics is based on a `` 3 + 1 '' foliation of spacetime .
leaves of the foliation are labeled by a parameter called the `` time variable '' .
phase space of this dynamics is composed of all the possible field cauchy data on a given leaf . the `` 3 + 1 '' decomposition provides also an identification between different leaves of the foliation , which makes the field dynamics and its hamiltonian function uniquely defined . here , we consider the hamiltonian description within a simple model : the massless scalar field , satisfying the wave equation in two- or four - dimensional minkowski spacetime .
this means that the space is one - dimensional : @xmath1 or three - dimensional : @xmath2 .
contrary to the standard ( `` adm '' ) formulation , the field initial data will not be assigned to spatially flat cauchy surfaces , but to hypersurfaces which extend to null infinity , namely spacelike hyperboloids .
naively , such a cauchy problem can not be described as a hamiltonian system : future evolution of the system is well defined , but the past evolution is absolutely non - unique and may be arbitrarily modified by radiation .
nevertheless , we are going to show , that the hamiltonian formulation of the field evolution is possible .
for this purpose we have to complete cauchy data on a hyperboloid by appropriate _ radiation data _ at light infinity . for pedagogical reasons ,
we begin our analysis with a ( much simpler ) finite case , where we restrict field dynamics to a finite light - cone and describe initial data on its ( characteristic ! ) boundary and on the finite part of the hyperboloid , contained within the cone . at the end
, we may consider the limiting case ( in our notation , this corresponds to @xmath3 ) , where the finite cone is shifted to infinity .
this way we obtain the hamiltonian description of initial data on the entire hyperboloid and on the _ scri _ @xmath4 ( conformal boundary of the spacetime ) . in the subsequent section
we describe the two - dimensional `` toy model '' .
the complete , four - dimensional theory is analyzed in section [ sec:13case ] .
let us consider a one - parameter family of past oriented light cones in the two - dimensional minkowski space time : @xmath5 where @xmath6 is the time coordinate of the vertex of @xmath7 .
we introduce new coordinates @xmath8 connected with minkowski coordinates @xmath9 in the following way : @xmath10 where @xmath11 . for @xmath12 new coordinates parameterize the whole cone @xmath7 .
moreover , surfaces @xmath13 const.@xmath14 correspond to hyperboloids . in order to describe field dynamics in a hamiltonian way
, we begin with the standard , relativistic lagrangian for wave equation : @xmath15 expressing the lagrangian density `` @xmath16 '' in new coordinates we obtain : @xmath17 where @xmath18 \left(\frac{\partial \varphi}{\partial\xi}\right)^2\ , .\ ] ] denoting @xmath19\ ] ] we obtain the autonomous ( i.e. @xmath20-independent ) lagrangian function : @xmath21 now , the standard procedure leading from the lagrangian to the hamiltonian description may be applied .
we first define conjugate momenta : @xmath22 where @xmath23 , and calculate the variation of the lagrangian : @xmath24 field equation ( in our case it is always the wave equation @xmath25 ) is equivalent to vanishing of the euler - lagrange term in , hence , equivalent to the following equation @xmath26 integrating over the volume @xmath27 , we obtain an identity valid for fields satisfying wave equation @xmath28 where `` dot '' denotes derivative with respect to the new time variable @xmath20 , while @xmath29 is the restriction of the field @xmath30 to the cauchy surface @xmath31 : @xmath32 for @xmath33 .
the time component of the momentum : @xmath34 , provides , together with @xmath29 , the complete description of cauchy data on this surface .
legendre transformation between @xmath35 and @xmath36 gives us : @xmath37_{\partial v_\epsilon},\end{aligned}\ ] ] where the hamiltonian @xmath38 is defined by @xmath39 canonical structure in the space of cauchy data is given by the standard symplectic form : @xmath40 formulae and imply the following relations between `` velocity '' @xmath35 and momentum @xmath36 : @xmath41 taking into account and , the hamiltonian may be written explicitly in terms of the cauchy data on @xmath42 @xmath43}(\phi,\pi ) & = \frac 12 \int_{-\frac{1-\epsilon}{1+\epsilon}}^{\frac{1-\epsilon}{1+\epsilon } } \left\ { \kappa \left ( \pi -\kappa^{-1}\xi \frac { \partial \phi}{\partial \xi}\right)^2 + \left(\kappa-\kappa^{-1}\xi^2\right ) \left(\frac{\partial \phi}{\partial \xi}\right)^2 \right\ } { { \rm d}}\xi \;.\end{aligned}\ ] ] the factor @xmath44 is positive for @xmath45 .
moreover , we have : @xmath46 this implies positivity of the hamiltonian for @xmath47 , i.e inside the cone @xmath7 . the following hamiltonian equations ( equivalent to euler - lagrange equations )
may be derived from the hamiltonian : @xmath48 provided no boundary terms remain after the integration by part of its variation , which , _ a priori _ , is not true !
this apparent paradox is implied by the fact , that evolution of cauchy data on a hyperboloid is well defined only forward in time and , whence , does not correspond _ a priori _ to any hamiltonian system . to overcome this difficulty , we take into account missing data on the light cone below hyperboloid and treat it as a part of cauchy data ( cf .
@xcite ) . for this purpose
we extend parametrization , beyond the volume @xmath49 , taking into account also the corresponding points on the boundary of the cone : @xmath50 and consider the data @xmath51 on the entire surface @xmath52 .
thus @xmath53 for @xmath54 , @xmath55 for @xmath56 .
equation implies that @xmath57 , for @xmath58 , whereas implies @xmath59 for @xmath60 . within these regions of the cauchy surface ,
the dynamics consists in transporting the field data @xmath51 over @xmath61 along the field @xmath62 , according to following equations : @xmath63 where @xmath64 denotes the lie derivative along the vector field @xmath62 .
the above equations can be derived from the following hamiltonians ( generators of space translations ) : @xmath65 } h_{[\frac{1-\epsilon}{1+\epsilon } , \infty)}(\phi , \pi ) & : = \int_{\frac{1-\epsilon}{1+\epsilon}}^{\infty } ( \pi \dot{\phi } - \mathcal{l } ) { { \rm d}}\xi = \int_{\frac{1-\epsilon}{1+\epsilon}}^{\infty } ( -\pi \partial_{\xi } \phi ) { { \rm d}}\xi\ ; , \end{aligned}\ ] ] @xmath66 } h_{(-\infty,-\frac{1-\epsilon}{1+\epsilon}]}(\phi , \pi ) & : = \int^{-\frac{1-\epsilon}{1+\epsilon}}_{-\infty } ( \pi \dot{\phi } - \mathcal{l } ) { { \rm d}}\xi = \int^{-\frac{1-\epsilon}{1+\epsilon}}_{-\infty } ( \pi \partial_{\xi } \phi ) { { \rm d}}\xi\ ; , \end{aligned}\ ] ] where @xmath67 vanishes identically as a pull - back of the scalar density @xmath68 via the degenerate coordinate transformation - , and the momentum @xmath36 on the cauchy surface @xmath61 is equal to the pull - back of the ( odd ) differential form @xmath69 to @xmath70
. moreover , momentum @xmath71 coincides with @xmath72 for @xmath73 , and with @xmath74 for @xmath75 , as the pull - back of the same form to the hypersurface @xmath76 .
hence , we obtain the following constraints : @xmath77 the phase space of cauchy data on the entire @xmath61 is described by the pairs @xmath78 defined on the whole @xmath79 and fulfilling constraint or in the corresponding regions . the total hamiltonian function @xmath80 on the entire phase space @xmath81 is equal to the sum of these partial hamiltonians : @xmath82}+ h_{[-\frac{1-\epsilon}{1+\epsilon } , \frac{1-\epsilon}{1+\epsilon}]}+ h_{[\frac{1-\epsilon}{1+\epsilon } , \infty ) } \ .\end{aligned}\ ] ] variation of @xmath80 gives @xmath83_{-\infty}^{-\frac{1-\epsilon}{1+\epsilon } } + \big[\pi^1\delta\phi\big]_{-\frac{1-\epsilon}{1+\epsilon } } ^{\frac{1-\epsilon}{1+\epsilon } } + \big[\pi^1\delta\phi\big]^{\infty}_{\frac{1-\epsilon}{1+\epsilon}}\end{aligned}\ ] ] with appropriate values for @xmath84 in the respective regions of @xmath61 .
the functional @xmath80 defines the hamiltonian dynamics of the total system , if the boundary terms in formula cancel .
this requires corner conditions at @xmath85 and @xmath86 and sufficient strong fall - off condition at infinity . to analyse these conditions
it is useful to reformulate our hamiltonian description .
taking into account constraints and and using and we have @xmath65_phi } h_{[\frac{1-\epsilon}{1+\epsilon } , \infty ) } & = \int_{\frac{1-\epsilon}{1+\epsilon}}^{\infty } ( \partial_{\xi } \phi)^2 { { \rm d}}\xi\ , \\ \label{eq : h_[-infty]_phi } h_{(-\infty,-\frac{1-\epsilon}{1+\epsilon } ] } & = \int^{-\frac{1-\epsilon}{1+\epsilon}}_{-\infty } ( \partial_{\xi } \phi)^2 { { \rm d}}\xi\ , \end{aligned}\ ] ] and the corresponding symplectic structures @xmath87 } & = \int^{-\frac{1-\epsilon}{1+\epsilon}}_{-\infty } ( \partial_\xi \delta\phi \wedge \delta \phi ) { { \rm d}}\xi \quad \text{for } \xi\leq -\frac{1-\epsilon}{1+\epsilon}\ ; .\end{aligned}\ ] ] changing variables in the following way : @xmath88 and denoting : @xmath89 we see that functions @xmath90 and @xmath91 do not depend upon @xmath20 ( see and ) , hence they are single variable functions .
now we can write formulas - and - jointly @xmath92 } + h_{[\frac{1-\epsilon}{1+\epsilon},\infty ) } \\
\label{eq : h_ext } & = \int_{-\infty}^{0 } \big\{\big(\partial_{\lambda } f^{-}_{\epsilon}\big)^2 + \big(\partial_{\lambda } g^{-}_{\epsilon}\big)^2 \big\}{{\rm d}}\lambda \\
\notag \omega_{\exter,\epsilon } & : = \omega_{(-\infty,-\frac{1-\epsilon}{1+\epsilon } ] } + \omega_{[\frac{1-\epsilon}{1+\epsilon},\infty)}\ ; , \\
\label{eq : omega_ext } & = \int^{0}_{-\infty } \big\ { \partial_\lambda \delta f^{-}_{\epsilon } \wedge \delta f^{-}_{\epsilon } + \partial_\lambda \delta g^{-}_{\epsilon } \wedge \delta g^{-}_{\epsilon } \big\ } { { \rm d}}\lambda\ ; , \end{aligned}\ ] ] where @xmath93 formula shows , that the canonical structure of `` external data '' ( i.e. data outside of the hyperboloid ) can be described by the `` @xmath94 '' symplectic form . to find the appropriate functional - analytic framework for the problem and , in particular , to obtain correct formulation of the corner condition ,
it is useful to reformulate also the `` internal data '' ( on the hyperboloid ) in a similar way .
indeed , we shall prove in the sequel that the transition between the hyperboloidal data and that part of the light - cone data , which lies _ above the hyperboloid _ is a canonical transformation , which converts the canonical form `` @xmath95 '' into the faddeev form `` @xmath96 '' , ( cf .
@xcite ) . for this purpose , assume that we know the light - cone data @xmath97 , where @xmath98 is a function which lives on the left piece of the light - cone , whereas function @xmath99 lives on the right one .
we use convenient coordinates @xmath100 and in particular the left piece of @xmath101 is given by @xmath102 , while the right one is given by @xmath103 .
therefore @xmath104 where @xmath30 is the general solution of the wave equation , i.e. @xmath105 where @xmath106 and @xmath107 are functions of one variable .
due to and , we can express the general solution in terms of the light - cone data @xmath108 : @xmath109 where @xmath110 .
formula implies the following transformation between the hyperboloidal data @xmath78 and the light - cone data @xmath108 : @xmath111 where we used definition of the momentum @xmath36 .
integrating over intervals @xmath112 $ ] and @xmath113 $ ] , we obtain , due to , the following transformation : @xmath114 and @xmath115 substitution of , into the symplectic form defined on the space of cauchy data @xmath51 over the hyperboloid @xmath42 gives us : @xmath116 changing variables in in the following way @xmath117 and denoting functions @xmath118 ( we use superscript `` @xmath119 '' , because these functions are defined over the positive half - line @xmath120 ) we can write the formula in the following form @xmath121 } \omega_{\inter,\epsilon } & = \int_{0}^{\infty } \big\ { \partial_\lambda \delta f^{+}_{\epsilon } \wedge \delta f^{+}_{\epsilon } + \partial_\lambda \delta g^{+}_{\epsilon } \wedge \delta g^{+}_{\epsilon}\big\ } { { \rm d}}\lambda\ .\end{aligned}\ ] ] equation shows that , indeed , the `` internal '' part of the canonical structure can also be described by the `` @xmath94 '' symplectic form defined on the light - cone data .
this ends our proof .
now , the corner conditions at points @xmath122 ( necessary for cancellation of boundary terms in the hamiltonian formula ) are expressed into the compatibility condition between the external ( given by ) and the internal ( given by ) structures , which must be satisfied at the point @xmath123 .
an obvious condition is that the total symplectic form : @xmath124 where @xmath125 , @xmath126 are equal to @xmath127 , @xmath128 for @xmath129 and to @xmath130 , @xmath131 for @xmath132 , respectively , must be well defined . in particular , a step discontinuity is excluded , because its derivative would produce the dirac delta , which can not be integrated with a non - continuous function .
we see that , due to constrains , the faddeev - takhtajan - dubrovin symplectic form , typical for the hamiltonian theory of solitons , arises in a natural way in hamiltonian description of characteristic value problem for standard , hyperbolic equations .
transition between the standard cauchy data and the radiation ( characteristic ) data is a symplectomorphism .
we conclude that the product of @xmath125 and @xmath126 by their derivatives must be integrable on real line @xmath133 , so must belong to @xmath134 . moreover , functions @xmath125 , @xmath126 and @xmath135 , @xmath136 have to belong to mutually dual spaces because they representpositions " and momenta " correspondingly .
this implies : that @xmath137 and , then , @xmath138
. equations imply that we have also : @xmath139 and @xmath140 .
the hamiltonian system we have obtained is conservative , because the total hamiltonian does not depend explicitly on time .
hence , the total energy is conserved : @xmath143 but , due to and , the external energy @xmath144}$ ] and @xmath145 are monotonically increasing function on time .
indeed , they are obtained by integration of a non - negative function @xmath146 over a part of the boundary @xmath70 of the cone which increases when the time increases .
we conclude that the trautman - bondi internal @xmath147}$ ] on the hyperboloid must be monotonically decreasing function on time .
we have also showed that @xmath147}$ ] is positive inside the cone .
it represents the amount of energy still remaining in the system , whereas @xmath144}$ ] and @xmath145 describe already radiated energy .
the above construction , with appropriate modifications , is valid also in four - dimensional minkowski spacetime . consider a one - parameter family of past - oriented light cones : @xmath148 where @xmath149 is the time coordinate of the vertex of @xmath150 .
we introduce new coordinates @xmath151 ( @xmath152 ) , related to minkowskian coordinates @xmath153 in a way analogous to - : @xmath154 where @xmath11 . for @xmath155 ,
the new coordinates parameterize the entire cone @xmath150 and the surfaces @xmath156 correspond to hyperboloids . to derive the hamiltonian description of the wave equation in these coordinates , we begin with the lagrangian : @xmath157 expressing it in terms of new coordinates we obtain : @xmath158 where @xmath159 \delta^{kl } \frac{\partial \hat\varphi}{\partial\xi^k } \frac{\partial \hat\varphi}{\partial\xi^l } \right\ } \left ( \frac{2e^{-\epsilon\tau}}{1-\|\xi\|^2}\right)^2 \ , .\
] ] this , apparently non - autonomous ( i.e. @xmath20-dependent ) , lagrangian becomes autonomous after an appropriate re - scaling of the field variable :
@xmath160 indeed , we obtain the formula analogous to : @xmath161 ^ 2 \\ & - \frac12 \kappa \delta^{kl}\left ( \partial_{\xi^k}\varphi -\frac{2\xi_k}{1-\|\xi\|^2}\varphi\right ) \left ( \partial_{\xi^l}\varphi -\frac{2\xi_l}{1-\|\xi\|^2}\varphi\right ) \ , , \end{aligned}\ ] ] where we denote : @xmath162 \ , .\ ] ] observe that the function @xmath163 differs from the original lagrangian @xmath164 by a complete divergence : @xmath165 where @xmath166 and @xmath167\varphi^2 \ , .\ ] ] this implies that both lagrangians @xmath164 and @xmath67 lead to the same equation of motion for the field @xmath30 , so we use the latter in the sequel . to derive the hamiltonian description , we integrate equation over the volume @xmath168 in the cauchy surface @xmath169 .
we obtain an identity valid for fields satisfying wave equation : @xmath170 where `` dot '' denotes derivative with respect to the new time variable @xmath20 , while @xmath29 is the restriction of the field @xmath30 to the cauchy surface @xmath31 .
moreover , we have introduced momentum @xmath34 , which is a part of cauchy data on this surface .
performing legendre transformation between @xmath35 and @xmath36 we get : @xmath171 where the hamiltonian @xmath172 is defined by @xmath173 and the symplectic form in phase space of cauchy data is given by : @xmath174 the lagrangian implies the following relation between `` velocity '' @xmath35 and momentum @xmath36 : @xmath175 thus , the hamiltonian my be written explicitly in terms of the cauchy data on @xmath176 @xmath177 where @xmath178 one can check that the quadratic form @xmath179 is positive definite for @xmath180 .
but , we have : @xmath181 this implies that hamiltonian is positive for @xmath182 , i.e. inside the cone @xmath7 .
the euler - lagrange equation coincides with the following hamiltonian equations , derived directly from : @xmath183 provided no boundary terms remain after the integration by part of its variation . to assure their cancellation we proceed in a way analogous to the previous section : we take into account missing radiation data on the light cone .
therefore , we extend parametrization beyond the volume @xmath176 , taking into account also the corresponding points on the boundary of the cone : @xmath184 and consider the data @xmath51 on the entire surface @xmath185 . equations imply that @xmath186 , for @xmath187 .
the dynamics consists in transporting the field data @xmath51 over the surface @xmath61 according to the following field equations : @xmath188 where follows from the fact that the momentum @xmath36 is not a scalar field ( like @xmath29 ) but a scalar density .
the above equations can be also derived from the standard hamiltonian formula : @xmath189 where @xmath190 and @xmath67 vanishes identically as a pull - back of the scalar density @xmath68 via the degenerate coordinate transformation .
variation of the above hamiltonian gives : @xmath191 where @xmath192 are given by and , whereas the boundary term comes from integration by parts .
the momentum @xmath36 on the cauchy surface @xmath61 is equal to the pull - back of the differential ( odd ) form @xmath193 to @xmath70 .
moreover , momentum @xmath194 coincides with @xmath36 , as the pull - back of the same form to the the hypersurface @xmath195 , so we obtain the following constraints @xmath196 the phase space of cauchy data on the entire @xmath61 is described by the pairs @xmath78 defined on the whole @xmath197 fulfilling constraints outside of the hyperboloid .
moreover , functions @xmath29 and @xmath36 should satisfy compatibility conditions ( `` corner conditions '' ) at points @xmath198 , because otherwise the total dynamics is not well defined . to formulate these conditions we proceed as in the previous section . summing up formula for @xmath199 and formula for @xmath199
we define the total energy @xmath80 , defined on the total phase space @xmath200 @xmath201 variation of @xmath80 gives us : @xmath202 where the dynamics @xmath84 is given given by ( inside ) and by outside of the sphere @xmath203 .
the global dynamics generated by @xmath80 is well defined if the boundary terms in formula vanish . to analyze the resulting corner conditions we reformulate our hamiltonian description as follows .
taking into account constraint we obtain from : @xmath204 and the corresponding symplectic structure @xmath205 where @xmath206 and @xmath207 denotes the volume element on the two - dimensional unit sphere @xmath208 . changing variables in the integral @xmath209 the remaining variables being unchanged , and denoting @xmath210
we obtain that @xmath211 does not depend on variable @xmath20 ( see formulas and ) .
hence , we have : @xmath212 where @xmath213 expression for external " symplectic form suggests to consider , instead of the phase space @xmath214 with constraint , the phase space of functions defined on the half of the tube @xmath215 , corresponding to negative values of @xmath216 .
we will show in the sequel that `` internal '' data @xmath51 on the hyperboloid can also be represented by canonically equivalent data on the remaining half - tube . the proof is based on the euler - lagrange equations , equivalent to the following identity : @xmath217 where by @xmath218 we denote generalized momenta . integrating equation over any region @xmath219 in spacetime we obtain identity : @xmath220 which holds for any configuration @xmath30 satisfying the field equations .
in particular , let @xmath219 be the set of points lying between the boundary of the cone @xmath221 and the hyperboloid @xmath176 , then @xmath222 where the signs come from the orientation of both surfaces @xmath223 and @xmath224 .
now , we treat these expressions as exterior one - forms on the space of cauchy data on @xmath223 and @xmath224 respectively , and calculate exterior derivative of both sides . because the left - hand - side is already an exterior derivative , its further exterior differentiation gives zero .
this way we prove the identity : @xmath225 equation means that the transition from the space of cauchy data on the hyperboloid @xmath224 to the space of boundary data on the cone @xmath226 , defined by the field dynamics , is a canonical transformation ( a _ symplectomorphism _ ) . using on the cauchy surface @xmath227
we obtain : @xmath228 on the other hand , on @xmath229 we have : @xmath230 where @xmath98 is a function which lives on @xmath223 .
thus equation takes the form @xmath231 where @xmath232 formulae and prove that the global cauchy data can be described by a single function @xmath233 , equal to @xmath234 for @xmath235 and to @xmath236 for @xmath132 .
tailoring these two partial phase spaces into a single phase space , we have to impose compatibility condition ( corner condition " ) at @xmath237 , namely : the symplectic form : @xmath238 must be well defined .
it means that the product of function which lives on a tube @xmath239 and its derivative along the tube must be integrable .
moreover , functions @xmath125 and @xmath240 have to belong to mutually dual spaces because they representpositions " and momenta " correspondingly .
this implies that @xmath241 and @xmath242 . | cauchy initial value problem on a hyperboloid is proved to define a hamiltonian system , provided the radiation data at null infinity are also taken into account , as a part of cauchy data .
the `` trautman - bondi mass '' , supplemented by the `` already radiated energy '' assigned to radiation data , plays role of the hamiltonian function .
this approach leads to correct description of the corner conditions .
* keywords : * trautman - bondi energy , wave equation , initial - characteristic value problem , hamiltonian field theory . |
polarization and reflects the photons in @xmath0 polarization .
, width=264 ] in this section , we introduce the mdi - sarg04 qkd protocol . first , we summarize the assumptions and mathematical definitions made in this paper , and then we describe how the protocol runs .
we assume that each of alice and bob has a phase randomized photon source , _ i.e. _ the vacuum , a single photon , and multi photons are emitted probabilistically .
the probabilities of the @xmath1-photon emission from alice and bob are @xmath2 and @xmath3 , respectively , which satisfy @xmath4 .
we encode the bit information in polarization of photons , and we assume that the preparation of the polarization is precise without any flaw . for simplicity , we consider the asymptotic case to neglect any statistical fluctuation , i.e. , the number of the signals sent by alice and bob is infinite . in our paper , horizontal and vertical polarization states of a single photon
are represented by @xmath5-basis qubit states , namely @xmath6 and @xmath7 , respectively .
we also define @xmath8 ( rectilinear)-basis states as @xmath9 for @xmath10 . by using a creation operator @xmath11 for a single photon in a polarization @xmath12 and the vacuum state @xmath13
, we denote an @xmath1-photon number state with polarization @xmath12 by @xmath14 .
( note that when the subscript @xmath12 is @xmath15 or @xmath16 , it refers to the qubit state rather than the photon number state ) .
other definitions we use are as follows : @xmath17 for @xmath10 and @xmath18 for @xmath19 .
@xmath20 , where @xmath21 , which satisfies @xmath22 for all @xmath23 . @xmath24 and @xmath25 .
we denote @xmath26 .
the protocol runs as follows : ( a1 ) alice and bob choose a bit value @xmath23 and @xmath27 @xmath28 , respectively , and they encode the bit value into the photonic states of their pulses as @xmath29 and @xmath30 .
( a2 ) alice and bob rotate the polarization of their pulses by applying rotation @xmath31 and @xmath32 with randomly - chosen values of @xmath33 and @xmath34 , respectively , where @xmath31 is defined by @xmath35 . after the rotation
, alice and bob send the pulses to eve s measurement unit ( mu ) through quantum channels .
( a3 ) eve performs a measurement on the incoming pulses and announces to alice and bob over the authenticated public channel whether her measurement outcome is successful or not . when the outcome is successful , she also announces types of the successful events , either type1 or type2 .
( a4 ) alice and bob broadcast @xmath36 and @xmath37 , over the authenticated public channel .
if the measurement outcome in ( a3 ) is successful with type1 and @xmath38 , they keep their bit values @xmath23 and @xmath27 in ( a1 ) , and alice flips her bit . if the measurement outcome in ( a3 ) is successful with type2 and @xmath39 , they keep their bit values @xmath23 and @xmath27 in ( a1 ) . in all the other cases ,
they discard their bit values .
type1 ( @xmath40&@xmath41 or @xmath42&@xmath43 ) & @xmath44 + type2 ( @xmath40&@xmath43 or @xmath42&@xmath41 ) & @xmath45 + ( a5 ) alice and bob repeat from ( a1 ) to ( a4 ) until the number of the successful events with rotation @xmath46 in type1 becomes @xmath47 and @xmath39 in type2 becomes @xmath48 .
let @xmath49 be the number of the successful detection event of type @xmath23 .
alice and bob announce randomly - chosen @xmath50 bits over the authenticated public channel , where @xmath51 is much smaller than 1 , and estimate the error rate @xmath52 in the remaining code bits .
the estimated number of the bit error in the code bits is denoted by @xmath53 .
( a6 ) alice and bob perform error correction and privacy amplification on the remaining @xmath54 bits by their discussion over the public channel . as a result
, they share a final key of length @xmath55 . at eve s mu in ( a3 ) , honest eve performs the bell measurement in order to establish quantum correlations between alice and bob to generate the key . in fig .
[ fig : actual ] , the experimental setup for the bell measurement is depicted .
it employs a half beam splitter ( bs ) , two polarization bss ( pbss ) , and the photon detectors . in the case where both alice and bob emit a single photon
, the simultaneous photon detection events matching the pattern type1 ( type2 ) , listed in table [ tbl : type ] , corresponds to the detection of @xmath44 ( @xmath45 ) .
we emphasize that in the security proof we assume that eve is malicious and has a control over the quantum channels , and all the bit errors are attributed to the consequence of the eavesdropping . in prepare & measure setting ,
the sarg04 protocol is different from the bb84 protocol only in the post - processing part , i.e. , no modification is needed in the experimental setup . in the mdi setting , however , the experimental setup for the bb84 protocols @xcite can not be directly used in mdi - sarg04 as it induces a high bit error rate , and this is a significant qualitative difference of mdi setting from prepare & measure setting , implying that not all the prepare & measure qkd protocols can not be directly converted to mdi setting .
therefore , we need to consider an alternative experimental scheme for mdi - sarg04 . in this section ,
we first discuss why the setup for mdi - bb84 gives the high bit error rate , and then we propose alternative experimental schemes for mdi - sarg04 . for the explanation we denote by @xmath56 the joint probability that eve receives @xmath1 and @xmath57 photons from alice and bob , respectively , and obtains the successful measurement outcome . note that while we do not deal with the types of eve s successful outcomes separately , the following discussion is valid for both types .
for simplicity , we neglect all the losses , including those in the quantum channel and the photon detectors , and therefore we can also regard @xmath56 as @xmath58 , which is the joint probability that alice and bob respectively emit @xmath1 and @xmath57 photons and eve obtains the successful measurement outcome . like in the mdi - bb84 protocols , we assume that alice and bob use a phase randomized weak coherent light whose average photon number is much smaller than 1 .
thus , we have @xmath59 for @xmath60 . for simplicity , we assume eve is honest , namely the bit error rate for @xmath61 is zero , and all photon detectors have unit quantum efficiency and no dark counting . in the following ,
we show that even with this simplification favorable to alice and bob , no significant key is expected . to see this
, we consider the bit error rate , and the total bit error rate @xmath62 is expected to be @xmath63 where @xmath64 is the bit error probability under the condition that alice emits @xmath1 photons and bob emits @xmath57 photons , and eve announces the successful outcome .
note that equation ( [ eq : etotapprox ] ) holds in both the mdi - bb84 and mdi - sarg04 protocols .
it is clear from equation ( [ eq : etotapprox ] ) that the bit error is caused by the case where one party emits two photons and the other party emits the vacuum .
it is also clear that @xmath65 can not be zero since the vacuum emission carries no bit information . in the case of mdi - bb84 , this event is always discarded from the sifted key , and consequently the bit error rate in the key generation basis , i.e. , rectilinear basis , is zero .
this is so because the two - photon states @xmath66 and @xmath67 , which contribute to the bit values , are orthogonal and they never produce the successful outcomes in eve s projection measurement for the basis @xmath68 .
therefore , in the experiment of mdi - bb84 , the bit error rate is very small . in the case of mdi - sarg04 , however , two states @xmath69 and @xmath70 consisting bit values are not orthogonal .
this means that the two - photon emission contributes to the successful outcome .
more precisely , @xmath71 holds from the direct calculation of @xmath72 . and
@xmath73 hold for any linear loss transmittance channel . ]
therefore , we conclude that the use of the phase randomized coherent light source gives no significant key in mdi - sarg04
. , @xmath74 and other problematic photon number configurations . ( a ) eve performs the qnd measurements on the pulses from alice and bob , and she does not perform the interference measurement for @xmath75 or @xmath76 .
eve accepts only when @xmath77 and @xmath78 are satisfied .
( b ) a quasi single - photon source used by alice and bob , which is composed of the heralded spdc process . when detector @xmath79 clicks , alice / bob sends her / his pulse at the remaining mode to eve s mu . , width=302 ] in order to generate a key in the mdi - sarg04 protocol , eve s mu or the photon sources should be modified such that the probability of obtaining the successful outcome due to the two photons and the vacuum state is suppressed . in order to suppress the probability , we propose two experimental setups : ( i ) eve performs quantum nondemolition ( qnd ) measurement on the two incoming pulses from alice and bob just before mixing them as shown in fig .
[ fig : qnd+parame](a ) .
the qnd measurement discriminates whether the photon number in the pulse is 0 , 1 or more .
eve accepts only the case where @xmath77 and @xmath78 and discards the other cases with multiple photons .
thanks to the qnd measurement , the total bit error rate is suppressed even if the phase randomized coherent light is used as a photon source .
( ii ) without the modification of eve s mu , alice and bob replace the phase randomized coherent light by a heralded single photon source based on a spontaneous parametric down - conversion ( spdc ) and a threshold photon detector ( see fig . [
fig : qnd+parame](b ) ) .
this dramatically reduces the probabilities of the events of @xmath80 and @xmath74 .
we will show that these setups enable us to generate the key later .
in this section , we discuss the unconditional security proof ( i.e. , the security proof against most general attacks ) of our scheme .
the security proof is independent of the specific device models like in fig .
[ fig : qnd+parame ] , namely it is valid for any eve s mu and any photon sources of alice and bob .
our proof employs the security proof based on the entanglement distillation protocol ( edp ) @xcite , where the distillation of @xmath81 is considered for type1 and that of @xmath82 is considered for type2 .
the proposed edp - based virtual protocol , which is equivalent to the mdi - sarg04 qkd from eve s viewpoint , runs as follows . ( v1 ) alice and bob prepare @xmath83 , where @xmath84 for @xmath85=a , b . here
@xmath86 and @xmath87 are randomly chosen .
the probability distribution of the photon number is equal to that of the photon source in the actual protocol .
alice and bob send the @xmath1 and @xmath57 photon states in @xmath88 and @xmath89 to eve s mu , respectively .
( v2 ) eve performs a measurement on the photons coming from alice and bob , and announces to them whether the measurement is successful ( including the type of the event ) or not . if the measurement result is not successful , alice and bob discard their qubits .
( v3 ) alice and bob broadcast the labels @xmath36 and @xmath37 , respectively . in the cases of @xmath90 with the announcement of type2 or @xmath91 , alice and
bob discard their qubits .
( v4 ) alice and bob repeat ( v1 ) ( v3 ) many times until the number of the successful events for @xmath92 becomes @xmath93 for @xmath94 , where @xmath23 corresponds to the type of the events .
( v5 ) let @xmath95 be the number of the successful detection event for type @xmath23 .
alice and bob announce randomly chosen @xmath50-photon pairs over the authenticated public channel , where @xmath51 is much smaller than 1 , and then they perform @xmath5-basis measurement on their qubits of the chosen pairs . by sharing their measurement results over the authenticated public channel
, they estimate the bit error rate on the code qubits denoted by @xmath96 . as a result
, the number of the bit error is estimated to be @xmath97 .
( v6 ) they estimate the upper bound on the phase error rate @xmath98 for @xmath1 and @xmath57 photons from the bit error rate @xmath99 for @xmath1 and @xmath57 photons . here
the phase error is defined by the bit error that would have been obtained if they had measured the qubit pairs by @xmath8 basis , which is the complementarity basis of the computational basis .
( v7 ) when the bit and the phase errors are smaller than a threshold value for entanglement distillation , they perform the distillation for @xmath54 qubit pairs . for the cases of type1 and type2 , they distill the photon pairs in states @xmath44 and @xmath45 , respectively .
we denote the number of the distilled maximally entangled qubit pairs as @xmath100 .
finally , by performing @xmath5-measurements on the distilled photon pairs , they obtain the key . the important quantities in the proof is the bit and phase errors , and the phase error rate determines the amount of privacy amplification . the bit error rate in the code bits of the virtual protocol , which is exactly the same as the one of the actual protocol ,
is directly estimated by test bits .
on the other hand , the phase error rate is defined by the complementary basis @xmath8 , which alice and bob never employ , and therefore this rate is not directly estimated in the protocols .
note that we are allowed to work on alice s @xmath1-photon emission and bob s @xmath57-photon emission separately , because alice s and bob s photon sources in the protocols are phase randomized . in the following subsections , we present the estimation of the phase error rates for the cases of type1 and type2 independently .
we derive an upper bound on the phase error @xmath101 for @xmath94 , where the superscript @xmath102 denotes @xmath61 and the subscript represents the type of the successful outcome , and derive an upper bound on the phase error @xmath103 .
we show that in the case of @xmath104 , no key can be generated when the probability of eve s successful outcome for the two - photon emission event is smaller than 1/16 .
we note that in the cases of either @xmath105 or @xmath106 , eve can perform an unambiguous state discrimination to one of the three - photon emission part @xcite , and thus we can not extract the key from such events , given that the channel is lossy enough .
fnally , we note that given the phase error rates , @xmath107 and @xmath108 , the asymptotic key rate for type @xmath23 is written by @xcite @xmath109+q_i^{(1,2)}[1-h(e^{(1,2)}_{i,\u{ph } } ) ] \nonumber\\ & + q_i^{(2,1)}[1-h(e^{(2,1)}_{i,\u{ph}})]-f(e_i^\u{tot})q_i^\u{tot}h(e_i^\u{tot } ) .
\label{eq : keyrate}\end{aligned}\ ] ] here @xmath110 is the binary shannon entropy . by the analysis based on the virtual protocol , we give the phase error estimation formula for @xmath111 and @xmath113 .
the estimation is performed for type1 and type2 , separately , and we detail the derivation of the phase error estimation in methods section . in the case of type 1 , we have @xmath114 for @xmath111 and @xmath115 for @xmath113 , where @xmath116 in the case of type 2 , we have @xmath117 for @xmath111 and @xmath118 for @xmath113 , where @xmath119 is the maximal solution of the following equation for @xmath16 @xmath120 and ( b ) for @xmath113 .
, width=264 ] we depict the dependencies of the phase error rates on the bit error rates in fig .
[ fig : pb ] . for the case of @xmath104 , the key can not be obtained for @xmath104 in type1 and type2 by giving an explicit eve s attack which give a phase error of 0.5 , as long as the success probability of eve s measurement conditioned that both alice and bob emit two photons is not larger than 1/16 .
we show the proof in methods section . we will prove that we can not generate a key from @xmath104 in the virtual protocol , and it follows that we can not generate a key from @xmath104 in the actual protocol either . to see this , note that the virtual protocol differs from the actual protocol only in the way to prepare the state , and the state prepared and post data - processing are exactly the same in both protocols .
in other words , only the local operation needed in state - preparation process by the legitimated parties are different in the two protocols . by recalling that any local operation can not convert a separable state into a non - separable state , we conclude that if we can not generate a key from a virtual protocol , then we can not generate a key from the actual protocol . here
we show the results of the key generation rate for the two experimental setups as shown in figs .
[ fig : qnd+parame](a ) and ( b ) by using typical experimental parameters taken from gobby - yuan - shields ( gys ) experiment @xcite , where the quantum efficiency and the dark counting of the all detectors in eve s mu are @xmath121 and @xmath122 , respectively , the loss coefficient of the quantum channel is @xmath123db / km , and the inefficiency of the error correcting code is @xmath124 . in the simulation
, we use infinite number of decoy states @xcite in order to obtain @xmath125 , @xmath126 , @xmath127 and @xmath128 . assuming that the bit error is stemmed only from dark countings of the detectors , we ignore the other imperfections such as the misalignment of the devices .
we also assume that the mean photon numbers of the signal pulses prepared by alice and bob are the same , and the mu in eve is the middle of alice and bob .
the mean photon number for the signal is optimized for maximizing the key generation rate at each distance . by using equation ( [ eq : keyrate ] ) with the above parameters and assumptions
, we calculate the key generation rate as a function of the distance between alice and bob ( i ) when eve postselects the events with @xmath77 and @xmath78 with the qnd measurement as shown in fig .
[ fig : qnd+parame](a ) and alice and bob use the coherent pulses , and ( ii ) when eve uses the mu in fig .
[ fig : actual ] and alice and bob use quasi single photon sources prepared by the spdc in fig .
[ fig : qnd+parame](b ) . only .
middle : the key rate of the mdi - sarg04 protocol from @xmath111 , @xmath112 and @xmath129 .
top : the key rate of the mdi - bb84 protocol .
( b ) the upper and lower solid lines are the key rates from @xmath111 , @xmath112 and @xmath129 for type1 and type2 , respectively .
the upper and lower dashed lines are the key rates from @xmath111 for type1 and type2 , respectively .
, width=302 ] .
for the key rates from @xmath111 , the three lines show the mean photon number when we consider only type1 , both types and only type2 from the top .
the mean photon numbers for the key rates from @xmath111 , @xmath112 and @xmath129 show a similar tendency .
the dashed line is for the mdi - bb84 protocol .
, width=264 ] _ case ( i ) _ the simulation result of the key rate is shown in fig . [
fig : qnd](a ) , and the mean photon number which maximizes the key rate is shown in fig .
[ fig : qnd_alpha ] .
we also plot the key rates of type1 and type2 separately in fig .
[ fig : qnd](b ) . the details for obtaining these figures
are shown in supplementary .
when the distance is zero , since there is no photon loss before the bs and the multi - photon emissions are excluded , the events of multi - photon input have no contribution to the key rate .
in fact , in fig . [
fig : qnd](a ) , the two key rates at zero distance obtained from only @xmath111 and from both @xmath111 , @xmath112 and @xmath129 are exactly the same . when the distance becomes longer , we see from fig . [
fig : qnd_alpha ] that the contribution of the multi photons becomes larger .
for the key rate from only @xmath111 , the mean photon number is monotonically decrease because the multi - photon emissions give only adverse effect . on the other hand ,
when we extract the key additionally from the multi photons , the mean photon number does not decrease monotonically , which shows an advantage in using multi - photon emission . , and
the successful events of @xmath113 and @xmath130 give little contribution to the key rate .
this is so because the probability of the two - photon component in the heralded photon source is negligibly small compared with the probability of the single - photon component .
the lines are for mdi - bb84 ( black ) , for both types ( red ) , type1 ( green ) and type2 ( blue ) of the mdi - sarg04 .
, width=264 ] .
the upper line ( black ) is the mean photon number for the mdi - bb84 protocol .
the lower three lines are for the key rates of the mdi - sarg04 protocol obtained from type1 ( green ) , both types ( red ) and type2 ( blue ) from the top . ,
width=283 ] _ case ( ii ) _ alice and bob use quasi single photon sources by spdc as shown in fig .
[ fig : qnd+parame](b ) .
detector @xmath79 is the same as that used in eve s mu , namely it is the threshold detector with the quantum efficiency of @xmath121 and the dark counting of @xmath122 .
eve s mu is the same as that shown in fig .
[ fig : actual ] .
the key rate is shown in fig .
[ fig : pdc ] . the details for calculating the key rates
are shown in supplementary . the mean photon number which maximizes the key rate
is shown in fig .
[ fig : pdc_alpha ] . from fig .
[ fig : pdc ] , we see that the key rate only from type1 and that both from type1 and type2 intersect . for the distribution distance
longer than the cross point , type2 has no contribution of the key , which is shown by the blue line in the figure , and therefore it is better to generate a key from type1 only . from fig .
[ fig : pdc_alpha ] , we see that the mean photon number is very small .
this is so because the use of larger mean photon numbers results in two - photon emission , which increases the bit error rate . from all the figures of the key rate
, one sees that the key rates of mdi - sarg04 are lower than those of mdi - bb84 .
this tendency holds also for prepare & measure sarg04 @xcite , and the higher phase error rates of sarg04 protocol than that of bb84 is the main reason of this tendency .
in recent years , many proposals and experimental demonstrations of the mdi qkd have been studied .
so far , all of them except for the continuous - variable qkd protocol @xcite are based on the bb84 protocol .
we first proved the unconditional security of the mdi qkd based on the sarg04 protocol . in our security proof
, we gave the upper bounds on the phase error rate when alice and bob emit single photons and when one party emit one photon and the other half emit two photons .
for the case of the two photon emissions from both parties , we proved that a key can not be generated as long as the probability of success in her measurement conditioned that both alice and bob emit two photons is not larger than 1/16 .
another important issue to be addressed in mdi setting is what kind of measurement should be implemented experimentally at eve s laboratory .
we have shown that the measurement setup for bb84 in mdi setting can not be used in sarg04 in mdi setting , and we proposed two measurement schemes for mdi sarg04 . in the first one , alice and bob use heralded single photon sources prepared by spdc . in the second one , eve performs qnd measurement on the two pulses coming from alice and bob individually . in our simulation
based on these experimental setups , it was confirmed that these setup can generate a key .
while eve accesses only the photons in modes @xmath131 and @xmath132 in the actual protocol , we pessimistically suppose that she can prepare any state in @xmath133 and @xmath134 for simplicity of the proof .
, width=321 ] here , we give the phase error estimation for @xmath61 . for this , it is convenient to recall a mathematical property of the maximally entangled state that @xmath135 is satisfied for any operator @xmath136
. therefore @xmath137 in ( v1 ) is expressed as @xmath138 , where @xmath139 .
physically , this identification can be interpreted as the situation where @xmath140 is prepared by each of the parties , the filtering operation , of which successful case is described by @xmath141 , is applied , and then each party sends the photons to eve only when the filtering operation succeeds ( also see fig . [
fig : nm11 ] ) . for the simplicity of the security proof
, we make an overestimation of eve s ability in terms of the accessibility of the photons , namely , we imagine eve who has a direct access to photons of @xmath133 and @xmath134 rather than @xmath131 and @xmath132 , and she can prepare any joint state of the photons of @xmath133 and @xmath134 . for later convenience ,
we denote by @xmath142 the state prepared by eve . in the following ,
we first discuss the case of type1 .
we define @xmath143 as the joint probability that the photons in @xmath142 pass through the filtering operation and induces a bit / phase error to the state @xmath44 after the rotation . here
@xmath144 and @xmath145 are povm elements of the bit and phase error measurements on @xmath142 , respectively .
the probability that the two photons in @xmath142 pass through the successful filtering operation is described by @xmath146 , where the povm element of the successful filtering operation on the two photons is @xmath147 where @xmath26 .
the povms for the bit and the phase errors are written as @xmath148 applying the bayes rule , the bit error rate @xmath149 and the phase error rate @xmath150 in the final state in modes @xmath151 and @xmath152 are described by @xmath153 the phase error estimation can be established by directly writing down the explicit form of equation ( [ eq : pibit11type1 ] ) comparing each matrix element , and one can conclude that @xmath154 thus from equations ( [ eq : pifil11type1 ] ) and ( [ eq : phasebitperfil11type1 ] ) , the phase error rate is precisely estimated , by using the bit error rate , as shown in equation ( [ eq:11phasebittype1 ] ) . thanks to azuma s inequality @xcite , equation ( [ eq:11phasebittype1 ] ) holds for any eavesdropping including coherent attacks .
next , we estimate the phase error rate for type2 . because only the cases of @xmath155 are accepted for type2 , the definition of the povm element of the successful filtering operation is changed to @xmath156 and the probability that the two photons in @xmath142 pass through the successful filtering operation
is expressed by @xmath157 .
we describe a joint probability that the two photons in @xmath142 pass through the successful filtering operation after the rotation and then the photons in modes @xmath151 and @xmath152 have a bit / phase error to the state @xmath45 by @xmath158 . like in the case of type1 ,
the povm elements of @xmath159 and @xmath160 are written by @xmath161 and @xmath162 by using the bayes rule , the bit / phase error rate of @xmath163 in the final state is expressed by @xmath164 in order to see the relation between the bit and phase error rates , we consider an inequality to bound the phase error as @xmath165 , where @xmath166 is a real number , which is equivalent to @xmath167 for @xmath168 . by considering a non - negativity condition of @xmath167
, we see that this inequality always holds when @xmath169 , and therefore , we have the relation between the phase error rate and the bit error as shown in equation ( [ eq:11phasebittype2 ] ) . and @xmath171 .
by eve s announcement for the successful measurement on the photons in @xmath88 , @xmath89 and @xmath172 , the three - photon state @xmath173 is prepared .
, width=321 ] below , we give the phase error estimation for @xmath170 and @xmath171 . by using the similar argument as @xmath61 , @xmath174 at bob s side in ( v1 )
is defined by @xmath175 as in fig .
[ fig : nm12 ] , where @xmath176 . here
we note that two - photon emission part is simulated by preparing two pairs of @xmath140 followed by the rotation and the filtering operation on two qubits ( see also fig .
[ fig : nm12 ] ) . in this virtual protocol
, while we consider two photons in different modes , this never underestimates eve s ability .
this is so because two photons in the different modes and two photons in a single mode can be converted just by an unitary transformation as @xmath177 .
we note that because the photon in mode @xmath178 is in @xmath179 after the filtering operation , and it is decoupled from all the other systems , the component is not related to the security proof .
again , we employ the overestimation that eve has the control over the state of the systems of @xmath180 , @xmath181 and @xmath182 , and we denote the three - photon state by @xmath173 , which is prepared by eve after her announcement of the success . like in the case for @xmath61
, we estimate a phase error for each case of type1 and type2 separately . for type1 ,
define a povm element of the successful filtering operations on @xmath173 as @xmath183 here the probability of the successful filtering operation is written by @xmath184 .
we define @xmath185 as a joint probability that the photons in @xmath173 pass through the filtering operation and induces a bit / phase error to the state @xmath44 after the rotation . the successful filtering operation
after the rotation is performed on the two photons in @xmath173 and then the photons in modes @xmath151 and @xmath152 have a bit / phase error to the state @xmath44 . here
, povm elements of @xmath186 are written by @xmath187 the actual bit error rate @xmath188 and phase error rate @xmath189 for @xmath170 and @xmath171 are obtained by accommodating the normalization by @xmath190 , and they are expressed as @xmath191 in order to see the relation between the bit and phase error rates , we consider an inequality to bound the phase error as @xmath192 , where @xmath193 and @xmath194 are real numbers . by using equations ( [ eq : pifil12type1 ] ) ( [ eq : eebit12type1 ] ) , and the linearity of the trace
, we obtain an inequality as @xmath195 , which is satisfied when @xmath196 therefore the phase error rate is given by using the bit error as shown in equation ( [ eq:12phasebittype1 ] ) . for type2 , we define a povm element @xmath197 of the successful filtering operation by limiting the summation only to @xmath198 and by replacing @xmath199 with @xmath200 in equation ( [ eq : pifil12type1 ] ) . the probability of the successful filtering operation is described by @xmath201 .
we also define joint probabilities of @xmath173 passing through the filtering and presenting bit and phase errors to the state @xmath45 by @xmath202 and @xmath203 .
we define the povm element of @xmath204 by limiting the summation only to @xmath198 and replacing @xmath199 with @xmath200 in equation ( [ eq : pibit12type1 ] ) , and that of @xmath205 is defined by limiting the summation only to @xmath198 , replacing @xmath199 with @xmath200 , and @xmath206 with @xmath207 for mode @xmath152 . in a similar manner as the case of type1 for @xmath170 and @xmath171 , by using the bit error rate defined by @xmath208 , the phase error rate as @xmath209 and real numbers @xmath210 and @xmath211 , we consider an inequality as @xmath212 , which leads to @xmath213 . from this inequality ,
we obtain @xmath214 , where @xmath119 is the maximal solution of equation ( [ eq : gs2 ] ) .
using @xmath119 , we have the relation between the phase error rate and the bit error as shown in equation ( [ eq:12phasebittype2 ] ) . .
by eve s announcement for the successful measurement on the photons in @xmath88 , @xmath215 , @xmath89 and @xmath172 , the four - photon state @xmath216 is prepared .
, width=321 ] for the case of @xmath104 , like in the previous subsection , @xmath217 at alice s side in ( v1 ) is obtained by @xmath218 , and @xmath174 at bob s side is prepared by the same manner . as a result , the virtual protocol for @xmath104 is equivalent to the successful situation of the filtering operations , which we depict in fig . [
fig : nm22 ] .
we denote the state of alice s and bob s four qubits after eve s successful announcement by @xmath219 . in the following , we prove that the key can not be obtained for @xmath104 by giving an explicit eve s attack , namely we give explicit states of @xmath180 , @xmath220 , @xmath181 and @xmath182 which give a phase error of 0.5 .
the key ingredient is that while eve can not manipulate these four qubits , she conclusively prepare such a state on their qubits by announcing the success of her measurement only when she succeeds an eavesdropping measurement on eve s photons @xmath88 , @xmath215 , @xmath89 and @xmath172 .
this attack gives eve the perfect information on the bit values when her measurement succeeds .
for type1 , the probability of the successful filtering operation is expressed by @xmath221 , where @xmath222 the joint probability , that the filtering operation succeeds and the bit / phase error to the state @xmath44 is detected , is expressed by @xmath223 , where @xmath224 the bit / phase error rate is expressed as @xmath225 .
one can confirm by direct calculation that a four - photon state of @xmath226 gives @xmath227 and @xmath228 , and another four - photon state @xmath229 , which is orthogonal to @xmath230 , gives @xmath231 and @xmath228 .
therefore , although eve can not touch the four modes @xmath180 , @xmath181 , @xmath220 and @xmath182 , eve can prepare the two states by a projective measurement on the four photons in @xmath88 , @xmath89 , @xmath215 and @xmath172 as @xmath232 .
one sees this fact from the equation @xmath233 , which also implies that the preparation succeeds with a probability of 1/16 .
thus a malicious eve achieves the phase error rate of 0.5 for any bit error rate by distributing these states with a relevant probability .
this means that the state in @xmath234 and @xmath235 is separable , and it follows that no key can be generated for @xmath236 , where @xmath237 is the probability of eve s successful detection of type @xmath23 conditioned that both alice and bob emit two photons . for type2 , with the same fashion as the case of @xmath170 and @xmath171 , povm elements @xmath238 for the successful filtering operation and @xmath239 for the bit / phase error are defined by replacing the summation range of @xmath36 , the prefactor and the proper inversion of the bit value of the projection in equations ( [ eq : pifil22 ] ) and ( [ eq : pibit22 ] ) .
we consider the following four orthogonal four - photon states for systems @xmath180 , @xmath181 , @xmath220 and @xmath182 @xmath240 , @xmath241 , @xmath242 and @xmath243 .
each state can be prepared by eve s projective measurement @xmath244 on the four photons in @xmath88 , @xmath89 , @xmath215 and @xmath172 with a probability of @xmath245 . by calculating the error probabilities
, we see that mixed states @xmath246 and @xmath247 give @xmath248 and @xmath249 , respectively .
therefore eve achieves any bit error rate below 0.5 while keeping @xmath250 by distributing the above two mixed states with an appropriate probability . as a result
, we conclude that for @xmath251 , the key can not be obtained .
this work was supported by the funding program for world - leading innovative r & d on science and technology ( first ) , mext grant - in - aid for scientific research on innovative areas 21102008 , mext grant - in - aid for young scientists(a ) 23684035 , jsps grant - in - aid for scientific research(a ) 25247068 and ( b ) 25286077 .
kt acknowledges support from the national institute of information and communications technology ( nict ) of japan ( project `` secure photonic network technology '' as part of `` the project uqcc '' ) .
99 mayers , d. unconditional security in quantum cryptography .
_ j. acm _ * 48 * ( 3 ) , 351 ( 2001 ) .
lo , h. k. & chau , h. f. unconditional security of quantum key distribution over arbitrarily long distances .
_ science _ * 283 * , 2050 ( 1999 ) .
shor , p. w. & preskill , j. simple proof of security of the bb84 quantum key distribution protocol .
lett . _ * 85 * , 441 ( 2000 ) .
lydersen , l. _ et al .
_ hacking commercial quantuk cryptography systems by tailored blight illumination .
_ nature photonics _ * 4 * , 686 ( 2010 ) .
lo , h. k. , curty , m. & qi , b. measurement - device - independent quantum key distribution .
_ _ * 108 * , 130503 ( 2012 ) .
bennett , c. h. & brassard , g. quantum cryptography : public key distribution and coin tossing .
_ in proceedings of ieee international conference on computers , systems , and signal processing _ 175 - 179 ( ieee press , new york , 1984 ) .
tamaki , k. , lo , h. k. , fung , c. h. f. & qi , b. phase encoding schemes for measurement - device - independent quantum key distribution with basis - dependent flaw .
_ _ * 85 * , 042307 ( 2012 ) .
xu , f. , curty , m. , qi , b. & lo , h. k. practical aspects of measurement - device - independent quantum key distribution .
_ new j. phys . _ * 15 * , 113007 ( 2013 ) .
tang , z. _ et al_. experimental demonstration of polarization encoding measurement - device - independent quantum key distribution . _ arxiv_:1306.6134 .
liu , y. _ et al_. experimental measurement - device - independent quantum key distribution .
_ _ * 111 * , 130502 ( 2013 ) .
bennett , c. h. quantum cryptography using any two nonorthogonal states .
_ _ * 68 * , 3121 ( 1992 ) .
inoue , k. , waks , e. & yamamoto , y. differential phase shift quantum key distribution .
_ _ * 89 * , 037902 ( 2002 ) .
gisin , n. towards practical and fast quantum cryptography .
_ arxiv : quant - ph/_0411022 .
scarani , v. , acin , a. , ribordy , g. & gisin , n. quantum cryptography protocols robust against photon number splitting attacks for weak laser pulse implementations . _ _ * 92 * , 057901 ( 2004 )
. sasaki , m. _ et al .
_ field test of quantum key distribution in the tokyo qkd network .
express _ * 19 * , 10387 ( 2011 ) .
huttner , b. , imoto , n. , gisin , n. & mor , t. quantum cryptography with coherent states .
a _ * 51 * , 1863 ( 1995 ) .
brassard , g. , ltkenhaus , n. , mor , t. & sanders , b. c. limitations on practical quantum cryptography .
_ _ * 85 * , 1330 ( 2000 ) .
tamaki , k. & lo , h. k. unconditionally secure key distillation from multi - photons .
_ _ * 73 * , 010302(r ) ( 2006 ) .
koashi , m. security of quantum key distribution with discrete rotational symmetry .
_ arxiv : quant - ph/_0507154 .
bennett , c. h. , divincenzo , d. p. , smolin , j. a. & wootters , w. k. mixed state entanglement and quantum error correction .
_ _ * 54 * , 3824 ( 1996 ) .
chefles , a. unambiguous discrimination between linearly independent quantum states .
lett . a _ * 239 * , 339 ( 1998 ) .
chefles , a. unambiguous discrimination between linearly dependent states with multiple copies .
_ _ * 64 * , 062305 ( 2001 )
. gottesman , d. , lo , h. k. , ltkenhaus , n. , & preskill , j. security of quantum key distribution with imperfect devices .
_ quantum information and computation _ * 5 * , 325 ( 2004 ) .
gobby , c. , yuan , z. l. & shields , a. j. quantum key distribution over 122 km of standard telecom fiber . _
phys . lett .
_ * 84 * , 3762 ( 2004 ) . lo , h. k. , ma , x. , & chen , k. decoy state quantum key distribution .
_ _ * 94 * , 230504 ( 2005 ) .
fung , c. h. f. , tamaki , k. , & lo , h. k. on the performance of two protocols : sarg04 and bb84 .
_ _ * 73 * , 012337 ( 2006 ) .
li , z. , zhang , y. c. , xu , f. , peng , x. , & guo , h. continuous - variable measurement - device - independent quantum key distribution .
_ arxiv_:1312.4655 .
azuma , k. weighted sums of certain dependent random variables .
_ tohoku math .
j. _ * 19 * , 357 ( 1967 ) .
here we describe the details of our simulation .
the quantum efficiency and the dark counting of the detectors are @xmath121 and @xmath122 , respectively .
the loss coefficient of the quantum channel is @xmath123db / km . we denote @xmath252 and @xmath253 .
we define that @xmath254 is a probability that the photons are detected as the successful event of type @xmath23 conditioned that alice and bob emit @xmath1 and @xmath57 photons in the states @xmath255 and @xmath256 for @xmath257 , respectively .
@xmath258 is the probability of the successful detection of type @xmath23 conditioned that alice and bob emit photons in @xmath255 and @xmath256 , respectively . assuming that eve is in the middle of alice and bob , the channel transmittance to eve from alice is the same as that from bob .
denoting that @xmath259 is the distance between alice and bob , the channel transmittance for alice and bob is @xmath260 in the following , we give the experimental data for the simulation ( i ) when eve postselects the events with @xmath77 and @xmath78 by the qnd measurement before mixing the pulses from alice and bob ( see fig .
2(a ) ) , and ( ii ) when alice and bob use quasi single photon sources by the spdc ( see fig .
2(b ) ) .
each of alice and bob uses a phase randomized weak coherent pulse with the mean photon number of @xmath261 .
the probability @xmath262 for @xmath1-photon emission is @xmath263 for later use , we define the equations @xmath264 in the following , we give @xmath265 and @xmath266 . from ref .
@xcite , the probability distribution function of the thermal state conditioned that the detector @xmath79 clicked in fig .
2(b ) is @xmath271 where @xmath272 is the probability that the detector @xmath79 clicks , which is described by @xmath273 by defining @xmath274 , the probability that @xmath1 photons exist before the bs conditioned on the click of @xmath79 is @xmath275 for type1 , the relevant equations for @xmath276 and @xmath277 are expressed by @xmath278 where @xmath279 we note that in equation ( [ eq : pes ] ) , we took the pessimistic scenario that all of the events for @xmath75 and @xmath76 are detected as the bit error . here
@xmath280 are given by @xmath281 | the measurement - device - independent quantum key distribution ( mdi qkd ) was proposed to make bb84 completely free from any side - channel in detectors .
like in prepare & measure qkd , the use of other protocols in mdi setting would be advantageous in some practical situations . in this paper
, we consider sarg04 protocol in mdi setting .
the prepare & measure sarg04 is proven to be able to generate a key up to two - photon emission events . in mdi
setting we show that the key generation is possible from the event with single or two - photon emission by a party and single - photon emission by the other party , but the two - photon emission event by both parties can not contribute to the key generation
. on the contrary to prepare & measure sarg04 protocol where the experimental setup is exactly the same as bb84 , the measurement setup for sarg04 in mdi setting can not be the same as that for bb84 since the measurement setup for bb84 in mdi setting induces too many bit errors . to overcome this problem , we propose two alternative experimental setups , and we simulate the resulting key rate .
our study highlights the requirements that mdi qkd poses on us regarding with the implementation of a variety of qkd protocols .
the security of quantum key distribution ( qkd ) can be guaranteed based on some mathematical models of the users devices @xcite .
unfortunately , the actual devices do not necessarily follow mathematical models , and we need to close the gap ( side - channel ) between the actual device and the mathematical model to implement secure qkd systems in practice . among side - channels ,
the side - channel of a photon detector seems to be most easily exploited by an eavesdropper ( eve ) since it accepts any input from eve who can generate an arbitrary optical state such that it causes an unexpected behavior in the detector .
in fact , the famous bright - pulse illumination attacks are based on side - channel in detectors @xcite . in order to countermeasure such attacks , measurement - device - independent ( mdi ) qkd @xcite
was proposed to make bb84 @xcite free from any possible side - channel in a detector . in mdi qkd
, alice and bob do not perform any measurement but only send quantum signals to be measured by eve . therefore
, bit strings generated by alice and bob are free from side - channels in photon detectors since they do not employ photon detectors . since its invention , mdi qkd has been actively studied both theoretically @xcite and experimentally @xcite . as is the case in prepare & measure scheme , implementation of protocols other than bb84 in mdi
setting could be suitable for some practical situations .
in fact , many experiments for non - bb84 type prepare & measure schemes , including b92 @xcite , dps qkd @xcite , coherent one - way protocol @xcite , sarg04 @xcite , etc , have been reported @xcite . therefore , it is useful in practice to use non - bb84 type protocols in mdi setting , and in this paper we consider to use sarg04 protocol in mdi setting , which we refer to as mdi sarg04 .
sarg04 was originally proposed to make bb84 robust against photon number splitting ( pns ) attacks @xcite just by changing the classical post - processing part in bb84 .
it is proven that sarg04 can indeed generate a key from two - photon emission event by alice in addition to single - photon emission event @xcite , showing robustness against pns attack in some parameter regimes . note in mdi setting is that both alice and bob are the sender of the signals , and as a result , the information leakage from the signals seems to be larger than the one in prepare & measure setting .
therefore , it is not trivial whether both single and two - photon emission events can contribute to the key generation or not .
our work answers this question , and we have found that the single - photon emission event by both alice and bob , or single - photon and two - photon emission by each of alice and bob can contribute to generating a key , but two - photon emission by the two parties can not make the contribution when a probability of eve s announcement of the successful measurement for the two - photon emission event is smaller than 1/16 .
another important issue to be addressed in mdi setting is what kind of measurement setup should be implemented experimentally at eve s laboratory . naively thinking , as sarg04 differs from bb84 only in the post - processing part , the same measurement setup for mdi bb84 should also work for mdi - sarg04 protocol . on the contrary , however , it turns out that the measurement setup for mdi bb84 results in high bit error rate when applied to mdi - sarg04 protocol , and consequently , no significant key can be generated . to generate a key in practice
, we propose two alternative measurement schemes for the mdi - sarg04 protocol , and simulate the resulting key generation rate . |
after many years of research in lattice qcd , it was possible to formulate qcd with chiral fermions on the lattice .
let @xmath0 be the size of the extra dimension , @xmath1 the wilson - dirac operator , and @xmath2 the bare fermion mass .
then , the theory with _ domain wall _
fermions is defined by the action : @xmath3 \\
p_{+}(\psi_{n+1 } + m \psi_1 ) = 0 , ~~p_{-}(\psi_0 + m \psi_n ) = 0 \end{array}\ ] ] where @xmath4 is the five - dimensional fermion matrix of the regularized theory and @xmath5 with @xmath6 being a mass parameter .
i define also a theory with _
truncated overlap fermions _ in complete analogy with the domain wall fermions by substituting @xcite : @xmath7 both theories can be compactified in the walls of the extra fifth dimension as low energy effective theories given by the action : @xmath8 where @xmath9 is the chiral dirac operator satisfying the ginsparg - wilson relation @xcite : @xmath10 where @xmath11 is the lattice spacing and @xmath12 is a local operator trivial in the dirac space .
untill now , the computations with chiral fermions with the standard algorithms have been very expensive .
the extra fermion flavors introduce a large overhead .
one multiplication with the fermion matrix costs @xmath15 @xmath1-multiplications with @xmath16 for domain wall fermions and much larger for the overlap operator @xcite . in this letter
i propose a multigrid algorithm along the fifth dimension which makes these simulations much faster .
the key observation is the lack of gauge connections along this dimension .
it is well - known that the overhead of such algorithms scales like @xmath0log@xmath0 .
here it is the @xmath17 algorithm : _
algorithm1 ( generic ) for solving the system @xmath18 : _
@xmath19 where by @xmath20 is denoted a vector with zero entries and @xmath21 are tolerances .
@xmath22 is typically orders of magnitude larger than @xmath23 such that the work per @xmath24 inversion is minimized . in fig .
1 we compare the norm of the residual @xmath26 of the conjugate residual ( cr ) algorithm ( which is optimal since @xmath24 is normal @xcite ) and @xmath25 .
i gain about an order of magnitude ( in average ) on @xmath27 @xmath28-configurations at @xmath29 and @xmath30 .
let @xmath32 be the following orthogonal transformation : @xmath33 i computed the inverse of @xmath9 in the @xmath25 using @xmath34 where @xmath35 is the same as @xmath4 but with @xmath36 @xcite and the subscript @xmath37 stands for the @xmath38 block of an @xmath0x@xmath0 partitioned matrix along the fifth dimension . | lattice regularization of chiral fermions is an important development of the theory of elementary particles .
nontheless , brute force computer simulations are very expensive , if not prohibitive . in this letter
i exploit the non - interacting character of the lattice theory in the flavor space and propose a multigrid approach for the simulation of the theory .
already a two - grid algorithm saves an order of magnitude of computer time for fermion propagator calculations .
-10 mm 11.15 ha , 11.30.rd , 11.30.fs key words : lattice qcd , chiral fermions , algorithms |
quantum critical systems are systems that exhibit a competition between the tendency for magnetic moments to order upon cooling and the tendency of those moments to be shielded by the conduction electrons through the kondo shielding mechanism . through the application of hydrostatic pressure or chemical pressure ( doping ) ,
the phase transition to a long - range ordered state can be suppressed to zero kelvin ; when this happens , such systems have been observed to display non - fermi liquid behavior over a range of temperatures@xcite , and some such systems display dynamical scaling .
when a quantum critical system displays dynamical scaling , then the response is no longer a function of the energy @xmath3 transferred to the system at temperature @xmath4 , but of the ratio @xmath5 only .
this scaling was first observed in ucu@xmath6pd@xcite , and has later been identified in other quantum critical systems@xcite , also as a function of applied magnetic field@xcite ( @xmath7-scaling ) . + in order for a system to display dynamical scaling , low - energy degrees of freedom must be available for excitations to couple to and relax back to equilibrium .
if this is the case , then the @xmath5-scaling is similar to the emergence of scaling behavior in high - energy physics where we also have the situation that @xmath8 .
the heavily doped quantum critical system ce(fe@xmath0ru@xmath1)@xmath2ge@xmath2 represents a very interesting case in the sense that this system displays @xmath5-scaling@xcite , but two plausible scenarios@xcite for the origin of low - energy excitations have been ruled out .
experiments on poly - crystalline ce(fe@xmath0ru@xmath1)@xmath2ge@xmath2 have already shown@xcite that there is no divergence of the local susceptibility in this system ( that would reflect the emergence of local moments at the expense of kondo shielded moments@xcite ) , and that the dynamic susceptibility has an energy dependence that can not be described by lorentzian lineshapes , excluding a scenario@xcite where the dynamical scaling is ascribed to fluctuations associated with a transition to a long - range spin - density wave ( sdw ) ordered state .
+ a solution to this observed @xmath5-scaling conundrum could be the presence of clusters .
ce(fe@xmath0ru@xmath1)@xmath2ge@xmath2 has been shown@xcite to form isolated , magnetically ordered clusters upon cooling down .
when the temperature is lowered , the magnetic moments on the ce ions get shielded by the conduction electrons through the kondo shielding mechanism . however , since the system has one ru ion for every three fe ions ( randomly distributed ) , the ensuing distribution in interionic distances has been shown to result in a distribution of kondo shielding temperatures . as a result ,
some moments get shielded before others , and the system displays percolative behavior@xcite upon cooling .
the clusters that peel of the lattice spanning , infinite cluster are subject to quantum mechanical finite - size effects@xcite ; as a result , the moments of the ions that constitute the cluster align with their neighbors , giving rise to fully ordered magnetic clusters with ising symmetry .
the superspin of such a cluster is free to pick a direction along the positive or negative z - direction ( c - axis ) . as such
, these clusters have one degree of freedom left , namely the direction of the superspin along the ising axis .
reorientations of the superspin could provide the sought - after low - energy excitations that this system must be capable of in order to explain the observed @xmath5-scaling behavior .
+ the formation of these isolated clusters has been observed directly by means of neutron scattering experiments@xcite , and the unusual behavior of the low temperature specific heat was shown@xcite to be consistent with the entropy shedding expected for clusters becoming isolated and forced to order magnetically .
monte carlo computer simulations for the relevant percolative system@xcite have shown that the emergence of magnetic clusters should play a dominant role in the low temperature behavior of heavily - doped quantum critical systems , while a comparison to a purely classical system@xcite that houses magnetic clusters has demonstrated that the observed dynamical scaling in quantum critical systems is likely to be the consequence of the presence of isolated clusters . in this paper , we demonstrate that superspin reorientations of isolated clusters do indeed occur in ce(fe@xmath0ru@xmath1)@xmath2ge@xmath2 , and that their flipping rates are qualitatively given by the predictions of hoyos and vojta@xcite for isolated ising clusters of varying sizes .
the predicted@xcite dynamic behavior of isolated clusters in an ising system with dissipation depends on the size of the clusters : the rate at which the superspin of a cluster can flip depends on the size of the cluster , and above a critical cluster size the spins are frozen in .
we argue that quantum critical ce(fe@xmath0ru@xmath1)@xmath2ge@xmath2 is such a system . + ] in reciprocal lattice units ( rlu ) for energy transfer @xmath3= 2.5 mev , corresponding to excitations with wave number @xmath9 , with @xmath10 the length of the c - axis .
both the flipper on ( magnetic , top two sets ) and flipper off ( nuclear , bottom two data sets ) are shown for @xmath4= 1.56 k and 50 k ( circles : 1.56 k , magnetic ; squares : 50 k , magnetic ; stars : 1.56 k nuclear ; diamonds : 50 k nuclear ) .
the magnetic data shows that there is broad scattering present at 50 k , and that this scattering increases upon cooling to 5 k over a broad @xmath11-range .
the feature at @xmath11= 0.65 is a spurion .
b ) the bottom panel displays the magnetic , flipper on data for @xmath3= 1.25 mev .
the increase in scattering between 50 k ( diamonds ) and 1.56 k ( circles ) occurs over a broad @xmath11-range centered around @xmath11= 0.45 .
the width of the additional scattering is about twice the width of the additional scattering observed at @xmath3= 0 mev ( errorbars with solid line through the points ) .
the elastic data have been divided by a factor of 5 for plotting clarity .
all lines in both panels are guides to the eye.,width=264 ] in order to demonstrate the simultaneous presence of frozen as well as dynamic clusters in ce(fe@xmath0ru@xmath1)@xmath2ge@xmath2 , we have performed new inelastic neutron scattering experiments using the bt7 triple axis spectrometer at nist in full polarization mode ; we combine these data with previously published data where spectra were collected on the same crystal using the hb3 spectrometer at hfir , and the dcs spectrometer at nist .
bt7 was operated @xcite using neutrons with fixed final energy of 14.7 mev , allowing for the removal of higher order neutrons by means of pg - filters in the scattered beam .
@xmath12he polarizers were utilized in the incident and scattered beam .
the time - dependence of these polarizers was well characterized during the course of the 5 day experiment on bt7 , allowing for a full separation between the spin - flip and non - flip signals using the technology and software developed at nist .
all polarized scattering experiments were set up with neutron polarization direction parallel to the direction of momentum transfer so that only the magnetic scattering would show up in the spin - flip channel .
we show characteristic data in fig .
the errorbars in all our figures represent plus or minus one standard deviation .
+ the observance of frozen clusters is straightforward , and has already been reported on : previous dcs experiments@xcite showed that the additional scattering that developed below 1 k at the ordering wave vector @xmath13= ( 0,0,0.45 ) was resolution limited , implying that the cluster reorientation frequency had to be longer than 80 ps , therby identifying these clusters as frozen and verifying one part of the prediction of reference [ ] . + in order to interpret the scattering associated with the reorientations of the superspins of the isolated clusters , we need to demonstrate that the observed quasi - elastic scattering is magnetic in origin ; this is shown in fig .
second , we need to show that the reorientation rate of smaller clusters is higher than that of larger clusters .
+ our task is facilitated by the fact that once clusters form upon cooling in ce(fe@xmath0ru@xmath1)@xmath2ge@xmath2 , these clusters persist upon further cooling@xcite .
that is , once a cluster has formed and been forced to order through finite size effects , then this cluster is no longer subject to kondo screening . as a result , by doing temperature dependent studies , we can first observe the dynamics of the smallest clusters at the higher temperatures . then , upon further cooling , we observe the dynamics of larger clusters as additional scattering on top of the scattering due to the smaller clusters .
we show in fig .
[ fig2 ] that this behavior can also be observed in quasi - elastic scattering by polycrystals .
+ ru@xmath1)@xmath2ge@xmath2 at @xmath14 = 3 nm@xmath15 , demonstrating the emergence of magnetic clusters . these data ( replotted from reference [ ] )
were taken on the in6 spectrometer at the ill .
the energy resolution of the spectrometer is 0.07 mev .
upon cooling from 100 k to 2 k magnetic correlations are seen to develop in the imaginary part of the dynamic susceptibility @xmath16`"`@xmath17 with the values at any given @xmath4 exceeding those at higher @xmath4 .
there is no sign of intensity migrating from the higher energies to lower ones , as emphasized by the tail of the data shown on an expanded vertical scale .
thus , the clusters present at the high @xmath4 persist down to the lowest @xmath4 , and that the additional scattering is due to newly minted clusters that peel off the infinite cluster upon cooling .
note that the data at 4.4 k and 1.8 k mostly differ in the region @xmath18 0.2 mev , implying that very close to the quantum critical point the infinite cluster is rapidly breaking apart , but still spawning fluctuating clusters .
all lines are guides to the eye . for plotting clarity ,
only one in every three errorbars has been plotted in the main figure , and only every second errorbar in the inset.,width=264 ] for emerging clusters that persist down to the lowest @xmath4 while still fluctuating , we expect to observe the following . on cooling , we should see the scattered intensity increase in the vicinity of @xmath13 , with the intensity at lower @xmath4 exceeding that at higher@xmath4 for all energy and momentum transfers in the vicinity of @xmath13 .
thus , any new intensity should appear on top of the intensity already present for the case where smaller clusters survive further cooling and keep their reorientation capabilities .
in addition , the new intensity should correspond to larger clusters , which corresponds to a sharper signal in momentum space .
this intensity should also have a smaller characteristic energy width , reflecting the lower flipping rate of the larger clusters .
we have observed all this . + the data in fig .
[ fig2 ] shows that , on cooling , the additional intensity appears on top of the existing intensity , with a decreasing characteristic energy - width .
we have verified this on bt7 for @xmath19 1 mev ( where resolution effects do not hamper the interpretation ) for our single crystal .
since these data do not add to fig .
[ fig2 ] , we show them elsewhere . in addition , it had already been shown that the elastic scattering around @xmath13 followed a similar pattern@xcite : on cooling , additional scattering was observed@xcite to appear on top of the intensity already present , and this additional scattering was narrower in momentum transfer , reflecting the formation of larger clusters without affecting the clusters already present .
+ the data in fig .
[ fig1]b is a good demonstration of the above .
when measuring at constant energy transfer ( @xmath3= 0 and 1.25 mev ) , we see that the data at @xmath3= 1.25 mev are wider than the elastic data .
this is exactly as expected for a system that sees smaller , fast - fluctuating clusters emerge at higher @xmath4 , but predominantly slower , larger ones at low @xmath4 .
when this happens , the intensity for @xmath20 ( with @xmath21 the flipping rate of a large superspin ) stops evolving on further cooling and its quasi - eleastic width reflects the width at the temperature below which predominantly larger clusters emerge .
e.g. , the larger clusters that emerge at the lowest @xmath4 fluctuate so slowly that they do not add to the intensity at @xmath3= 1.25 mev . as a result , while the width in momentum transfer narrows upon cooling for the elastic channel , the width at @xmath3= 1.25 mev remains unchanged , reflecting the ( persisting ) presence of the smaller clusters . for reference
, the width at 1.25 mev corresponds to the elastic width at @xmath22 4 k@xcite .
note that normal critical scattering does not exhibit this ( fig .
[ fig1]b ) behavior .
+ thus , smaller clusters exhibit a higher reorientation rate than larger clusters , as predicted theoretically@xcite .
however , the superspin of larger clusters can still flip , as witnessed in the temperature dependence of the scattering at lower energy transfers ( and clearly visible ( for @xmath23 0.1 mev ) in the high - resolution polycrystalline data shown in fig .
[ fig2 ] ) . combining this with the emergence of resolution - limited magnetic intensity that emerges at the very lowest @xmath4 on the high - resolution spectrometer dcs@xcite
, we conclude that ce(fe@xmath0ru@xmath1)@xmath2ge@xmath2 houses both frozen and dynamic clusters , and that the reorientation rate of the dynamic clusters depends on the size of the magnetic clusters .
+ in summary , we have shown that isolated , ordered clusters emerge , upon cooling , in a strongly doped quantum critical system .
these clusters can be frozen as well as dynamic , and the degree of freedom associated with the reorientation of the superspin of the dynamic clusters persists down to the lowest temperatures , providing the system with a readily accessible branch of low - energy excitations .
this branch is likely the reason behind the observed but unexplained@xcite @xmath5-scaling .
+ 10 g.r .
stewart , rev .
73 * , 797 ( 2001 ) ; * 78 * , 743 ( 2006 ) .
et al . _ , phys .
lett . * 75 * , 725 ( 1995 ) .
a. schrder _
et al . _ , nature ( london ) * 407 * , 351 ( 2000 ) .
w. montfrooij _
et al . _ ,
lett . * 91 * , 087202 ( 2003 ) .
qimiao si _
_ , nature ( london ) * 413 * , 804 ( 2001 ) .
john a. hertz , phys .
b * 14 * , 1165 ( 1976 ) .
a.j . millis , phys . rev .
b * 48 * , 7183 ( 1993 ) .
dietrich stauffer and amnon aharony , _ introduction to percolation theory _
( crc , boca raton , fl , 1994 ) .
t. heitmann _
et al . _ ,
j. of physics conference series * 391 * , 012018 ( 2012 ) .
w. montfrooij _
et al . _ ,
b * 76 * , 052404 ( 2007 ) .
et al . _ , j. appl
115 * , 17e110 ( 2014 ) .
t. heitmann _
et al . _ , j. of mod .
phys . * 5 * , 649 ( 2014 ) .
t. heitmann _
et al . _ ,
b * 81 * , 014411 ( 2010 ) .
jos a. hoyos and thomas vojta , phys .
b * 74 * , 140401 ( r ) ( 2006 ) . w. c. chenit et al .
, physica b 397 , 168 ( 2007 ) . | we have identified low - energy magnetic excitations in a doped quantum critical system by means of polarized neutron scattering experiments .
the presence of these excitations could explain why ce(fe@xmath0ru@xmath1)@xmath2ge@xmath2 displays dynamical scaling in the absence of local critical behavior or long - range spin - density wave criticality .
the low - energy excitations are associated with the reorientations of the superspins of fully ordered , isolated magnetic clusters that form spontaneously upon lowering the temperature .
the system houses both frozen clusters and dynamic clusters , as predicted by hoyos and vojta [ phys . rev .
b 74 , 140401 ( r ) ( 2006 ) ] . |
the stellar populations of galaxies carry a record of their star forming and chemical histories , from the epoch of formation to the present .
the global properties of galaxies are determined by the nature and evolution of their stellar components .
their studies thus provide a powerful tool to explore the physics of galaxy formation and evolution ( @xcite ; @xcite ) . for local group galaxies and some very nearby galaxies that can be resolved into individual stars with hst and 10-m class telescopes ,
the stellar population properties may be studied by means of direct observations . however , for objects at larger distances , individual stars ( except for some giants ) are unresolved , even with 10-m class telescopes .
the integrated light of such galaxies is expected to contain valuable information about their physical properties .
the most common approach used to interpret the integrated spectrophotometric properties of galaxies is _ stellar population synthesis_. there are two main types of population synthesis studies : evolutionary population synthesis and empirical population synthesis ( eps hereafter ) . both types of studies require a complete library of input ( observed or model atmosphere ) spectra of stars or star clusters ( e.g. , @xcite ) . in the evolutionary population synthesis approach , pioneered by tinsley ( 1967 ) ,
the main adjustable parameters are the stellar initial mass function ( imf ) , the star formation history and , in some cases , the rate of chemical enrichment .
assumptions about the time evolution of these parameters allow one to compute the age - dependent distribution of stars in the hertzsprung - russell diagram , from which the integrated spectral evolution of the stellar population can be obtained . in recent years
, a number of groups have developed evolutionary population synthesis models , which allow one to investigate the physical properties of observed galaxies ( @xcite ; arimoto & yoshii 1987 ; buzzoni 1989 ; bressan , chiosi & fagotto 1994 ; worthey 1994 ; weiss , peletier & matteucci 1995 ; fioc & rocca - volmerange 1997 ; mas - hesse & kunth 1999 ; moeller , fritze - v .
alvensleben & fricke 1999 ; leitherer et al .
1999 ; and bruzual & charlot 1993 , 2003 ) .
such models are convenient tools for studying the spectral _ evolution _ of galaxies , as they allow one to predict the past and future spectral appearance of galaxies observed at any given time .
however , modern evolutionary population synthesis models still suffer from serious limitations ( e.g. , charlot 1996 ) .
the empirical population synthesis approach , also known as ` stellar population synthesis with a data base ' , was introduced by faber ( 1972 ) . in this technique ,
one reproduces the integrated spectrum of a galaxy with a linear combination of spectra of individual stars or star clusters of various types taken from a comprehensive library .
the empirical population synthesis approach has been employed successfully by several authors to interpret observed galaxy spectra ( faber 1972 ; oconnell 1976 ; pickles 1985 ; bica 1988 ; pelat 1998 ; boisson et al .
2000 ; cid fernandes 2001 ) .
an appealing property of this approach is that the results do not hinge on a priori assumptions about stellar evolution , the histories of star formation and chemical enrichment , nor in the case of a library of individual stellar spectra the imf .
in return , it does not allow one to predict the past and future spectral appearance of galaxies . in 1996 , a workshop devoted to the comparison of various evolutionary population synthesis and empirical population synthesis codes showed broad agreement in general and a number of discrepancies in detail ( leitherer et al .
1996 ) . in this paper
, we use the empirical population synthesis approach to constrain the stellar content and star formation histories of a sample of 73 blue compact galaxies ( bcgs ) .
our approach builds on earlier studies by bica ( 1988 ) and cid fernandes ( 2001 ) , and we adopt the library of observed star - cluster spectra assembled by bica & alloin ( 1986 ; see also bica 1988 ) .
our primary goal is to illustrate the constraints that can obtained on the ages and metallicities of the stellar populations of bcgs that can not be resolved into individual stars , on the basis of their integrated spectra . in a forthcoming study
, we will use the results of this analysis to model the spectral evolution of bcgs using evolutionary population synthesis models .
another goal of the present work is to refine previous measurements of the balmer emission - line fluxes of the bcgs in our sample , through an accurate modeling of the underlying stellar absorption spectrum .
this issue is critical to spectroscopic analyses of bcgs ( e.g. , @xcite ; @xcite ) .
the paper is organized as follows . in section 2
, we briefly review the principle of the empirical population synthesis approach .
we describe the properties of the bcg spectral sample in section 3 . in section 4
, we estimate the stellar components in bcgs by fitting the observed equivalent widths and continuum colors with the empirical population synthesis approach . in section 5
we discuss the age - metallicity degeneracy and the possible application of our results .
section 6 summarizes our main results .
in a recent paper , cid fernandes ( 2001 ) revisit the classical problem of synthesizing the spectral properties of a galaxy by using a ` base ' of star - cluster spectra , approaching it from a probabilistic perspective .
their work improves over previous eps studies at both the formal and computational levels , and it represents an efficient tool for probing the stellar population mixture of galaxies . to provide a quantitative description of the stellar components in the nuclear regions of bcgs
, we will apply this eps method to our bcg sample .
it is based on spectral - group templates built from star clusters of different ages and metallicities , and on bayes probabilistic theorem and the eps metropolis algorithm . in this section ,
we first describe the star cluster base .
we then briefly review bayes probabilistic formulation and the eps metropolis algorithm
. empirical population synthesis by with a data base ( a base of @xmath0 spectra of stars or star clusters ) is a classical tool designed to study the stellar population of galaxies .
the recent eps algorithm of cid fernandes ( 2001 ; see also bica 1988 ) relies on a data base of star - cluster spectra rather than on one of individual stellar spectra .
the star cluster base used by cid fernandes ( 2001 ) was introduced by schmidt ( 1991 ) .
it consists of 12 population groups , spanning five age bins 10 gyr ( representing globular cluster - like populations ) , 1 gyr , 100 myr , 10 myr and ` hii'-type ( corresponding to current star formation ) and four metallicities 0.01 , 0.1 , 1 and 4 times z@xmath1 ( see table 1 of cid fernandes 2001 ) .
all metallicities are not available for clusters of all ages .
in particular , spectra of star clusters younger than 1 gyr are only available for the metallicities z@xmath1 and 4z@xmath1 .
since blue compact galaxies are expected to have predominantly low abundances , the base of star - cluster spectra used by cid fernandes ( 2001 ) is not optimal for our analysis .
the base of 35 star - cluster spectra used by bica ( 1988 ; see also bica & alloin 1986 ) is better appropriate for our purpose , as it contains spectra of young star clusters with metallicities down to @xmath2 ( see table 1 ) .
each component , corresponding to a specific age and metallicity , is characterized by a set of six metallic features ( k @xmath33933 , cn @xmath34200 , g band @xmath34301 , @xmath35175 , @xmath38543 , @xmath38662 ) and three balmer lines ( , , and ) , as well as @xmath4 continuum fluxes at selected pivot wavelengths ( 3290 , 3660 , 4020 , 4510 , 6630 , 7520 and 8700 ) , which are normalized at @xmath35870 .
the library contains an hii region component corresponding to current star formation .
this is represented by a pure continuum based on the spectrum of 30 dor , which is used at all metallicities ( see the recent description by schmitt 1996 ) .
table 1 lists the ages and metallicities of each of the 35 components of the bica ( 1988 ) base of star - cluster spectra .
the top line lists the ages of the components , while the rightmost column lists the metallicities , with [ z / z@xmath1 ] = log(z / z@xmath1 ) .
.ages ( in units of yr ) , metallicities and numbering convention for the star clusters in the base . [ cols="^,^,^,^,^,^,^,^,>",options="header " , ] the results of our analysis are summarized in tables 3 and 4 for the galaxies with asymptotic ( young - star ) metallicities [ z / z@xmath1]=0.5 and 0.0 , respectively . for each galaxy , we report the percentage contributions from each age - metallicity component to the integrated flux at @xmath6 .
the first and second lines of each table indicate the metallicities and the ages of the different stellar components . for reference
, the third line indicates the indices of the stellar components in the cluster data base of table 1 . for each galaxy
, we also list the inferred @xmath7-band attenuation @xmath8 and the absorption equivalent widths of , , and , as measured from the synthetic stellar population spectra .
figure [ eqw - syn - obs ] shows the equivalent widths ( @xmath9 ) of all the stellar absorption features used to constrain the fits in the model spectra against those in the observed spectra ( @xmath10 , for @xmath11 ) , for all 73 bcgs in our sample .
clearly , the feature strengths in the model spectra are in very good agreement with those in the observed spectra .
a first noticeable result in tables 3 and 4 is that all bcgs show an underlying component of stars older than @xmath12 .
the fractional contribution of this component to the total light at @xmath35870 exceeds 15% for most galaxies , except for some low - luminosity bcgs , such as i zw 18 , ii zw 40 , which have very strong emission line spectra , and have a marginally detected old component .
we note that the spectra used here sample the inner regions of the galaxies .
the contribution from old and intermediate - age stars to the integrated light could be even larger in the extended ( off - nuclear ) regions of galaxies .
the presence of significant populations of old and intermediate - age stars indicates that blue compact galaxies have experienced substantial episodes of star formation in the past .
this supports the results from deep imaging observations of the color - magnitude diagrams of a few nearby bcgs that these are old galaxies ( see section 5.2 below ) .
as expected , stars younger than @xmath13 tend to dominate the emission at 5870 , consistent with the observational character of bcgs , i.e. blue colors and strong emission lines .
another interesting result of tables 3 and 4 is that blue compact galaxies present a variety of star formation histories . in ii zw 67 and i
zw 56 , for example , stars with age 10 gyr contribute as much as 20 per cent of the flux at @xmath35870 .
in contrast , in iii zw 43 and mrk 57 , old stars do not contribute significantly to the flux at @xmath35870 , that is produced in half by intermediate - age stars .
the galaxies i zw 18 and ii zw 40 differ from these cases in that their emission at @xmath35870 is accounted almost entirely by young stars .
the star formation history of bcgs , therefore , appears to vary significantly on a case by case basis . for simplicity , in what follows
we arrange stars into four age bins : * old * stars with age @xmath14 ( proportion ) ; * int*ermediate - age stars with ages between @xmath12 and @xmath15 ( proportion ) ; * y*oung * s*tars with ages between @xmath16 and @xmath17 ( proportion ) ; and newly - born stars in regions ( proportion ) .
hence + + + = 1 .
figure [ fig - age ] illustrates graphically the variety of star formation histories reported in tables 3 and 4 for the 73 galaxies in our sample . in each panel
, the horizontal axis represents , , , and ( from left to right ) , while the vertical axis shows the percentage contributions at 5870 of stars in these four age bins .
stars younger than @xmath17 dominate the emission in most bcgs , but the galaxies also contain substantial fractions of older stars .
figure [ fig - spec ] illustrates the results of the spectral fits obtained for four galaxies in our sample .
also shown are the contributions to the integrated spectrum by stars in the four age groups defined in section 4.3 above ( the relative contributions by the different stellar components to the total flux at @xmath35870 are those listed in tables 3 and 4 for these galaxies ) .
figure [ fig - spec ] shows that the synthetic spectra ( * syn * ) inferred from our population synthesis analysis provide good fits to the observed spectra of bcgs ( * obs * ) .
the absorption wings of , and in the observed spectra are also well reproduced by the models ( the synthetic spectra do not include nebular emission lines ) .
we find that , for some strong star - forming galaxies , the synthetic spectra do not provide very good fits to the observed continuum spectra at wavelengths between 4300 and 4800 .
this may arise from the presence of wolf - rayet ( wr ) features , such as features at @xmath34511 4535 , @xmath34565 , @xmath34605 , 4620 , and @xmath34686 , in the observed spectra .
we plan to investigate this small discrepancy using evolutionary population synthesis models in a future paper .
the range of star formation histories inferred for the bgcs in our sample may be interpreted in terms of an evolutionary sequence . following cid fernandes ( 2001b ) ,
we represent graphically the histories of star formation of the galaxies in figure [ fig - ed ] , in a plane with abscissa = + and ordinate .
also shown as dashed lines in the figure are lines of constant ( + + = 1 ) .
the main result from figure [ fig - ed ] is that the 73 bcgs in our sample define a sequence in and over a relatively small range of .
the range in is slightly smaller for galaxies with asymptotic ( young - star ) metallicity [ z / z@xmath1]=0.5 ( stars ) than for those with asymptotic metallicity [ z / z@xmath1]= 0.0 ( squares ) .
some galaxies with ` extreme ' stellar populations are labelled in fig .
[ fig - ed ] . in i
zw 56 and ii zw 67 , for example , old stars contribute up to @xmath18% of the integrated flux at @xmath35870 ( see section 4.3 above ) .
the galaxy iii zw 43 is that where intermediate - age stars contribute the most to the optical continuum emission .
in contrast , some galaxies , such as i zw 18 , ii zw 40 , appear to be almost ` pure starbursts ' .
the small percentage of old and intermediate - age stars found here for these galaxies is consistent with the results from other recent studies , e.g. , smoker ( 1999 ) , aloisi ( 1999 ) ( see section 5.2 below ) .
since the location of an individual galaxy in figure [ fig - ed ] reflects the evolutionary state of its stellar population , we can interpret the _ distribution _ of galaxies in this diagram as an _
evolutionary _ sequence .
in particular , starbursting galaxies whose spectra are entirely dominated by young massive stars ( bottom right part of the diagram ) will presumably evolve toward larger /values ( top left part of the diagram ) over a timescale of a few @xmath19 , once their bursts have ceased .
this interpretation is supported by several observational facts .
first , metal absorption - line features become stronger and galaxy colors become progressively redder as one moves from large to large along the sequence .
second , galaxies with large ( such as i zw 56 , ii zw 67 ) have spectra typical of a `` post - starburst '' galaxies , with pronounced balmer absorption lines characteristic of a - type stars and with no strong emission lines .
finally , most bcgs in which wr features have been detected are located in the large- region of the diagram , consistent with the young burst ages implied by the presence of wr stars ( @xcite ) .
accurate measurements of the h - balmer emission lines are crucial to constrain the attenuation by dust , the star formation rate , the gas - phase abundances of chemical elements and the excitation parameter in galaxies ( e.g. , @xcite ) . to measure with accuracy the balmer emission - line fluxes of bcgs , we must account for the contamination by underlying stellar absorption .
different approaches have been used to correct balmer emission - line fluxes for underlying stellar absorption in bcgs .
the simplest approach consists in adopting a constant equivalent width ( 1.52 ) for all the balmer absorption lines ( e.g. , @xcite ; @xcite ) .
another standard correction consists in determining the absorption equivalent width through an iterative procedure , by assuming that the equivalent width is the same for all balmer lines and by requiring that the color excesses derived from / , / , and / be consistent ( e.g. , @xcite ; @xcite ) . in reality
, however , the absorption equivalent width may not be the same for all h - balmer lines .
the advantage of the empirical population synthesis method used above to fit the observed spectra of bcgs is that it provides simultaneous fits to the continuum and stellar absorption features of the galaxies .
we have measured the absorption equivalent widths of , , , and in the synthetic spectra fitted to all 73 galaxies in our sample ( last four columns of table 3 and 4 ) .
figure [ fig - abs ] shows the distributions of the equivalent widths of the four lines .
the equivalent widths of , , and range typically between 2 and 5 ( the distributions have different shapes for different lines ) , while that of is typically less than 2 .
hence , figure [ fig - abs ] indicates that adopting a constant absorption equivalent width for all h - balmer lines is only a crude approximation .
our results are consistent with those of mas - hesse & kunth ( 1999 ) and olofsson ( 1995 )
. we can correct with accuracy , therefore , the fluxes of balmer emission lines for underlying stellar absorption in the spectra of bcgs . in a forthcoming study
, we will exploit the nebular emission - line spectra ( obs syn in figure [ fig - spec ] ) of bcgs to constrain the rates of star formation and gas - phase chemical element abundances in these galaxies .
it is worth noting that the reason why we are able to constrain simultaneously the ages and metallicities of the different stellar components in bcgs is that our approach is based on fitting stellar absorption lines , in addition to continuum fluxes .
age and metallicity have similar effects on the continuum spectra of galaxies .
however , some stellar absorption features such as , , and the g - band have been shown to depend mostly on age , while others , such as and fe@xmath35709 , have been shown to depend mostly on metallicity ( e.g. , @xcite ; @xcite ; @xcite ) . in this paper
, we use the k @xmath33933 , @xmath34102 , cn @xmath34200 , g band @xmath34301 , @xmath34340 and @xmath35176 absorption features to determine the stellar population of bcgs .
these include both age - sensitive and metal - sensitive features
. some of these absorption features may not be available in some spectra , which may lack , for example , and .
also , some absorption features may be very shallow , and hence difficult to measure with high accuracy .
for most galaxies , however , we find that the results of the empirical population synthesis analysis provide coarse but useful constraints on the histories of star formation and metal enrichment . the most straightforward way to observe the star formation history of galaxies is through deep single - star photometry .
this allows one to directly identify stars in various evolutionary phases through their positions in a color - magnitude diagram ( cmd ) containing the fossil record of the star formation history . during the past few years
, intense activity has focused on the reconstruction of the star formation histories of nearby galaxies using this approach .
several of these studies were performed on bcgs and led to new constraints on the ages of the oldest stars present in these galaxies ( @xcite ; @xcite ; @xcite ; @xcite ) .
the general conclusions from these studies are that bcgs contain evolved stellar populations , and that their star formation histories have been discontinuous .
the basic shortcoming of this approach is that it can not be used in galaxies at large distances .
in addition , the cmd - analysis method is also subject to several uncertainties such as distance determination , extinction , and contamination of stellar colors by gaseous emission .
our results of the eps analysis of 73 bcgs are consistent with those of the cmd analysis of a few nearby galaxies , supporting the finding that bcgs are old galaxies with intermittent star formation .
the evolutionary diagram constructed in figure [ fig - ed ] above shows that the eps method is not only able to recognize composite systems from a handful of observable absorption - line features and continuum fluxes , but also to provide a rough description of the evolutionary state of the stellar components .
therefore , the eps method provides a convenient tool for the study of stellar components and star formation histories of galaxies .
it can be easily applied to the spectra gathered by large spectroscopic galaxy surveys .
we have presented the results of an empirical population synthesis study of a sample of 73 blue compact galaxies . our main goal was to study the stellar components of bcgs .
we have constrained the star formation histories of bcgs by comparing observed stellar absorption features and continuum fluxes with a library of star cluster spectra .
our conclusions can be summarized as follows : bcgs present a variety of star formation histories , as inferred from the wide spread in stellar absorption equivalent widths and continuum colors from galaxy to galaxy .
bcgs are typically age - composite stellar systems , in which different stellar components are clearly distinguished : the current starburst , an underlying older population , and some intermediate - age population .
a quantitative analysis indicates that the nuclei of some bcgs are dominated by young components and the star - forming process is still ongoing . in most of bcgs ,
stars older than 1 gyr contribute significantly to the integrated optical emission .
the contribution by these stars can exceed 40% in some cases .
overall , the stellar populations of bcgs suggest that they are old galaxies undergoing intermittent star formation episodes ; a typical bcgs is not presently forming its first generation of stars .
we also find that the attenuation by dust is typically very small in the bcgs in our sample .
our results are consistent with the results from deep imaging observations using hst and large ground - based telescopes .
the virtue of the eps approach is that it is applied to integrated galaxy spectra .
this method should be useful , therefore , for interpreting the spectra garthered by large spectroscopic galaxy surveys .
the eps approach also provides accurate spectral fits of observed galaxy spectra . from these spectral fits , it is possible to measure with accuracy , in particular , the absorption strengths of stellar balmer lines and to correct the observed balmer emission - line fluxes for underlying stellar absorption .
the pure emission - line spectra of the bcgs in our sample , resulting from the subtraction of the synthetic spectral fits from the observed spectra , will be presented in a forthcoming paper .
we thank an anonymous referee for helpful comments and constructive suggestions , which helped us improve the paper .
we are very grateful to dr .
r. cid fernandes for giving us the eps computer program , and kindly assisting us in its use .
this work is based on observations made with the 2.16 m telescope of the beijing astronomical observatory(bao ) and supported by the chinese national natural science foundation ( cnnsf 10073009 ) .
thanks the alexander von humboldt foundation , the federal ministry of education and research , and the programme for investment in the future ( zip ) of the german government for support .
x.k . has been financed by the special funds for major state basic research projects of china and the alexander von humboldt foundation of germany . | this is the third paper of a series dedicated to the study of the star formation rates , star formation histories , metallicities and dust contents of a sample of blue compact galaxies ( bcgs ) .
we constrain the stellar contents of 73 blue compact galaxies by analyzing their continuum spectra and the equivalent widths of strong stellar absorption features using a technique of empirical population synthesis based on a library of observed star - cluster spectra .
our results indicate that blue compact galaxies are typically age - composite stellar systems ; in addition to young stars , intermediate - age and old stars contribute significantly to the 5870 continuum emission of most galaxies in our sample .
the stellar populations of blue compact galaxies also span a variety of metallicities .
the ongoing episodes of star formation started typically less than a billion years ago .
some galaxies may be undergoing their first global episode of star formation , while for most galaxies in our sample , older stars are found to contribute up to half the optical emission .
our results suggest that bcgs are primarily old galaxies with discontinuous star formation histories .
these results are consistent with the results from deep imaging observations of the color - magnitude diagrams of a few nearby bcgs using hst and large ground - based telescopes .
the good quality of our population synthesis fits of bcg spectra allow us to estimate the contamination of the , , and balmer emission lines by stellar absorption .
the absorption equivalent widths measured in the synthetic spectra range from typically 1.5 for , to 25 for , , and . the implied accurate measurements of emission - line intensities will be used in a later study to constrain the star formation rates and gas - phase chemical element abundances of blue compact galaxies . |
recent solar @xcite , atmospheric @xcite , reactor @xcite and accelerator @xcite neutrino experiments have convincingly verified the hypothesis of neutrino oscillations , which can naturally happen if neutrinos are slightly massive and lepton flavors are mixed .
this discovery indicates the existence of new physics beyond the standard model ( sm ) .
although it is easy to add either dirac or majorana neutrino mass terms to the sm , it is highly non - trivial to reveal the essential meaning behind such terms and find a natural and qualitative explanation of the smallness of neutrino masses .
a complete theory of neutrino mass generation has been lacking and is eagerly desirable . a low - energy effective theory responsible for
the generation of neutrino masses might give rise to slight violation of the unitarity of the neutrino mixing matrix .
if three light neutrinos are mixed with other degrees of freedom ( e.g. , with the light sterile neutrinos @xcite , the heavy majorana neutrinos @xcite , or the whole tower of kaluza - klein states in some models with extra dimensions @xcite ) , their @xmath1 flavor mixing matrix @xmath2 appearing in the sm charged - current interactions will in general be non - unitary .
therefore , the deviation of @xmath2 from unitarity can serve as an indicator of new physics beyond the sm
. a generic non - unitary neutrino mixing matrix @xmath2 can be parametrized as @xmath3 , flavor mixing matrix .
for example , the non - unitary matrix @xmath2 can also be expressed as @xmath4 , where @xmath5 is a lower ( or upper ) triangle matrix and @xmath6 is a unitary matrix @xcite .
although different parametrizations are mathematically equivalent , they may have their own advantages in describing different phenomena of neutrino physics . ] where @xmath7 is a hermitian matrix which can be written as @xmath8 with @xmath9 being real and @xmath10 ( @xmath11 ) being complex , and @xmath12 is a unitary matrix which is usually parametrized in terms of three mixing angles and one dirac cp - violating phase @xcite as @xmath13 here we have omitted two majorana cp - violating phases of @xmath14 since they are irrelevant to neutrino oscillations .
given the lepton mixing matrix @xmath2 , the constraints on the moduli of the elements of @xmath15 have been deduced in ref . @xcite by combining the experimental data on both neutrino oscillations and precision electroweak tests : @xmath16 it is then easy to find that @xmath17 , @xmath18 , @xmath19 , and @xmath20 should hold .
we can further denote @xmath9 as @xmath21 , @xmath22 and @xmath23 ; namely , @xmath24 where @xmath25 , @xmath26 , @xmath27 are all real . bounds on @xmath25 , @xmath26 and @xmath27 can also be obtained from eq . ( 3 ) : @xmath28 , @xmath29 and @xmath30 .
the effects of non - unitarity of @xmath2 on neutrino oscillations have been discussed in some literature @xcite .
in particular , the authors of ref .
@xcite have used the same parametrization of @xmath2 as given above and explored the effects of cp violation induced by those non - unitary complex parameters in neutrino oscillations . in this paper , we start from an intriguing point of view that the realistic neutrino mixing matrix @xmath2 might result from a non - unitary correction to the well - known tri - bimaximal mixing pattern @xcite .
the latter is compatible with current experimental data very well and can be derived from a number of flavor symmetries and their spontaneous or explicit breaking mechanisms @xcite .
instead of building a specific neutrino model to realize such a phenomenological conjecture , here we shall concentrate on the consequences of @xmath2 on neutrino oscillations . in our new neutrino mixing scenario , @xmath31 with @xmath14 being given by @xmath32 and @xmath7 being an arbitrary hermitian matrix as shown in eq .
( 1 ) or ( 4 ) . to be specific , the non - unitary neutrino mixing
matrix @xmath2 reads @xmath33 it is clear that the parameters of @xmath7 lead simultaneously to the unitarity violation and the deviation from @xmath14 .
the resulting smallest mixing angle @xmath0 to be measured in reactor @xmath34 oscillation experiments is attributed to the small parameters @xmath35 and @xmath36 .
so are the cp - violating phases of @xmath2 .
the remaining parts of this paper are organized as follows . in sections
ii and iii , we develop a complete set of series expansion formulas for neutrino oscillation probabilities both in vacuum and in matter of constant density , respectively , by taking account of the non - unitary mixing matrix @xmath2 given in eq .
( 6 ) . in section
iv , we discuss the possibility of determining some parameters of @xmath7 by constructing the deformed unitarity triangles " .
section v is devoted to a short discussion about incorporating our parametrization of the non - unitary neutrino mixing matrix into a generic type - ii seesaw model .
finally some conclusions are drawn in section vi .
suppose that the non - unitary v in eq .
( 6 ) describes the mixing between the neutrino fields in the mass basis and those in the flavor basis , @xmath37 where @xmath38 and @xmath39 .
the probability of neutrino oscillation @xmath40 ( @xmath41 ) can be derived in a similar way to that in the unitary case .
the procedures of deriving the formulas for neutrino oscillation probabilities can be found in ref .
@xcite . here
we follow another way , in order to be concise .
the derivation may be easily understood as follows .
a typical neutrino oscillation process @xmath40 can be divided into three parts @xcite : 1 ) @xmath42 being produced at the source through the charged - current interaction which can be denoted as @xmath43 .
here , @xmath44 is a superposition of mass eigenstates @xmath45 ; 2 ) @xmath45 propagates from the source to the detector ; 3 ) @xmath46 ( a superposition of @xmath45 ) being catched by the detector through the charged - current interaction @xmath47 .
therefore , the amplitude of the neutrino oscillation @xmath48 can be correspondingly divided into three parts : @xmath49 in the case of non - unitary neutrino mixing , it follows that @xmath50 .
the factor @xmath51 ensures that the total rate @xmath52 .
similarly , we have @xmath53 .
the expression of @xmath54 is the same as that in the unitary case : @xmath55 .
finally , the amplitude of the neutrino oscillation @xmath48 is given by @xmath56 then @xmath41 , the probability of neutrino oscillation @xmath48 , is given by @xmath57 \ ; \nonumber\\ & \ ; = \ ; & \frac{1}{\left ( v_{}^ { } v_{}^{\dagger } \right ) _ { \alpha \alpha } \left ( v_{}^ { } v_{}^{\dagger } \right ) _ { \beta \beta } } \cdot \left [ |\left ( v_{}^ { } v_{}^{\dagger } \right ) _ { \alpha \beta}|^2 - 4 a_{\alpha \beta}^{21 } \sin^2 \delta_{21}^ { } - 4 a_{\alpha \beta}^{31 } \sin^2 \delta_{31}^ { } - 4 a_{\alpha \beta}^{32 } \sin^2 \delta_{32}^ { } \right . \nonumber\\ & & \hspace{4 cm } \left . - 2 j_{\alpha \beta}^{21 } \sin2\delta_{21}^ { } - 2 j_{\alpha \beta}^{31 } \sin2\delta_{31}^ { } - 2 j_{\alpha \beta}^{32 } \sin2\delta_{32}^ { } \right ] \ ; , % ( 10)\end{aligned}\ ] ] where @xmath58 with @xmath59 , and @xmath60 $ ] , @xmath61 $ ] .
we can further absorb the renormalization factor @xmath62 into the redefinitions of @xmath63 and @xmath64 and rewrite the above equation as @xmath65 where @xmath66 , @xmath67 . the first term in eq .
( 11 ) is the so - called zero - distance " term .
it means that at @xmath68 we have @xmath69 note that if @xmath70 , @xmath71 ; namely , there are no zero - distance " effects in the disappearance experiments . if @xmath72 , the oscillation probability @xmath73 is in general nonzero , that is the zero - distance " effect .
one can find that although it is nonzero , this term add only a tiny constant to the oscillation probability , and does not change the oscillatory behavior . in our scenario , we have @xmath74 , @xmath75 and @xmath76 . another significant difference between the non - unitary and unitary cases is : if the mixing matrix @xmath2 is non - unitary , there may exist 9 different jarlskog invariants @xmath77 corresponding to 3 different oscillation channels instead of a unique jarlskog in the unitary case .
if @xmath78 is satisfied , we can expand eq .
( 11 ) as @xmath79 in our calculations , we find that although all the nine @xmath80 are of @xmath81 , only @xmath82 is of @xmath83 while @xmath84 and @xmath85 are both of @xmath86 .
this observation means that the most sensitive way at short baseline neutrino oscillation experiments to detect cp violation is to measure the @xmath87 channel .
such a point was also pointed out in refs .
@xcite . here
we present a complete set of formulas for neutrino oscillation probabilities @xmath41 to the second order in powers of @xmath35 , @xmath36 , @xmath88 , @xmath25 , @xmath26 , @xmath27 and @xmath89 .
these formulas are good approximations for the @xmath90-dominated oscillations , i.e. , for neutrino oscillation experiments with relative short baselines and relatively high energies .
@xmath91 \right ) ^2 \right ] \sin^2 \delta_{31}^ { } + \frac{2}{3 } \left ( 1 + 2 { \rm re}[\hat{\kappa}_{12}^ { } ] \right ) \delta_{21}^ { } \sin2\delta_{31}^ { } \nonumber\\ & & - \frac{4}{9 } \left ( 2 - 3 \sin^2 \delta_{31}^ { } \right ) \delta_{21}^{2 } \ ; , \\ % ( 15 ) p_{\tau \tau}^ { } & \ ; \approx \ ; & 1 - \left ( 1 - 4 \left ( { \rm re}[\hat{\kappa}_{23}^ { } ] \right ) ^2 \right ) \sin^2 \delta_{31}^ { } + \frac{2}{3 } \left ( 1 - 2 { \rm re}[\hat{\kappa}_{13}^ { } ] \right ) \delta_{21}^ { } \sin2\delta_{31}^ { } \nonumber\\ & & - \frac{4}{9 } \left ( 2 - 3 \sin^2 \delta_{31}^ { } \right ) \delta_{21}^{2 } \ ; , \\ % ( 16 ) p_{e \mu}^ { } & \ ; \approx \ ; & 4 |\hat{\kappa}_{12}^{}|^2 - \left ( 3 |\hat{\kappa}_{12}^{}|^2 - |\hat{\kappa}_{13}^{}|^2 + 2 { \rm re}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] \right ) \sin^2 \delta_{31}^ { } + \frac{4}{9 } \delta_{21}^{2 } \nonumber\\ & & + \frac{2}{3 } { \rm re}[\hat{\kappa}_{12}^ { } + \hat{\kappa}_{13}^ { } ] \delta_{21}^ { } \sin2\delta_{31}^ { } + 2 { \rm im}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] \sin2\delta_{31}^ { } \nonumber\\ & & + \frac{4}{3 } \left ( 2 { \rm im}[\hat{\kappa}_{12}^ { } ] - { \rm im}[\hat{\kappa}_{12}^ { } + \hat{\kappa}_{13}^ { } ] \sin^2 \delta_{31}^ { } \right ) \delta_{21}^ { } \ ; , \\ % ( 17 ) p_{e \tau}^ { } & \ ; \approx \ ; & 4 |\hat{\kappa}_{13}^{}|^2 - \left ( 3 |\hat{\kappa}_{13}^{}|^2 - |\hat{\kappa}_{12}^{}|^2 + 2 { \rm re}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] \right ) \sin^2 \delta_{31}^ { } + \frac{4}{9 } \delta_{21}^{2 } \nonumber\\ & & - \frac{2}{3 } { \rm re}[\hat{\kappa}_{12}^ { } + \hat{\kappa}_{13}^ { } ] \delta_{21}^ { } \sin2\delta_{31}^ { } - 2 { \rm im}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] \sin2\delta_{31}^ { } \nonumber\\ & & - \frac{4}{3 } \left ( 2 { \rm im}[\hat{\kappa}_{13}^ { } ] - { \rm im}[\hat{\kappa}_{12}^ { } + \hat{\kappa}_{13}^ { } ] \sin^2 \delta_{31}^ { } \right ) \delta_{21}^ { } \ ; , \\ % ( 18 ) p_{\mu \tau}^ { } & \ ; \approx \ ; & 4 |\hat{\kappa}_{23}^{}|^2 + \left [ 1 - |\hat{\kappa}_{12}^ { } + \hat{\kappa}_{13}^{}|^2 - 4 \left ( \right ) ^2 \right ) \right ] \sin^2 \delta_{31}^ { } \nonumber\\ & & - \frac{2}{3 } \left ( 1 + { \rm re}[\hat{\kappa}_{12}^ { } - \hat{\kappa}_{13}^ { } ] \right ) \delta_{21}^ { } \sin2\delta_{31}^ { } + \frac{4}{9 } \left ( 1 - 3 \sin^2 \delta_{31}^ { } \right ) \delta_{21}^{2 } \nonumber\\ & & + \left [ \left ( 2 - \epsilon_{b}^ { } - \epsilon_{c}^ { } \right ) { \rm im}[\hat{\kappa}_{23}^ { } ] - { \rm im}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] \right ] \sin2\delta_{31}^ { } \nonumber\\ & & - \frac{4}{3 } \left ( 2 { \rm im}[\hat{\kappa}_{23}^ { } ] - { \rm im}[\hat{\kappa}_{12}^ { } + \hat{\kappa}_{13}^ { } + \hat{\kappa}_{23}^ { } ] \sin^2 \delta_{31}^ { } \right ) \delta_{21}^ { } \ ; .
% ( 19)\end{aligned}\ ] ] the first term in each of the above six equations is the zero - distance " term .
the last two terms in eq .
( 17 ) , ( 18 ) or ( 19 ) are the cp - violating " terms .
suppose that the absolute values of those non - unitary parameters are around their upper bounds , i.e. , @xmath92 , @xmath93 , @xmath94 , @xmath95 , @xmath96 and @xmath97 .
we notice that @xmath98 is only sensitive to @xmath99 through the detection of @xmath100 oscillation . by measuring the probability of @xmath101 oscillation it is possible to determine or constrain the value of @xmath102 $ ] .
if the small difference between @xmath103 and @xmath104 can be well measured , then we are able to obtain the information on @xmath105 $ ] . combined with the value of @xmath106 ,
@xmath107 can be determined . as for @xmath108 $ ]
, the most effective way is to probe the cp - violating terms in the @xmath109 channel .
when a neutrino beam passes through matter , only @xmath110 can interact with electrons in the medium via the charged - current interactions , while @xmath110 , @xmath111 and @xmath112 can all interact with electrons , protons and neutrons in the medium via the neutral - current interactions .
the coherent forward scattering from the constituents of matter modifies the evolution behaviors of the neutrino beam . in the vacuum mass eigenbasis ,
the evolution equation of neutrinos can be written as @xmath113 we use tildes to denote the quantities in matter .
for the propagation of neutrinos in matter of constant density , the hamiltonian @xmath114 is given by @xcite @xmath115 where @xmath116 is the energy matrix in the mass eigenbasis in vacuum , @xmath117 , with @xmath118 and @xmath119 ( @xmath120 and @xmath121 are the electron and neutron densities , respectively ) . here
@xmath2 is just the non - unitarity mixing matrix in eq .
the hermitian matrix @xmath114 can be diagonalized by a unitary transformation @xmath122 , where @xmath123 is the effective energy matrix in matter .
the solution to eq .
( 20 ) can be expressed as @xmath124 where we have inserted @xmath125 . from eq .
( 22 ) we can work out the neutrino oscillation probabilities in matter : @xmath126 \ ; , % ( 23)\end{aligned}\ ] ] where @xmath127 , @xmath128 $ ] , @xmath129 $ ] and @xmath130 .
comparing eq .
( 23 ) with eq . ( 11 ) , we find that the matrix @xmath131 , which is also non - unitary , can be regarded as the effective neutrino mixing matrix in matter . in appendix
a , we present the details of the approximate diagonalization of the hamiltonian @xmath114 by using the perturbation theory . in the results to be presented below , those terms in proportion to @xmath132 will be neglected .
the reason is simple : for ordinary earth matter , which is electrically neutral , we have @xmath133 to a good degree of accuracy , and thus we can safely set @xmath134 .
it is necessary to mention that the subsequent analytical approximations are not very good for a relative large @xmath135 or for the @xmath136-dominated oscillation .
in addition , we can not directly obtain eqs . ( 14 ) @xmath137 ( 19 ) from eqs . ( 27 ) @xmath137 ( 32 ) by setting @xmath138 . this is because the expansion of @xmath139 in eqs .
( a4 ) @xmath137 ( a7 ) is improper if @xmath140 .
the eigenvalues of @xmath141 , @xmath142 @xmath143 , are related to the effective neutrino masses in matter by the relations @xmath144 , where @xmath145 is the energy of neutrinos . and @xmath146 to denote the effective neutrino masses and the mass squared differences in matter instead of @xmath147 and @xmath148 , because we did not ordering @xmath142 according to their magnitude and the mass spectrum . after ordering @xmath142 , we have @xmath149 and @xmath150 . ]
then the effective mass squared differences in matter are given by @xmath151 which are shown in eqs .
( 24 ) , ( 25 ) and ( 26 ) to the second order in @xmath35 , @xmath36 , @xmath88 , @xmath25 , @xmath26 , @xmath27 and @xmath89 .
@xmath152 \right ) - \frac{2 \left ( \delta m_{21}^{2 }
\right ) ^2}{9 e_{\nu}^ { } v_{cc}^ { } } \nonumber\\ & & - \frac{\left ( 2 e_{\nu}^ { } v_{nc}^ { } \right ) ^2}{\delta m_{31}^{2 } } \left [ \left ( \epsilon_{b}^ { } - \epsilon_{c}^ { } \right ) ^2 + 4 \left ( { \rm i m } [ \hat{\kappa}_{23}^ { } ] \right ) ^2 \right ] \nonumber\\ & & + e_{\nu}^ { } \left ( v_{cc}^ { } - v_{nc}^ { } \right ) \left ( \left | \hat{\kappa}_{12}^ { } - \hat{\kappa}_{13}^ { } \right |^2 - 2 \epsilon_{a}^{2 } \right ) \nonumber\\ & & - e_{\nu}^ { } v_{nc}^ { } \left ( \left | \epsilon_{b}^ { } - \hat{\kappa}_{23}^ { } \right |^2 + \left | \epsilon_{c}^ { } - \hat{\kappa}_{23}^ { } \right |^2 - 2 \left | \hat{\kappa}_{12}^ { } \right |^2 - 2 \left | \hat{\kappa}_{13}^ { } \right |^2 \right ) \ ; , \\
% ( 24 ) \delta \tilde{\lambda}_{31}^{2 } & \ ; \approx \ ; & \delta m_{31}^{2 } - 2 e_{\nu}^ { } v_{cc}^ { } - \frac{1}{3 } \delta m_{21}^{2 } - 4 e_{\nu}^ { } \left ( v_{cc}^ { } - v_{nc}^ { } \right ) \epsilon_{a}^ { } \nonumber\\ & & - 2 e_{\nu}^ { } v_{nc}^ { } \left ( \epsilon_{b}^ { } + \epsilon_{c}^ { } - 2 { \rm re } [ \hat{\kappa}_{23}^ { } ] \right ) - \frac{\left ( \delta m_{21}^{2 } \right ) ^2}{9 e_{\nu}^ { } v_{cc}^ { } } \nonumber\\ & & + \frac{\left ( 2 e_{\nu}^ { } v_{nc}^ { } \right ) ^2}{\delta m_{31}^{2 } } \left [ \left ( \epsilon_{b}^ { } - \epsilon_{c}^ { } \right ) ^2 + 4 \left ( { \rm i m } [ \hat{\kappa}_{23}^ { } ] \right ) ^2 \right ] \nonumber\\ & & + e_{\nu}^ { } \left ( v_{cc}^ { } - v_{nc}^ { } \right ) \left ( \left | \hat{\kappa}_{12}^ { } + \hat{\kappa}_{13}^ { } \right |^2 - 2 \epsilon_{a}^{2 } \right ) \nonumber\\ & & - e_{\nu}^ { } v_{nc}^ { } \left ( \left | \epsilon_{b}^ { } + \hat{\kappa}_{23}^ { } \right |^2 + \left | \epsilon_{c}^ { } + \hat{\kappa}_{23}^ { } \right |^2 - 2 \left | \hat{\kappa}_{12}^ { } \right |^2 - 2 \left | \hat{\kappa}_{13}^ { } \right |^2 \right ) \ ; , \\
% ( 25 ) \delta \tilde{\lambda}_{32}^{2 } & \ ; \approx \ ; & \delta m_{31}^{2 } - \frac{2}{3 } \delta m_{21}^{2 } + 8 e_{\nu}^ { } v_{nc}^ { } { \rm re } [ \hat{\kappa}_{23}^ { } ] + \frac{\left ( \delta m_{21}^{2 } \right ) ^2}{9 e_{\nu}^ { } v_{cc}^ { } } \nonumber\\ & & + \frac{2 \left ( 2 e_{\nu}^ { } v_{nc}^ { } \right ) ^2}{\delta m_{31}^{2 } } \left [ \left ( \epsilon_{b}^ { } - \epsilon_{c}^ { } \right ) ^2 + 4 \left ( { \rm i m } [ \hat{\kappa}_{23}^ { } ] \right ) ^2 \right ] \nonumber\\ & & + 4 e_{\nu}^ { } \left ( v_{cc}^ { } - v_{nc}^ { } \right ) { \rm re } [ \hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] - 4 e_{\nu}^ { } v_{nc}^ { } \left ( \epsilon_{b}^ { } + \epsilon_{c}^ { } \right ) { \rm re } [ \hat{\kappa}_{23}^ { } ] \ ;
. % ( 26)\end{aligned}\ ] ] in eqs .
( 27 ) to ( 32 ) , we present the expansion forms of eq .
( 22 ) for all six neutrino oscillation probabilities to the second order in @xmath35 , @xmath36 , @xmath88 , @xmath25 , @xmath26 , @xmath27 and @xmath89 in terms of the quantities in vacuum .
@xmath153 \right ) ^2 + \frac{2 \sqrt{2 } \delta_{21}^ { } v_{nc}^ { } l}{3 \delta_{31}^ { } \left ( v_{cc}^ { } - 2 \delta_{31}^ { } \right ) } \left ( \epsilon_{b}^ { } -\epsilon_{c}^ { } \right ) \right . \nonumber\\ & & \left . - 4 \left ( { \rm re}[\hat{\kappa}_{23}^ { } ] \right ) ^2 + \frac{4 v_{cc}^ { } l}{\delta_{31}^ { } } \left ( \epsilon_{b}^ { } -\epsilon_{c}^ { } \right ) { \rm re}[\hat{\kappa}_{23}^ { } ] - \frac{\left ( v_{cc}^ { } l \right ) ^2}{\delta_{31}^ { } } \left ( \epsilon_{b}^ { } -\epsilon_{c}^ { } \right ) ^2 \right ] \sin^2 \delta_{31}^ { } \nonumber\\ & & + \left [ 2 \left ( \frac{1}{3 } \delta_{21}^ { } + 2 v_{nc}^ { } l { \rm re}[\hat{\kappa}_{23}^ { } ] \right ) - \frac{4 \delta_{21}^{2}}{9 v_{cc}^ { } l } - 2 \left ( v_{cc}^ { } - v_{nc}^ { } \right ) l { \rm re}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] \right . \nonumber\\ & & \left .
+ 2 v_{nc}^ { } l \left ( \epsilon_{b}^ { } + \epsilon_{c}^ { } \right ) { \rm re}[\hat{\kappa}_{23}^ { } ] - \frac{\left ( v_{nc}^ { } l \right ) ^2}{\delta_{31}^ { } } \left [ \left ( \epsilon_{b}^ { } - \epsilon_{c}^ { } \right ) ^2 + 4 \left ( { \rm i m } [ \hat{\kappa}_{23}^ { } ] \right ) ^2 \right ] \right ] \sin2\delta_{31}^ { } \nonumber\\ & & + 2 \left | \frac{2 \delta_{21}^{}}{3 v_{cc}^ { } l } + \hat{\kappa}_{12}^ { } \right |^2 \cdot \left [ 2 \sin^2 \delta_{31}^ { } \sin^2 \left ( \frac{v_{cc}^ { } l}{2 } \right ) + \frac{1}{2 } \sin2 \delta_{31}^ { } \sin \left ( v_{cc}^ { } l \right ) \right ] \nonumber\\ & & - 4 \left | \frac{2 \delta_{21}^{}}{3 v_{cc}^ { } l } + \hat{\kappa}_{12}^ { } \right |^2 \sin^2 \left ( \frac{v_{cc}^ { } l}{2 } \right ) - 4 \left ( \frac{1}{3 } \delta_{21}^ { } + 2 v_{nc}^ { } l { \rm re}[\hat{\kappa}_{23}^ { } ] \right ) ^2 \ ; , \\ % ( 28 ) \nonumber\\ \tilde{p}_{\tau \tau}^ { } & \ ; = \ ; & 1 - \left [ 1 - 8 \left ( \frac{1}{3 } \delta_{21}^ { } + 2 v_{nc}^ { } l { \rm re}[\hat{\kappa}_{23}^ { } ] \right ) ^2 - \frac{2 \sqrt{2 } \delta_{21}^ { } v_{nc}^ { } l}{3 \delta_{31}^ { } \left ( v_{cc}^ { } - 2 \delta_{31}^ { } \right ) } \left ( \epsilon_{b}^ { } -\epsilon_{c}^ { } \right ) \right . \nonumber\\ & & \left . - 4 \left ( { \rm re}[\hat{\kappa}_{23}^ { } ] \right ) ^2 - \frac{4 v_{cc}^ { } l}{\delta_{31}^ { } } \left ( \epsilon_{b}^ { } -\epsilon_{c}^ { } \right ) { \rm re}[\hat{\kappa}_{23}^ { } ] - \frac{\left ( v_{cc}^ { } l \right ) ^2}{\delta_{31}^ { } } \left ( \epsilon_{b}^ { } -\epsilon_{c}^ { } \right ) ^2 \right ] \sin^2 \delta_{31}^ { } \nonumber\\ & & + \left [ 2 \left ( \frac{1}{3 } \delta_{21}^ { } + 2 v_{nc}^ { } l { \rm re}[\hat{\kappa}_{23}^ { } ] \right ) - \frac{4 \delta_{21}^{2}}{9 v_{cc}^ { } l } - 2 \left ( v_{cc}^ { } - v_{nc}^ { } \right ) l { \rm re}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] \right . \nonumber\\ & & \left . + 2 v_{nc}^ { } l \left ( \epsilon_{b}^ { } + \epsilon_{c}^ { } \right ) { \rm re}[\hat{\kappa}_{23}^ { } ] - \frac{\left ( v_{cc}^ { } l \right ) ^2}{\delta_{31}^ { } } \left [ \left ( \epsilon_{b}^ { } - \epsilon_{c}^ { } \right ) ^2 + 4 \left ( { \rm i m } [ \hat{\kappa}_{23}^ { } ] \right ) ^2 \right ] \right ] \sin2\delta_{31}^ { } \nonumber\\ & & + 2 \left | \frac{2 \delta_{21}^{}}{3 v_{cc}^ { } l } - \hat{\kappa}_{13}^ { } \right |^2 \cdot \left [ 2 \sin^2 \delta_{31}^ { } \sin^2 \left ( \frac{v_{cc}^ { } l}{2 } \right ) + \frac{1}{2 } \sin2 \delta_{31}^ { } \sin \left ( v_{cc}^ { } l \right ) \right ] \nonumber\\ & & - 4 \left | \frac{2 \delta_{21}^{}}{3 v_{cc}^ { } l } - \hat{\kappa}_{13}^ { } \right |^2 \sin^2 \left ( \frac{v_{cc}^ { } l}{2 } \right ) - 4 \left ( \frac{1}{3 } \delta_{21}^ { } + 2 v_{nc}^ { } l { \rm re}[\hat{\kappa}_{23}^ { } ] \right ) ^2 \ ; , \\ % ( 29 ) \nonumber\\ \tilde{p}_{e \mu}^ { } & \ ; = \ ; & 4 |\hat{\kappa}_{12}^{}|^2 - \left ( 3 |\hat{\kappa}_{12}^{}|^2 - |\hat{\kappa}_{13}^{}|^2 + 2 { \rm re}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] \right ) \sin^2 \delta_{31}^ { } \nonumber\\ & & + 4 \left ( \frac{2 \delta_{21}^{}}{3 v_{cc}^ { } l } + { \rm re}[\hat{\kappa}_{12}^ { } ] \right ) { \rm re}[\hat{\kappa}_{12}^ { } + \hat{\kappa}_{13}^ { } ] \sin^2 \delta_{31}^ { } \sin^2 \left ( \frac{v_{cc}^ { } l}{2 } \right ) \nonumber\\ & & + \left ( \frac{2 \delta_{21}^{}}{3 v_{cc}^ { } l } + { \rm re}[\hat{\kappa}_{12}^ { } ] \right ) { \rm re}[\hat{\kappa}_{12}^ { } + \hat{\kappa}_{13}^ { } ] \sin2 \delta_{31}^ { } \sin \left ( v_{cc}^ { } l \right ) \nonumber\\ & & + 4 \left [ \left ( \frac{2 \delta_{21}^{}}{3 v_{cc}^ { } l } \right ) ^2 - \left | \hat{\kappa}_{12}^ { } \right |^2 \right ] \sin^2 \left ( \frac{v_{cc}^ { } l}{2 } \right ) \nonumber\\ & & \underline{+ 2 \left ( \frac{2 \delta_{21}^{}}{3 v_{cc}^ { } l } { \rm im}[\hat{\kappa}_{12}^ { } + \hat{\kappa}_{13}^ { } ] - { \rm im}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] \right ) \sin 2 \delta_{31}^ { } \sin^2 \left ( \frac{v_{cc}^ { } l}{2 } \right ) } \nonumber\\ & & \underline{+ 2 \left ( \frac{2 \delta_{21}^{}}{3 v_{cc}^ { } l } { \rm im}[\hat{\kappa}_{12}^ { } + \hat{\kappa}_{13}^ { } ] - { \rm im}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] \right ) + \sin^2 \delta_{31}^ { } \sin \left ( v_{cc}^ { } l \right ) } \nonumber\\ & & \underline{+ 2 { \rm im}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] \sin 2 \delta_{31}^ { } + \frac{8 \delta_{21}^{}}{3 v_{cc}^ { } l } { \rm im}[\hat{\kappa}_{12}^ { } ] \sin \left ( v_{cc}^ { } l \right ) } \ ; , \\ % ( 30 ) \nonumber\\ \tilde{p}_{e \tau}^ { } & \ ; = \ ; & 4 |\hat{\kappa}_{13}^{}|^2 - \left ( 3 |\hat{\kappa}_{13}^{}|^2 - |\hat{\kappa}_{12}^{}|^2 + 2 { \rm re}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] \right ) \sin^2 \delta_{31}^ { } \nonumber\\ & & - 4 \left ( \frac{2 \delta_{21}^{}}{3 v_{cc}^ { } l } - { \rm re}[\hat{\kappa}_{13}^ { } ] \right ) { \rm re}[\hat{\kappa}_{12}^ { } + \hat{\kappa}_{13}^ { } ] \sin^2 \delta_{31}^ { } \sin^2 \left ( \frac{v_{cc}^ { } l}{2 } \right ) \nonumber\\ & & - \left ( \frac{2 \delta_{21}^{}}{3 v_{cc}^ { } l } - { \rm re}[\hat{\kappa}_{13}^ { } ] \right ) { \rm re}[\hat{\kappa}_{12}^ { } + \hat{\kappa}_{13}^ { } ] \sin2 \delta_{31}^ { } \sin \left ( v_{cc}^ { } l \right ) \nonumber\\ & & + 4 \left [ \left ( \frac{2 \delta_{21}^{}}{3 v_{cc}^ { } l } \right ) ^2 - \left | \hat{\kappa}_{13}^ { } \right |^2 \right ] \sin^2 \left ( \frac{v_{cc}^ { } l}{2 } \right ) \nonumber\\ & & \underline{- 2 \left ( \frac{2 \delta_{21}^{}}{3 v_{cc}^ { } l } { \rm im}[\hat{\kappa}_{12}^ { } + \hat{\kappa}_{13}^ { } ] - { \rm im}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] \right ) \sin 2 \delta_{31}^ { } \sin^2 \left ( \frac{v_{cc}^ { } l}{2 } \right ) } \nonumber\\ & & \underline{- 2 \left ( \frac{2 \delta_{21}^{}}{3 v_{cc}^ { } l } { \rm im}[\hat{\kappa}_{12}^ { } + \hat{\kappa}_{13}^ { } ] - { \rm im}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] \right ) \sin^2 \delta_{31}^ { } \sin \left ( v_{cc}^ { }
l \right ) } \nonumber\\ & & \underline{- 2 { \rm im}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] \sin 2 \delta_{31}^ { } - \frac{8 \delta_{21}^{}}{3 v_{cc}^ { } l } { \rm im}[\hat{\kappa}_{13}^ { } ] \sin \left ( v_{cc}^ { } l \right ) } \ ; , \\ % ( 31 ) \nonumber\\ \tilde{p}_{\mu \tau}^ { } & \ ; = \ ; & 4 |\hat{\kappa}_{23}^{}|^2 + \left [ 1 + 4 \left ( \frac{1}{3 } \delta_{21}^ { } + 2 v_{nc}^ { } l { \rm re}[\hat{\kappa}_{23}^ { } ] \right ) ^2 - |\hat{\kappa}_{12}^ { } + \hat{\kappa}_{13}^{}|^2 \right . \nonumber\\ & & \left . - 4 |\hat{\kappa}_{23}^{}|^2 - 4\left ( { \rm im}[\hat{\kappa}_{23}^ { } ] \right ) ^2 - \frac{\left ( v_{cc}^ { } l \right ) ^2}{\delta_{31}^ { } } \left ( \epsilon_{b}^ { } -\epsilon_{c}^ { } \right ) ^2 \right ] \sin^2 \delta_{31}^ { } \nonumber\\ & & - \left [ 2 \left ( \frac{1}{3 } \delta_{21}^ { } + 2 v_{nc}^ { } l { \rm re}[\hat{\kappa}_{23}^ { } ] \right ) + \frac{8 \delta_{21}^{2}}{9 v_{cc}^ { } l } - 2 v_{nc}^ { } l \left ( \epsilon_{b}^ { } + \epsilon_{c}^ { } \right ) { \rm re}[\hat{\kappa}_{23}^ { } ] \right . \nonumber\\ & & \left .
+ 2 \left ( v_{cc}^ { } - v_{nc}^ { } \right ) l { \rm re}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] + \frac{\left ( v_{cc}^ { } l \right ) ^2}{\delta_{31}^ { } } \left [ \left ( \epsilon_{b}^ { } - \epsilon_{c}^ { } \right ) ^2 + 4 \left ( { \rm i m } [ \hat{\kappa}_{23}^ { } ] \right ) ^2 \right ] \right ] \sin2\delta_{31}^ { } \nonumber\\ & & - 2 \left ( \frac{4 \delta_{21}^{2}}{9 \left ( v_{cc}^ { } l \right ) ^2 } + \frac{2 \delta_{21}^{}}{3 v_{cc}^ { } l } { \rm re}[\hat{\kappa}_{12}^ { } - \hat{\kappa}_{13}^ { } ] - { \rm re}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] \right ) \nonumber\\ & & \cdot \left [ 2 \sin^2 \delta_{31}^ { } \sin^2 \left ( \frac{v_{cc}^ { } l}{2 } \right ) + \frac{1}{2 } \sin2 \delta_{31}^ { } \sin \left ( v_{cc}^ { } l \right ) \right ] \nonumber\\ & & + 4 \left ( \frac{1}{3 } \delta_{21}^ { } + 2 v_{cc}^ { } l { \rm re}[\hat{\kappa}_{23}^ { } ] \right ) ^2 \underline{+ \left [ \left ( 2 - \epsilon_{b}^ { } - \epsilon_{c}^ { } \begin{matrix } \cr \end{matrix } - \frac{4 \left ( v_{nc}^ { } l \right ) ^2}{\delta_{31}^ { } } { \rm re}[\hat{\kappa}_{23}^ { } ] \right . \right . } \nonumber\\ & & \underline{\left .
\left . - \frac{2 \sqrt{2 } \delta_{21}^ { } v_{nc}^ { } l}{3 \delta_{31}^ { } \left ( v_{cc}^ { } - 2 \delta_{31}^ { } \right ) } \right ) { \rm im}[\hat{\kappa}_{23}^ { } ] - { \rm im}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] \right ] \sin 2 \delta_{31}^{}}\nonumber\\ & & \underline{+ 2 \left ( \frac{2 \delta_{21}^{}}{3 v_{cc}^ { } l } { \rm im}[\hat{\kappa}_{12}^ { } + \hat{\kappa}_{13}^ { } ] - { \rm im}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] \right ) \sin 2 \delta_{31}^ { } \sin^2 \left ( \frac{v_{cc}^ { } l}{2 } \right ) } \nonumber\\ & & \underline{+ 2 \left ( \frac{2 \delta_{21}^{}}{3 v_{cc}^ { } l } { \rm im}[\hat{\kappa}_{12}^ { } + \hat{\kappa}_{13}^ { } ] - { \rm im}[\hat{\kappa}_{12}^ { } \hat{\kappa}_{13}^ { * } ] \right ) \sin^2 \delta_{31}^ { } \sin \left ( v_{cc}^ { } l \right ) } \nonumber\\ & & \underline { + 16 { \rm im}[\hat{\kappa}_{23}^ { } ] \left ( \frac{1}{3 } \delta_{21}^ { } + 2 v_{nc}^ { } l { \rm re}[\hat{\kappa}_{23}^ { } ] \right ) \sin^2 \delta_{31}^ { } } \nonumber\\ & & \underline { - 8 { \rm im}[\hat{\kappa}_{23}^ { } ] \left ( \frac{1}{3 } \delta_{21}^ { } + 2 v_{nc}^ { } l { \rm re}[\hat{\kappa}_{23}^ { } ] \right ) } \ ; .
% ( 32)\end{aligned}\ ] ] in order to obtain the probabilities of anti - neutrino oscillations @xmath154 , one needs to simultaneously change the signs of @xmath155 , @xmath156 and the terms underlined in the expressions of @xmath157 .
different from the case in vacuum , the terms of @xmath81 which can be strongly enhanced by large @xmath158 appear not only in the expression of @xmath159 but also in those of @xmath160 and @xmath161 .
if we omit the terms of @xmath86 , then we get very concise formulas for @xmath160 , @xmath162 and @xmath159 : @xmath163 \right ) \sin2\delta_{31}^ { } \ ; , \\
% ( 33 ) \tilde{p}_{\tau \tau}^ { } & \ ; = \ ; & 1 - \sin^2 \delta_{31}^ { } + 2 \left ( \frac{1}{3 } \delta_{21}^ { } + 2 v_{nc}^ { } l { \rm re}[\hat{\kappa}_{23}^ { } ] \right ) \sin2\delta_{31}^ { } \ ; , \\
% ( 34 ) \tilde{p}_{\mu \tau}^ { } & \ ; = \ ; & \sin^2 \delta_{31}^ { } + 2 \left ( \frac{1}{3 } \delta_{21}^ { } + 2 v_{nc}^ { } l { \rm re}[\hat{\kappa}_{23}^ { } ] \right ) \sin2\delta_{31}^ { } + 2 { \rm im}[\hat{\kappa}_{23}^ { } ] \sin2\delta_{31}^ { } \ ; .
% ( 35)\end{aligned}\ ] ] we carry out a numerical analysis to show the difference between the corrections from the non - unitary parameter @xmath88 and the corrections from the nonzero @xmath0 to the neutrino oscillation probabilities .
we compare between two special cases : case i , we consider a unitary and nearly tri - bimaximal mixing matrix with nonzero @xmath0 , as @xmath12 given in eq .
( 2 ) with @xmath164 and @xmath165 ; case ii , we consider the non - unitary mixing matrix as shown in eq .
the inputs of our numerical calculations are @xmath166 , @xmath167 ( sign @xmath168 " for the normal hierarchy and @xmath169 " for the inverted hierarchy ) , the matter density @xmath170 . and
in case i we choose @xmath171 , in case ii we choose @xmath172 , @xmath173 , @xmath174 , @xmath175 , @xmath176 and @xmath177 .
our numerical analysis is independent of our analytical results given above , but it confirms the results of our analytical approximations . as for the analytical approximations for case i , one may refer to @xcite , in which there exists a complete set of series expansion for three - flavor neutrino oscillation probabilities in matter in terms of small @xmath0 and @xmath178 .
1 shows the effective mass differences in matter as functions of the neutrino beam energy @xmath145 in case i and case ii for both the normal and the inverted hierarchies .
we can clearly see that the mass squared differences are strongly magnified if @xmath145 is large ( or equivalently if the matter density is large ) .
an interesting point is that in case ii the effective mass difference @xmath179 can reach zero at around @xmath180 in the normal hierarchy case while in case i the nonzero @xmath181 ensures the nonzero value of @xmath179 .
if we choose a nonzero @xmath0 for case ii , we will get similar curves as those of case i. we find that although the non - unitary parameter @xmath182 plays a very similar role as @xmath183 in the expressions of neutrino oscillation probabilities in vacuum , it has very different effects from @xmath184 in matter .
this finding provides us with a possibility of distinguishing the nonzero @xmath0 from the non - unitary parameters in @xmath2 . taking account of eqs .
( 27 ) to ( 29 ) in ref .
@xcite , we can easily see that the dirac phase @xmath185 dose not appear in the expressions of the eigenvalues of @xmath141 .
however , in the non - unitary case , all the effective mass squared differences contain the terms proportional to @xmath186 $ ] which is relevant to @xmath187 for fixed @xmath188 .
2 shows the effective mass differences in matter as functions of the dirac phase @xmath185 in case i or the phase of @xmath88 in case ii for both mass hierarchies , where we have chosen @xmath189 .
we find that the correction from the term @xmath186 $ ] can be around @xmath190 in this situation .
3 tells us how the probabilities of neutrino oscillations @xmath101 and @xmath191 in matter are modified with the changing of the baseline @xmath158 , where we have chosen @xmath192 and @xmath193 .
we can clearly see from the figure that the probability @xmath159 ( @xmath194 ) can be largely enhanced ( depressed ) by a long baseline if the matter effect is taken into account . at the baseline @xmath195
, @xmath159 is about @xmath196 .
4 shows the probability @xmath159 and @xmath160 as functions of the dirac phase @xmath185 in case i or the phase of @xmath88 in case ii for both mass hierarchies with the baselines @xmath197 and @xmath198 .
we find that if @xmath159 can be measured to the level of @xmath199 at @xmath200 from the source and @xmath188 can be well measured , @xmath201 may convincingly be determined .
another point worthwhile to point out is that from eqs .
( 33 ) to ( 35 ) we find that these three probabilities have approximate sign[@xmath202 - @xmath201 degeneracy , which means @xmath203 . this degeneracy is broken by the term @xmath204 , which increases with the increase of the baseline @xmath158 .
4 shows that this degeneracy breaking can reach @xmath196 at @xmath198 if the energy of neutrinos is taken to be @xmath189 .
we admit that it is very difficult to measure the transition probabilities to the accuracy of @xmath196 or even @xmath199 in the present or forthcoming neutrino oscillation experiments .
given the small effects of unitarity violation , however , our numerical results can at least serve to illustrate how sensitive an ambitious long - baseline neutrino experiment should be to this kind of new physics .
it is worth remarking two positive aspects of searching for the non - unitarity of @xmath2 in the @xmath205 disappearance or @xmath206 appearance experiments .
first , the signatures of the non - unitarity can be strongly enhanced by the matter effects , and thus a high energy and a very long baseline are essential to detect appreciable effects of the non - unitarity in the neutrino oscillation experiments . as for neutrinos of energy around @xmath207 , a baseline much longer than @xmath200 may have much better sensitivity to the non - unitary parameters in the neutrino mixing matrix , in which case the varying of the terrestrial matter density need to be taken into account @xcite .
second , the dependency of the non - unitary parameters @xmath10 on the energy of the neutrino beam is different from other neutrino mixing parameters , and thus measuring the energy dependency of the transition probabilities will help to identify the small effects of the non - unitarity .
in the unitary limit , the neutrino mixing matrix @xmath2 which relates the neutrino mass eigenstates @xmath208 to the neutrino flavor eigenstates @xmath209 is unitary .
the unitarity implies @xmath210 in the complex plane , eq .
( 36 ) corresponds to six unitarity triangles @xcite denoted as @xmath211 , @xmath212 , @xmath213 , @xmath214 , @xmath215 and @xmath216 respectively .
the area of each unitarity triangle is @xmath217 , where @xmath218 is the jarlskog invariant measure of cp violation for the unitary mns mixing matrix .
if there is no cp violation ( e.g. , the tri - bimaximal mixing ) , the unitarity triangles shrink to segments . in other words , introducing the unitarity triangles " provides a geometric way to describe cp violation ( which can be determined by the area of each triangle ) by measuring the cp - conserving quantities ( the sides of the triangles ) . if the mixing matrix @xmath2 is non - unitary , the orthogonal relations in eq .
( 36 ) are in general not satisfied and the unitarity triangles in the complex plane are in general open .
note that every two vector sides can determine one triangle in the complex plane , of which twice the area corresponds to one of the nine jarlskog invariants . to be explicit
, the jarlskog invariant @xmath77 equals to twice the area of the triangle determined by @xmath219 and @xmath220 ( or @xmath221 and @xmath222 ) . apparently , if all the six triangles are closed , these nine jarlskog invariants are all equal ( to @xmath218 ) . as for our special scenario ,
the cp - violating effects are attributed to the phases of @xmath35 , @xmath36 and @xmath88 .
thus we are able to determine those unitarity - violating parameters by constructing those deformed unitarity triangles " . in appendix b , we give all the eighteen sides of the six deformed triangles .
table i shows the ratios of two sides of @xmath211 , @xmath212 and @xmath213 to the first order of @xmath36 and @xmath88 .
here we omit the smallest parameter @xmath35 .
c|l + @xmath211 & @xmath223 + & @xmath224 + + @xmath212 & @xmath225 + & @xmath226 + + @xmath213 & @xmath227 + & @xmath228 + & @xmath229 + one can find from appendix b that for @xmath211 we have @xmath230 , therefore @xmath211 is almost closed . from table
i we find that in our special scenario , if the ratio @xmath231 of @xmath211 can be well measured , @xmath106 can be determined .
combined with the ratio @xmath232 , the phase of @xmath36 can also be obtained .
the results can be and should be checked by the measurements of @xmath231 and @xmath232 of @xmath212 , which is a validation of our scenario .
in addition , constructing the triangle @xmath213 can give us the correlation between @xmath188 and @xmath201 if @xmath36 is fixed .
in this section we would like to present a short discussion about the well - accepted seesaw mechanism @xcite , of which the unitary violation of the mixing matrix @xmath2 is a general consequence .
we show here how to incorporate our parametrization into a generic type - ii seesaw model . in the type - ii seesaw model @xcite which contains @xmath233 right - handed neutrinos ,
the neutrino mass terms can be written as @xmath234 where @xmath235 with @xmath236 being the charge conjugation matrix , likewise for @xmath237 . here
, @xmath238 is a @xmath239 matrix , @xmath240 is a @xmath241 matrix and @xmath242 is a @xmath243 matrix .
the overall @xmath244 neutrino mass matrix in @xmath245 , denoted as @xmath246 , can be diagonalized by the unitary transformation @xmath247 ; or explicitly , @xmath248 where @xmath249 and @xmath250 with @xmath251 and @xmath252 being the light and heavy majorana neutrino masses , respectively .
note that the submatrices @xmath2 , @xmath253 , @xmath254 and @xmath255 are all non - unitary .
suppose the mass eigenstates and the flavor eigenstates of the charged leptons are identical .
in the basis of mass states , the standard charged - current interactions of neutrinos can be written as @xmath256 ~ + ~ { \rm h.c . } \ ; .
% ( 39)\end{aligned}\ ] ] one can find that the matrix @xmath254 describes the strength of the charged - current interaction between the charged leptons and the heavy neutrinos .
in addition , the deviation of @xmath2 from unitary is also characterized by @xmath254 , since the unitarity of @xmath257 requires @xmath258 @xcite .
in fact the diagonalization of @xmath246 can be divided into two steps by decomposing @xmath257 into the product of two unitary matrices @xmath259 and @xmath260 : @xmath261 where @xmath259 and @xmath262 take the general forms @xmath263 the matrices @xmath264 , @xmath265 , @xmath236 and @xmath266 are in general non - unitary , but @xmath267 and @xmath268 are unitary matrices .
we can easily find that : @xmath269 , @xmath270
. let s count the degrees of freedom of these matrices .
the @xmath271 unitary matrix contains @xmath272 degrees of freedom .
suppose freedom left for @xmath259 is @xmath273 the parameters in @xmath267 and @xmath268 are all free , the degrees of @xmath274 .
an ansatz made for @xmath259 in ref .
@xcite is to suppose that @xmath265 is an arbitrary @xmath241 matrix which contains just @xmath275 degrees of freedom and then parametrize @xmath259 as @xmath276 comparing eqs .
( 38 ) , ( 40 ) , ( 41 ) and ( 42 ) with the parametrization @xmath3 shown in section i , we can simply choose : @xmath277 , then we can find that @xmath278 which contains the same degrees of freedom as the hermitian matrix in eq . (
1 ) . that is to say in the parametrization we have chosen , the unitarity - violating parameters in @xmath7 can be expressed as the functions of @xmath279 , where @xmath254 is of the order of @xmath280 . in other words , the deviation from the unitarity which is described by @xmath281 in eq .
( 4 ) is of the order of @xmath282 . in the canonical seesaw scenarios ,
the light neutrino masses are attributed to the leading - order contribution @xmath283 ( type - i seesaw ) or @xmath284 ( type - ii seesaw ) . in order to obtain the light neutrino mass scale @xmath285
, the mass scale of right - handed majorana neutrinos is expected to be as high as @xmath286 .
in such a case , one finds @xmath287 or equivalently @xmath288 , which is too small to be detected .
the possible ways out have recently been discussed ( see , e.g. , refs . @xcite and @xcite ) .
for instance , one may first impose certain flavor symmetries on the textures of @xmath289 , @xmath290 and @xmath291 to guarantee @xmath292 ( type - i seesaw ) or @xmath293 ( type - ii seesaw ) and then introduce slight perturbations to them so as to produce the tiny light neutrino masses . as a consequence ,
the mass scale of three light neutrinos in this approach is essentially decoupled from the magnitude of @xmath254 . in the models proposed in refs .
@xcite , for example , the right - handed majorana neutrinos are assumed to be around the tev scale such that @xmath294 holds and the elements of @xmath281 can be as large as @xmath295 , just close to their upper bounds constrained by current experimental data .
these kinds of seesaw models will be tested at the lhc or ilc @xcite ; and one of their low - energy consequences , which is just the unitarity violation of @xmath2 under discussion , will also be tested in the long - baseline neutrino oscillation experiments .
as we are about to enter an era of high precision neutrino physics , a general and comprehensive analysis of the non - unitary neutrino mixing matrix makes sense and will be useful for phenomenological explanations of future measurements and tests of type - i and type - ii seesaw models . in this paper we have investigated a new pattern of the neutrino mixing matrix which is the product of an arbitrary hermitian matrix and the well - known tri - bimaximal mixing matrix . starting with this non - unitary mixing matrix
, we have presented a complete set of series expansion formulas for neutrino oscillation probabilities both in vacuum and in matter of constant density .
we have carried out a numerical analysis to emphasize the importance of matter effects in the measurements of the non - unitary parameters and in distinguishing their effects from the effects induced by small @xmath0 .
we find that measuring the probability of @xmath109 or @xmath296 oscillation with large neutrino energy ( e.g. , @xmath297 ) and a relatively long baseline ( e.g. , several thousand km ) is a viable way to detect those non - unitary parameters .
we have also discussed the possibility of determining the small non - unitary perturbations and the extra cp - violating phases by constructing the deformed unitarity triangles " .
finally we have shown that our parametrization of the non - unitary mixing matrix can be naturally incorporated into the generic type - ii seesaw model .
i would like to thank prof .
xing for sharing valuable ideas with me and enlightening me on this subject .
i am greatly indebted to him for polishing up the manuscript with many suggestions and corrections .
i am also grateful to w. chao for patient and useful discussions .
this work was supported in part by the national natural science foundation of china .
in this appendix we use the perturbation theory to diagonalize the effect hamiltonian @xmath122 in matter of constant density . in the series expansion of @xmath139 , we regard @xmath35 , @xmath36 , @xmath88 , @xmath25 , @xmath26 , @xmath27 and @xmath298 as the small parameters of the same order , and perform the diagonalization to the second order of them .
now we are going to diagonalize the effective hamiltonian @xmath299 .
first we draw @xmath300 ( which contributes only a pure phase @xmath301 to the oscillation amplitude and nothing to the oscillation probabilities ) out of @xmath114 and rewrite @xmath114 as @xmath302 where @xmath303 .
we find that it is easier to perform the diagonalization of @xmath304 , with @xmath305 \ ; , % ( a2)\ ] ] and @xmath306 \nonumber\\ \nonumber\\ & & - v_{nc}^ { } \left ( 2 \mathbf{\epsilon}_{}^ { * } + { \mathbf{\epsilon}_{}^{*}}^2\right ) - v_{nc}^ { } \cdot \mathbf{1 } \ ; , % ( a3)\end{aligned}\ ] ] the term @xmath307 can be neglected for the same reason as omitting the term @xmath300
. then we will diagonalize @xmath308 with @xmath309 using the perturbation theory .
we can easily find that @xmath310 and @xmath311 .
the matrix @xmath131 , which describes the mixing in matter , can be expressed as @xmath312 , which is also non - unitary .
we write @xmath313 where @xmath314 for the eigenvalues and the eigenvectors , we also write @xmath315 and @xmath316 ( for @xmath317 ) . the unitary matrix @xmath318 .
@xmath319 can be easily diagonalized .
the eigenvalues and the eigenvectors of @xmath319 are : @xmath320 and @xmath321 then , the first and the second order corrections to the eigenvalues are given by @xmath322 and the corrections to the eigenvectors are calculated by @xmath323 \cdot v_{j}^{(0 ) } \nonumber\\ \nonumber\\ & & - \frac{1}{2 } \left [ \sum_{j \neq i}^ { } \frac { \left | \tilde{\cal{h'}}_{ji}^{(1 ) } \right |^2 } { \left ( \tilde{e'}_{i}^{(0 ) } - \tilde{e'}_{j}^{(0 ) } \right ) ^2 } \right ] \cdot v_{i}^{(0 ) } \ ; , % ( a13)\end{aligned}\ ] ] where @xmath324 . inserting eqs .
( a8 ) , ( a9 ) into eqs .
( a10 ) and ( a12 ) , we obtain @xmath325 \right ) \ ; , \\
% ( a15 ) \tilde{e'}_{3}^{(0 ) } & \ ; = \ ; & - v_{nc}^ { } \left ( \epsilon_{b}^ { } + \epsilon_{c}^ { } - 2 { \rm re } [ \hat{\kappa}_{23}^ { } ] \right ) \ ; ; % ( a16)\end{aligned}\ ] ] and @xmath326 \right ) & - \displaystyle \frac{v_{nc}^{}}{2 \sqrt{2 } \delta_{31}^ { } } \left ( \epsilon_{b}^ { } - \epsilon_{c}^ { } - 2 { \rm im}[\hat{\kappa}_{23}^ { } ] \right ) \cr - \displaystyle \frac{2 \delta_{21}^{}}{3 v_{cc}^ { } l } & \displaystyle \frac{v_{nc}^{}}{2 \sqrt{2 } \delta_{31}^ { } } \left ( \epsilon_{b}^ { } - \epsilon_{c}^ { } + 2 { \rm im}[\hat{\kappa}_{23}^ { } ] \right ) & ~ \displaystyle \frac{v_{nc}^{}}{2 \sqrt{2 } \delta_{31}^ { } } \left ( \epsilon_{b}^ { } - \epsilon_{c}^ { } - 2 { \rm im}[\hat{\kappa}_{23}^ { } ] \right ) \end{matrix } \right ) \ ; , % ( a17)\ ] ] where @xmath58 with @xmath59 ( for @xmath327 ) , and the terms proportional to @xmath328 are omitted .
we can further calculate @xmath329 to the first order in @xmath35 , @xmath36 , @xmath88 , @xmath25 , @xmath26 , @xmath27 and @xmath330 : @xmath331 \right ) } { 2 \sqrt{2 } \delta_{31}^ { } } & \displaystyle \frac { b + \hat{\kappa}_{23}^{}}{\sqrt{2 } } - \frac{v_{nc}^ { } \left ( \epsilon_{b}^ { } - \epsilon_{c}^ { } - 2 { \rm im}[\hat{\kappa}_{23}^ { } ] \right ) } { 2 \sqrt{2 } \delta_{31}^ { } } \cr - \displaystyle \frac{2 \delta_{21}^{}}{3 v_{cc}^ { } l } + \hat{\kappa}_{13}^ { * } & \displaystyle - \frac{c - \hat{\kappa}_{23}^{}}{\sqrt{2 } } + \frac{v_{nc}^ { } \left ( \epsilon_{b}^ { } - \epsilon_{c}^ { } + 2 { \rm im}[\hat{\kappa}_{23}^ { } ] \right ) } { 2 \sqrt{2 } \delta_{31}^ { } } & \displaystyle \frac{c + \hat{\kappa}_{23}^{}}{\sqrt{2 } } + \frac{v_{nc}^ { } \left ( \epsilon_{b}^ { } - \epsilon_{c}^ { } - 2 { \rm im}[\hat{\kappa}_{23}^ { } ] \right ) } { 2 \sqrt{2 } \delta_{31}^ { } } \end{matrix } \right ) \ ; .
% ( a18)\ ] ] in this paper we do not order the eigenvalues of @xmath141 according to their magnitude and the mass spectrum . this ordering does not change the oscillation probabilities . in this appendix
we give the results of @xmath332 , @xmath333 , @xmath334 and @xmath335 , which are enough for calculating the probabilities @xmath98 , @xmath336 and @xmath337 to the second order .
@xmath338 and @xmath339 , which are not shown here , only correct the terms of the second order in @xmath340 , @xmath341 and @xmath103 .
sides of six deformed unitarity triangles " to the first order in @xmath35 , @xmath36 , @xmath88 , @xmath25 , @xmath26 and @xmath27 .
* @xmath211 : @xmath342 with @xmath343 .
* @xmath212 : @xmath344 with @xmath345 .
* @xmath213 : @xmath346 with @xmath347 . * @xmath214 : @xmath348 -3 i { \rm im}[\hat{\kappa}_{12}^ { } - \hat{\kappa}_{13}^ { } ] \right ) \ ; , \\
% ( b10 ) s_{2}^ { } & \ ; = \ ; & v_{\mu 1}^ { } v_{\mu 2}^ { * } \ ; \approx \ ; - \frac{1}{3 \sqrt{2 } } \left ( 1 + 2 \epsilon_{b}^ { } \right ) + \frac{1}{3 \sqrt{2 } } \left ( { \rm re}[\hat{\kappa}_{12}^ { } + 2 \hat{\kappa}_{23}^ { } ] -3 i { \rm im}[\hat{\kappa}_{12}^ { } ] \right ) \ ; , \\ % ( b11 ) s_{3}^ { } & \ ; = \ ; & v_{\tau 1}^ { } v_{\tau 2}^ { * } \ ; \approx \ ; - \frac{1}{3 \sqrt{2 } } \left ( 1 + 2 \epsilon_{c}^ { } \right ) - \frac{1}{3 \sqrt{2 } } \left ( { \rm re}[\hat{\kappa}_{13}^ { } - 2 \hat{\kappa}_{23}^ { } ] -3 i { \rm im}[\hat{\kappa}_{13}^ { } ] \right ) \ ; , % ( b12)\end{aligned}\ ] ] with @xmath349 -3 i { \rm im}[\hat{\kappa}_{12}^ { } - \hat{\kappa}_{13}^ { } ] \right ) $ ] . * @xmath215 : @xmath350 \right ) \ ; , \\ % ( b14 ) s_{3}^ { } & \ ; = \ ; & v_{\tau 1}^ { } v_{\tau 3}^ { * } \ ; \approx \ ; \frac{1}{2 \sqrt{3 } } \left ( 1 + 2 \epsilon_{c}^ { } \right ) + \frac{1}{\sqrt{3 } } \left ( \hat{\kappa}_{13}^ { * } + i { \rm im}[\hat{\kappa}_{23}^ { } ] \right ) \ ; , % ( b15)\end{aligned}\ ] ] with @xmath351 \right ) $ ] .
* @xmath216 : @xmath352 \right ) \ ; , \\ % ( b17 ) s_{3}^ { } & \ ; = \ ; & v_{\tau 2}^ { } v_{\tau 3}^ { * } \ ; \approx \ ; - \frac{1}{\sqrt{6 } } \left ( 1 + 2 \epsilon_{c}^ { } \right ) + \frac{1}{\sqrt{6 } } \left ( \hat{\kappa}_{13}^ { * } - 2 i { \rm im}[\hat{\kappa}_{23}^ { } ] \right ) \ ; , % ( b18)\end{aligned}\ ] ] with @xmath353 \right ) $ ] .
kamland collaboration , k. eguchi _
et al . _ ,
lett . * 90 * , 021802 ( 2003 ) ; chooz collaboration , m. apollonio
_ et al . _ ,
b * 420 * , 397 ( 1998 ) ; palo verde collaboration , f. boehm _
et al . _ ,
lett . * 84 * , 3764 ( 2000 ) . for recent reviews with extensive references ,
see : h. fritzsch and z.z .
xing , prog .
* 45 * , 1 ( 2000 ) ; z.z .
xing , int .
a * 19 * , 1 ( 2004 ) ; g. altarelli and f. feruglio , new j. phys .
* 6 * , 106 ( 2004 ) ; r.n .
et al . _ , hep - ph/0510213 ; r.n .
mohapatra and a.yu .
smirnov , hep - ph/0603118 ; a. strumia and f. vissani , hep - ph/0606054 .
p. minkowski , phys .
b * 67 * , 421 ( 1977 ) ; t. yanagida , in _ proceedings of the workshop on unified theory and the baryon number of the universe _ , edited by o. sawada and a. sugamoto ( kek , tsukuba , 1979 ) ; m. gell - mann , p. ramond , and r. slansky , in _ supergravity _ , edited by p. van nieuwenhuizen and d. freedman ( north holland , amsterdam , 1979 ) ; s. l. glashow , in _ quarks and leptons _ , edited by m. l@xmath354vy _ et al . _ ( plenum , new york , 1980 ) ; r. n. mohapatra and g. senjanovic , phys . rev
44 * , 912 ( 1980 ) .
s. m. bilenky and c. giunti , phys .
b * 300 * , 137 ( 1993 ) .
m. czakon , j. gluza and m. zralek , acta phys .
b * 32 * , 3735 ( 2001 ) .
b. bekman , _ et al .
d * 66 * , 093004 ( 2002 ) .
j. holeczek , j. kisiel , j. syska and m. zralek , eur .
j. c * 52 * 905 ( 2007 ) ; j. lopez - pavon , aip conf .
proc . * 981 * 219 ( 2008 ) ; s. goswami and t. ota , arxiv:0802.1434 [ hep - ph ] .
some recent papers on possible deviations from the tri - bimaximal neutrino mixing can be found in : x. g. he and a. zee , phys .
b * 645 * , 427 ( 2007 ) ; s. luo and z.z .
xing , phys .
b * 646 * , 242 ( 2007 ) ; z.z .
xing and s. zhou , phys .
b * 653 * , 278 ( 2007 ) ; a. h. chan , h. fritzsch , s. luo and z. z. xing , phys .
d * 76 * , 073009 ( 2007 ) ; s. f. king , arxiv:0710.0530 [ hep - ph ] ; s. antusch , s. f. king and m. malinsky , arxiv:0711.4727 [ hep - ph ] ; a. mondragon , m. mondragon and e. peinado , arxiv:0712.2488 [ hep - ph ] ; s. luo and z.z .
xing , arxiv:0712.2610 [ hep - ph ] ; m. honda and m. tanimoto , arxiv:0801.0181 [ hep - ph ] ; c. h. albright and w. rodejohann , arxiv:0804.4581 [ hep - ph ] .
j. schechter and j. w. f. valle , phys .
d * 22 * , 2227 ( 1980 ) ; t. p. cheng and l. f. li , phys .
d * 22 * , 2860 ( 1980 ) ; m. magg and c. wetterich , phys .
b * 94 * , 61 ( 1980 ) ; g. lazarides , q. shafi and c. wetterich , nucl . phys
b * 181 * , 287 ( 1981 ) .
t. han and b. zhang , phys .
lett . * 97 * , 171804 ( 2006 ) ; f. del aguila , j. a. aguilar - saavedra , and r. pittau , j. phys .
ser . * 53 * , 506 ( 2006 ) ; f. m. l. de almeida _ et al .
d * 75 * , 075002 ( 2007 ) ; t. han , b. mukhopadhyaya , z. si , and k. wang , phys . rev .
d * 76 * , 075013 ( 2007 ) ; w. chao , z. si , z.z . xing , and s. zhou , arxiv:0804.1265 [ hep - ph ] . | we propose a new pattern of the neutrino mixing matrix which can be parametrized as the product of an arbitrary hermitian matrix and the well - known tri - bimaximal mixing matrix . in this scenario ,
nontrivial values of the smallest neutrino mixing angle @xmath0 and the cp - violating phases entirely arise from the non - unitary corrections .
we present a complete set of series expansion formulas for neutrino oscillation probabilities both in vacuum and in matter of constant density .
we do a numerical analysis to show the non - unitary effects on neutrino oscillations .
the possibility of determining small non - unitary perturbations and cp - violating phases is discussed by measuring neutrino oscillation probabilities and constructing deformed unitarity triangles " .
some brief comments on the non - unitary neutrino mixing matrix in the type - ii seesaw models are also given . |
so far the top quark properties measured at the tevatron are in good agreement with the standard model ( sm ) predictions except the inclusive discrepancy of @xmath3 for @xmath4 gev because it is not confirmed by the d0 collaboration . ]
forward - backward asymmetry @xmath0 @xcite , which , as reported by the cdf collaboration and the d0 collaboration , exceeds the sm prediction by about @xmath5 @xcite .
such an anomaly has been widely speculated as a harbinger of new physics and thus stimulated various explanations in extensions of the sm @xcite .
these extensions , albeit in quite different forms , usually have rich top quark phenomenology at colliders .
since the tevatron is going to be shut down very soon , the task to screen out the right theory is left for the lhc @xcite .
although the present top quark dataset at the lhc is moderate , it is already capable of scrutinizing the validity of some extensions .
for example , the non - observation of a clear resonance in the @xmath6 production searched by the atlas and cms collaborations at @xmath7 tev implies that an axigluon with strong couplings to light quarks should be heavier than 3.2 tev @xcite , which makes it less attractive as an explanation of @xmath0 @xcite ( however , as pointed in the last reference in @xcite , a light axigluon with an enlarged width and reduced couplings to light quarks is still allowed by the current lhc measurements ) .
meanwhile , since no excess of same - sign top quark events was observed by recent measurements from the lhc and tevatron @xcite , the light @xmath8 model based on flavor non - universal @xmath9 symmetry @xcite is also disfavored . among the surviving models two typical ones are the @xmath1 model @xcite and the diquark ( @xmath2 ) model @xcite , which , as pointed in @xcite , are preferred by the combined fit of @xmath10 and the total @xmath6 production rate measured at the tevatron . in this work
we focus on these two models and perform a comparative study by considering several observables at the lhc .
our study shows that each of these observables can be enhanced to the observable level and meanwhile exhibits different characteristics in these two models . as a result
, the @xmath1 model is found to be tightly constrained by the charge asymmetry in @xmath6 production at the lhc , while the diquark model can be readily explored once more luminosity is accumulated at the lhc .
we will consider the following observables : * top quark charge asymmetry in @xmath6 production at the lhc , which is defined by @xcite @xmath11 ( @xmath12 ) is the pseudo - rapidity of the top ( anti - top ) quark in the laboratory frame , and @xmath13 denotes cross section .
this asymmetry reflects whether the top quarks on average are more boosted than the anti - top quarks or not .
we note that the cms collaboration has recently measured this quantity with an integrated luminosity of 1.09 fb@xmath14 and obtained @xmath15 , which is consistent with its sm prediction @xmath16 @xcite .
a similar result is also reported by the atlas collaboration with larger uncertainties @xcite .
so this asymmetry can be used to limit new physics models @xcite . *
top quark polarization asymmetry in @xmath6 production at the lhc , defined by @xcite @xmath17 with the first ( second ) subscript of @xmath13 denoting the helicity of the top ( anti - top ) quark . unlike light quarks , top quark decays rapidly before forming any hadronic bound state .
so its spin information is preserved by its decay products and can be recovered by their angular distributions . for the @xmath18 production at the lhc ,
the top quark is not polarized at the leading order of the sm because the production proceeds mainly through the qcd interaction and the parity - violating electro - weak contribution to the polarization is negligibly small @xcite , but any addition of new parity - violating interaction of top quark may induce sizable polarization asymmetry @xcite .
* enhancement factor of the @xmath6 production rate in high invariant mass region of @xmath6 : @xmath19 where @xmath20 incorporates the contributions from the sm and the new physics . in exotic @xmath21-channel or @xmath22-channel @xmath6 production ,
the rutherford singularity can alter significantly the distribution of the @xmath6 invariant mass in high energy tail @xcite , so @xmath23 may deviate significantly from unity . *
charge asymmetry in the associated production of a single top with a particle @xmath24 : @xmath25 this asymmetry can be measured by requiring that the top quark decay semi - leptonically and @xmath24 decay hadronically , and looking for the asymmetry between the event numbers with one lepton and one anti - lepton in the signal respectively .
it was once suggested in searching for single top production in the sm and in limiting new physics models @xcite .
depending on @xmath26 and the initial partons in @xmath27 production , @xmath28 may be far larger or smaller than unity .
* charge asymmetry in @xmath29 production defined by @xmath30 like @xmath31 , this asymmetry reflects whether @xmath32 or @xmath33 is more boosted .
given the interactions of the particle @xmath24 with quarks , this asymmetry is determined by @xmath26 and the energy of the lhc . this paper is organized as follows . in sec .
ii , we briefly describe the features of the @xmath1 model and diquark model . then in sec .
iii we discuss some observables in @xmath6 production , single top production and @xmath34 ( @xmath2 ) pair production . finally , we draw our conclusion in sec
among various explanations of the @xmath0 anomaly , the model with a color singlet @xmath1 is a promising one @xcite .
this model can be realized in an asymmetric left - right framework @xcite presented in appendix a , which is based on the gauge group @xmath35 and assumes that only the first and third generation right - handed quarks transform nontrivially under the group @xmath36 .
the interaction relevant to our calculation is given as @xmath37 the @xmath6 production then gets additional contribution from the @xmath21-channel process @xmath38 via exchanging a @xmath1 , which may sizably alter @xmath0 at the tevatron .
note that in the framework presented in appendix a , besides @xmath1 , the newly predicted neutral and charged higgs bosons can also contribute to the @xmath6 production . since the size of such contribution is model - dependent and
may be negligible if these fields are heavy and/or the vev of @xmath39 is much higher than the electro - weak breaking scale @xcite , we in our study do not consider these contributions .
another model we are considering is the color - triplet diquark model @xcite , where a new scalar @xmath2 ( called diquark ) is assigned with the quantum number ( @xmath40 , @xmath41 , @xmath42 ) under the sm gauge group @xmath43 .
the relevant lagrangian is then given by @xmath44 where the coupling coefficients satisfy @xmath45 with @xmath46 being the flavor index , @xmath47 is the antisymmetric tensor in color space , and @xmath48 with @xmath49 being the charge conjugate matrix . in this framework
, the discrepancy of @xmath0 can be alleviated by the contribution of the @xmath22-channel process @xmath50 mediated by the triplet @xmath2 . in @xcite ,
a comparative study of @xmath0 was performed in diquark models where @xmath2 is assigned in different representations of the @xmath51 group , and it was found that the triplet model is better suited to explain the @xmath0 anomaly without conflicting with other experimental results . in our analysis , in order to escape constraints from low energy processes such as @xmath52@xmath53 mixing , we set @xmath54 to be zero except @xmath55 .
the common feature of the two models comes from the calculation of the @xmath6 production rate , where the interference of the new contribution with the sm qcd amplitude always partially cancels the pure new contribution .
in fact , this cancellation is essential for the models to explain the @xmath0 anomaly and at same time keeps other observables consistent with their measured values at the tevatron .
we checked that such cancellation persists in calculating @xmath56 discussed below , and the extent of the cancellation depends on the new particle mass and the collider energy .
we also checked that , partially due to the difference in parton distributions for the initial states , @xmath57 in the diquark model usually exceeds that in the @xmath1 model if @xmath58 and @xmath59 .
in this section we present the numerical results for the observables at the lhc with @xmath7 tev . we take the sm parameters as @xcite @xmath60 and use the parton distribution function cteq6l1 @xcite by setting @xmath61 with @xmath62 and @xmath63 denoting the renormalization scale and the factorization scale respectively . for the constraints from the @xmath6 production rates , we consider the tevatron measurements @xcite , which are so far the most precise results .
we require the predictions of the inclusive @xmath0 and the total @xmath6 production rate in each model to lie within @xmath64 region of their experimental values . as mentioned earlier
, we do not consider the discrepancy of the @xmath0 in large @xmath6 invariant mass region reported by the cdf collaboration ( about @xmath65 away from its sm prediction for @xmath66 gev@xcite ) since it is not confirmed by the d0 collaboration @xcite .
we also do not consider the constraint from the measured @xmath6 invariant mass distribution at the tevatron because the shape of such a distribution in high energy tail is sensitive to the cut efficiency of event selection and also to qcd corrections @xcite . before presenting our results for @xmath67
, we point out two features of @xmath0 . first , because the valence quark in proton always moves in parallel with the proton , @xmath68 observed at the tevatron means that the top quark tends to move along with the valence quark than to move in the opposite direction .
second , @xmath0 depends on the collider energy @xmath69 .
we found that as @xmath69 increases , @xmath10 increases monotonically in the @xmath1 model but decreases monotonically in the diquark model .
this means that if the two models predict a same @xmath0 at the tevatron , then as @xmath69 increases to the lhc energy , the tendency of top quark to move with the valence quark ( @xmath22 or @xmath70 ) in the @xmath1 model should be larger than that in the diquark model . in fig .
[ fig1 ] we show the correlation between @xmath71 at the tevatron and @xmath31 at the lhc in these two models .
such results are obtained by scanning over the two - dimension parameter space of the models and keeping only the samples surviving the tevatron constraints .
we see that @xmath71 and @xmath31 are of the same sign and with the increase of @xmath57 the value of @xmath31 increases too .
this behavior can be understood by noting the following three points .
the first is that in the @xmath6 rest frame the top and the anti - top outgo back to back .
so , regardless the underlying dynamics , we always have latexmath:[$|\eta_t| = process @xmath38 or the @xmath22-channel process @xmath73 at @xmath74 colliders like the lhc , the @xmath6 rest frame tends to be boosted along the direction of @xmath70 or @xmath22 quark since they are the valence quarks in proton . for a given event ,
the direction of the valence quarks is definite .
then , if the scattering angle @xmath75 ( @xmath76 ) between the outgoing top quark and the valence quark in @xmath6 rest frame is less ( larger ) than @xmath77 , @xmath78 defined in the laboratory frame tends to be larger ( less ) than @xmath79 .
and the last point is if the top quark has equal probability to move along and to move in opposite to the valence quark direction at the lhc ( corresponding to @xmath80 in @xmath81 collision ) , the number of events with @xmath82 should be same as that with latexmath:[$|\eta_t| < former probability exceeds the latter probability ( corresponding a positive @xmath0 in @xmath81 collision ) , more events with @xmath82 than with @xmath84 should be obtained and thus @xmath67 is positive .
this analysis shows that @xmath85 at the tevatron can be treated as an indicator of @xmath67 at the lhc .
[ fig1 ] also indicates that @xmath67 in the @xmath1 model is usually several times larger than that in the diquark model for a given value of @xmath0 .
one underlying reason is , as we mentioned before , the probability of the top quark to move along with the valence quark in the @xmath1 model exceeds that in the diquark model .
another reason is from the parton distribution of the initial states : at the tevatron we have @xmath86 , while at the lhc @xmath87 .
so when both models predict a same @xmath0 at the tevatron , the parton distribution in the @xmath1 model is relatively enhanced at the lhc .
another striking feature of fig .
[ fig1 ] is that a large portion of the samples in the @xmath1 model have been ruled out by the measured value of @xmath67 at @xmath88 level , which implies that the @xmath1 model has already been tightly limited by the charge asymmetry .
in contrast , in the diquark model the @xmath67 value always lie within @xmath88 range of its experimental central value .
we checked that the @xmath67 value in the diquark model will be further reduced at the lhc as @xmath69 is raised to 14 tev . in getting fig.1
, we note that , since the new physics contributions to the @xmath6 cross section are relatively small , both @xmath56 and @xmath3 can be approximated as the sm value plus the new physics effect : @xmath89 and @xmath90 . for the values of @xmath91 and @xmath92
, we use their nlo qcd results : @xmath93 @xcite and @xmath94 ( which is obtained by the mcfm package @xcite ) . in calculating @xmath95 and @xmath96 ,
we encounter two kinds of cross sections : the sm cross sections @xmath97 and the new physics corrections @xmath98 .
we use the tree - level expression of @xmath99 due to the absence of its high order qcd correction in literatures , while for the @xmath97 , we use its most precise nnlo result , which is obtained by multiplying its lo prediction by a k factor , i.e. @xmath100 for the lhc @xcite and @xmath101 for the tevatron @xcite . in fig .
[ fig2 ] we show the dependence of @xmath31 on the model parameters such as the coupling strength and the new particle mass . due to the difference in kinematic features of the @xmath21 and the @xmath22 channels , the mass ranges favored by @xmath3 and @xmath102 are @xmath103 and @xmath104 for the two models respectively .
this figure indicates that for a given new particle mass the coupling coefficient ( @xmath55 or @xmath105 ) is restricted in a certain region , and as the new particle becomes heavy , the region moves upward .
this is because we have required the samples shown in the figure to explain the @xmath0 anomaly and at same time to satisfy the @xmath106 constraint .
this figure also indicates that a heavy new particle along with a strong coupling can predict a large @xmath31 .
we checked this case and found it usually corresponds to a large @xmath0 at the tevatron . with @xmath23 and @xmath107 at the lhc respectively .
[ fig3],title="fig:",width=240,height=307 ] with @xmath23 and @xmath107 at the lhc respectively .
[ fig3],title="fig:",width=240,height=307 ] in the left frame of fig .
[ fig3 ] we show the correlation of the @xmath31 with the ratio @xmath23 defined by eq .
( [ r1 ] ) . as we mentioned before , for the @xmath21-channel or the @xmath22-channel @xmath6 production , the rutherford singularity tends to push more events to high @xmath108 region so that @xmath23 may be significantly larger than unity .
this is reflected in the @xmath1 model where @xmath23 is in the range of 2.0 and 7.7 and in the diquark model where @xmath23 varies from 1.2 to 2.7 .
since the predicted @xmath23 is in two separated regions , @xmath23 may be utilized to discriminate the models .
we checked the reason for the difference and found that the cancellation between the pure new physics contribution and the interference contribution in the @xmath1 model is not as strong as that in the diquark model .
we also note that the lhc with higher luminosity is capable of exploring the models with @xmath109 @xcite .
so we conclude that the quantity @xmath23 is complementary to @xmath67 in testing the models .
since the new interactions violate parity and hence can lead to top quark polarization asymmetry @xmath107 at the lhc , in the right frame of fig .
[ fig3 ] we show the correlation of @xmath31 with @xmath107 .
this figure indicates that the value of @xmath107 increases with the increase of @xmath31 with its maximum value reaching @xmath110 and @xmath111 for the two models respectively . to roughly estimate the observability of such asymmetry
, we calculate the statistical significance @xmath112 defined in @xcite for an integrated luminosity of 1 fb@xmath14 without considering the cut efficiency and the systematic uncertainties .
we find that for nearly all the samples in the models , the predicted @xmath107 can reach its @xmath113 sensitivity , which is @xmath114 for the @xmath34 model and @xmath115 for the diquark model . in the @xmath1 ( diquark ) model
, the associated production of single top quark with @xmath1 ( @xmath2 ) proceeds by the feynman diagrams shown in fig .
the total production rates ( top events plus anti - top events ) can reach 60 pb and 160 pb for the surviving samples in the two models respectively . due to the electric charge carried by @xmath116 ( @xmath117 ) ,
the production rates of the top quark and anti - top quark are not equal . since the initial state is @xmath118 ( @xmath119 ) for the top production and @xmath120 ( @xmath121 ) for the anti - top production , the parton distributions determine @xmath122 for the @xmath1 model and @xmath123 for the diquark model , where @xmath28 denotes the charge asymmetry of the associated production defined in eq .
( [ r2 ] ) . from fig .
[ fig5 ] , we find @xmath124 in the @xmath1 model while @xmath125 in the diquark model . in our calculation
we also find that , although the rate of the @xmath126 production decreases monotonically as @xmath1 becomes heavy , the ratio @xmath127 increases .
the reason is that the distribution function of the sea quark @xmath128 is more suppressed in high proton momentum fraction region . and @xmath28 at the lhc.,width=240,height=307 ] in order to further test two models , we investigate the kinematical distributions of the single top productions . as an illustration
, we take the best point for each model .
the best point is determined by minimizing the @xmath129 function defined as @xmath130 where the observables @xmath131 are @xmath0 and @xmath102 at the tevatron and @xmath31 at the lhc .
we add the experimental and the sm errors in quadrature to calculate @xmath132 . for the @xmath1 model the best point
is found to be at @xmath133 and @xmath134 gev , with @xmath135 ; while for diquark model the best point is at @xmath136 and @xmath137 gev , with @xmath138 . in table
i we present the predictions for the observables at the best points . 0.1 in .predictions of the @xmath1 model and the diquark model at the best point .
@xmath24 denotes @xmath34 or @xmath2 .
new physics contributions to the cross sections at the tevatron ( lhc ) are in unit of fb ( pb ) . [ cols="^,^,^,^,^,^,^,^,^,^,^ " , ] in our analysis we assume @xmath116 and @xmath117 mainly decay as @xmath139 and @xmath140 with the anti - top quark decaying hadronically so that @xmath1 and @xmath2 can be reconstructed . in this way , the associated productions may be disentangled from the @xmath6 production @xcite which acts as the main background . using the madgraph5/madevent @xcite , we study the signal @xmath141 at the parton level under the basic cuts at the lhc , where @xmath142 denotes the missing transverse energy . and @xmath143 in the single top productions at the lhc .
here the @xmath144-jet and the light jet are required from same new particle.,title="fig:",width=288,height=336 ] and @xmath143 in the single top productions at the lhc . here
the @xmath144-jet and the light jet are required from same new particle.,title="fig:",width=288,height=336 ] in fig .
6 we display the distributions of the total transverse energy @xmath145 and the angle between the @xmath144-jet and the light jet coming from @xmath34(@xmath2 ) , which are all defined in the laboratory frame .
the left panel of this figure shows that the most events from @xmath146 have lower @xmath145 than those from @xmath147 .
the reason is that in the considered case @xmath1 is lighter than the diquark state .
the right panel shows that the @xmath144-jet is inclined to fly along the light jet in the @xmath34 model , while to fly in opposite to the light jet in the diquark model .
this is because , although the decay products of @xmath148(@xmath2 ) are boosted along the direction of @xmath148(@xmath2 ) , the massive anti - top from the @xmath148(@xmath2 ) decay may kick its @xmath144-jet in certain direction so that the @xmath144-jet can deviate from the boost direction .
actually , we find that the @xmath144-jet from a left - handed anti - top quark ( as in the @xmath148 model ) tends to fly along the direction of the anti - top quark @xcite , which is also the direction of the light jet from the @xmath148 decay ; while the @xmath144-jet from a right - handed anti - top quark ( as in the case in the diquark model ) tends to fly in the opposite direction . for the charge asymmetry in single top production , due to the large jet multiplicities and moderate @xmath144-tagging efficiency in the process , the measurement will be somewhat challenging at the lhc .
however , we noted that the peak values of @xmath145 ( @xmath149gev ) in both models are much larger than that in the sm ( @xmath150gev ) . with higher luminosity and higher kinematic cuts ,
the measurements of the differential cross sections and the single top charge asymmetries versus @xmath145 will be useful to discover the signals @xcite .
moreover , the @xmath144-jet angular distribution may serve as a complementary discriminator for the background , since the distribution of @xmath143 in the sm is relatively flat in comparison with the signals .
the detailed analysis of the backgrounds depends on the full detector simulation which is partially studied in ref . @xcite . due to the interactions introduced in sec .
ii , the @xmath151 production proceeds only by the parton process @xmath153 through exchanging a top quark , while the @xmath152 production may proceed either by @xmath154 or by @xmath155 ( via @xmath156 and @xmath157 interactions ) .
we checked our results for the @xmath152 production and found that the gluon annihilation contribution is usually negligibly small .
one main reason is that for the surviving samples presented in fig .
[ fig2 ] , @xmath2 is usually heavy and thus the gluon distribution in proton is suppressed .
we also found that , for given @xmath158 , the @xmath152 production rate is slightly lower than the @xmath151 rate .
this is shown in fig .
[ fig7 ] , where one can learn that for @xmath159 gev , @xmath160 may exceed 6 pb while @xmath161 can only reach 4 pb .
although the pair production rates are moderate at the lhc with @xmath7 tev , the charge asymmetry @xmath162 can still be sizable because it only reflects the unbalance between the particle and its charge conjugate state in boosting along the valence quarks . in fig .
[ fig8 ] we show the charge asymmetry @xmath162 in the two models .
this figure indicates that in the @xmath1 model the @xmath163 fluctuates around zero , while in the diquark model @xmath164 varies between @xmath165 and @xmath166 .
these results can be understood from fig .
[ fig7 ] , which shows that for @xmath167 gev the cross section with @xmath168 is slightly larger than that with @xmath169 , and with the increase of @xmath170 this relation is reversed ; while in the diquark model the corresponding former rate is always larger than the latter rate to obtain a significant negative @xmath164 .
we note that in the sm the value of @xmath162 for the @xmath171 production is positive , while in the @xmath1 model the value of @xmath163 is negative for a light @xmath1 .
the difference comes from the masses of mediators . in the sm ,
the main contribution to the @xmath172 production is through the @xmath21 channel by mediating a massless light quark , while in the @xmath148 model , it is top quark that mediates the process of the @xmath151 production .
we checked that if we set @xmath173 to zero , @xmath162 in @xmath1 pair production will become positive as @xmath174 in the sm .
we also note that in the diquark model , even with the constraints from @xmath67 , the value of @xmath164 can still deviate significantly from zero .
we checked that at the lhc with @xmath175 tev the rates for these productions are usually enhanced by about @xmath176 times , while @xmath162 changes little in both models .
in this paper we discussed the potential of the lhc to discriminate the @xmath34 model and the diquark model which were used to explain the @xmath0 anomaly measured at the tevatron . with the constraints from the tevatron , we examine the charge and polarization asymmetries in @xmath6 production , the charge asymmetries in single top production and @xmath1@xmath177 pair production at the lhc with @xmath178 tev .
we found that the predictions of these observables may be large enough to reach their detectable levels at the lhc . in
particularly , the recent measurement of the charge asymmetry in @xmath6 production from the lhc has already imposed a strong limit on the @xmath1 explanation of the @xmath0 anomaly .
we also found that each observable in the two models shows different characteristics and a joint analysis of these observables at the lhc can help to discriminate the two models .
lei wu thanks fabio maltoni and johan alwall for helpful discussion of madgraph .
this work was supported in part by hastit under grant no .
2009hastit004 , by the national natural science foundation of china ( nnsfc ) under grant nos .
10821504 , 10725526 , 10775039 , 11075045 , by the project of knowledge innovation program ( pkip ) of chinese academy of sciences under grant no .
kjcx2.yw.w10 , and by the grant - in - aid for scientific research ( no .
14046201 ) from japan .
the asymmetric left - right model with light @xmath148 was proposed in @xcite .
it is based on the gauge group @xmath35 and assumes that only the first and third generation right - handed quarks transform nontrivially under the group @xmath36@xcite .
the symmetry breaking starts with @xmath179 to obtain the sm hypercharge @xmath180 , and subsequently @xmath181 to obtain @xmath182 .
for the first breaking , a @xmath36 triplet higgs field is introduced so that the neutral gauge bosons @xmath8 of the @xmath36 group is significantly heavier than the charged boson @xmath34@xcite .
two distinctive features of the model are exhibited in @xcite .
one is , after choosing specific rotation matrices to transform right - handed quarks from flavor basis to mass eigenstates , @xmath1 may couple to flavors in the combination @xmath183 with unsuppressed strength , while @xmath8 only has flavor conserving interactions , i.e. @xmath184 such specific choice , as shown in @xcite , is phenomenologically favored by several anomalies in top physics and b physics observed at the tevatron . the second feature is , unlike the traditional flavor universal left - right model where the quarks acquire masses by interacting with @xmath185 bi - doublet fields@xcite ,
the quark masses are generated in a complex way .
for example , the first and third generation right - handed quarks may have higgs terms like @xmath186 where flavor indices @xmath187 and @xmath188 are @xmath189 and @xmath190 , @xmath39 and @xmath191 are doublet fields under the group @xmath36 and @xmath192 respectively with @xmath193 and @xmath194 , and @xmath195 denotes the vacuum expectation value ( vev ) of the neutral component of @xmath39 ; whereas the second generation right - handed quarks take on the more conventional form @xmath196 obviously , once the field @xmath39 gets its vev the sm mechanism for mass generation is recovered with the quark yukawa coupling coefficients @xmath197 given by @xmath198 for @xmath199 and @xmath200 for @xmath201 .
in addition , as suggested by @xcite , the five dimension operators in eq.([operator ] ) may be generated by integrating out heavy @xmath202-singlet fermions with mass scale @xmath203 , which usually carry appropriate hypercharge . in the @xmath1 model ,
the additional contribution to the @xmath6 production comes from the @xmath21-channel process @xmath204 via the exchange of @xmath1 or neutral / charged component fields of the @xmath39 .
obviously , if the component fields are heavy and/or if @xmath195 is much larger than the electro - weak breaking scale so that the @xmath205 interactions are suppressed ( see eq.[operator ] ) , the latter contribution can be safely neglected , which was done in literature @xcite .
99 j. h. khn _ et al .
_ , 59 , 054017 ( 1999 ) ; m. t. bowen , s. d. ellis and d. rainwater , phys .
d * 73 * , 014008 ( 2006 ) ; v. ahrens _ et al .
_ , jhep * 1009 * , 097 ( 2010 ) ; arxiv:1106.6051 [ hep - ph ] ; n. kidonakis , arxiv:1105.5167 [ hep - ph ] .
w. hollik and d. pagani , arxiv:1107.2606 [ hep - ph ] .
t. aaltonen _ et al . _
[ the cdf collaboration ] , 83 , 112003 ( 2011 ) ; y. takeuchi _ et al . _ ,
http://www-cdf.fnal.gov/physics/new/top/2011/dilafb/note 10398 ; q .- h .
cao _ et al .
_ , phys .
rev . * d81 * , 114004 ( 2010 ) ; g. rodrigo and p. ferrario , nuovo cim .
c * 33 * , 04 ( 2010 ) ; m. i. gresham , i. w. kim and k. m. zurek , phys .
d * 83 * , 114027 ( 2011 ) ; j. f. kamenik , j. shu and j. zupan , arxiv:1107.5257 [ hep - ph ] ; s. westhoff , arxiv:1108.3341 [ hep - ph ] .
p. ferrario , g. rodrigo , 80 , 051701 ( 2009 ) ; p. ferrario and g. rodrigo , jhep * 1002 * , 051 ( 2010 ) ; p. h. frampton _
_ , 683 , 294 ( 2010 ) ; m. v. martynov , a. d. smirnov , mod .
a * 25 * , 2637 ( 2010 ) ; r. s. chivukula _ et al .
_ , 82 , 094009 ( 2010 ) ; y. bai _ et al .
_ , jhep * 1103 * , 003 ( 2011 ) ; a. djouadi _ et al . _ , 82 , 071702 ( 2010 )
; k. kumar _ et al .
_ , jhep * 1008 * , 052 ( 2010 ) ; g. burdman _ et al .
_ , phys .
d * 83 * , 035012 ( 2011 ) ; e. alvarez _ et al .
_ , jhep * 1105 * , 070 ( 2011 ) ; c. delaunay _ et al .
_ , arxiv:1101.2902 . m. bauer _ et al .
_ , jhep 1011 , 039 ( 2010 ) ; b. xiao _ et al .
_ , arxiv:1011.0152 [ hep - ph ] ; c. h. chen _ et al .
_ , phys .
b * 694 * , 393 ( 2011 ) ; r. foot , phys . rev .
d * 83 * , 114013 ( 2011 ) ; a. djouadi _ et al .
_ , phys .
b * 701 * , 458 ( 2011 ) ; r. barcelo _ et al .
_ , arxiv:1105.3333 [ hep - ph ] ; e. alvarez _ et al .
_ , arxiv:1107.1473 ; e. gabrielli and m. raidal , arxiv:1106.4553 ; h. wang _ et al .
_ , arxiv:1107.5769 ; g. z. krnjaic , arxiv:1109.0648 ; h. davoudiasl , t. mcelmurry and a. soni , arxiv:1108.1173 ; u. haisch , s. westhoff , jhep * 1108 * , 088 ( 2011 ) ; g. m. tavares and m. schmaltz , arxiv:1107.0978 [ hep - ph ] ; s. jung , h. murayama , a. pierce and j. d. wells , phys . rev .
d * 81 * , 015004 ( 2010 ) .
s. jung , a. pierce and j. d. wells , phys .
d * 83 * , 114039 ( 2011 ) . v. barger _
et al . _ , 81 , 113009 ( 2010 ) ; phys . lett .
b * 698 * , 243 ( 2011 ) ; j. cao _
et al . _ , 81 , 014016 ( 2010 ) ; i. dorsner _
et al . _ , 81 , 055009 ( 2010 ) ; b. xiao _ et al .
_ , 82 , 034026 ( 2010 ) ; b. bhattacherjee _ et al .
_ , phys .
d * 83 * , 091501 ( 2011 ) ; k. m. patel and p. sharma , jhep * 1104 * , 085 ( 2011 ) ; m. r. buckley , d. hooper , j. kopp , e. neil , phys .
* d83 * , 115013 ( 2011 ) ; g. isidori and j. f.
kamenik , phys .
b * 700 * , 145 ( 2011 ) ; e. r. barreto _ et al .
_ , phys .
d * 83 * , 054006 ( 2011 ) ; arxiv:1104.1497 ; a. rajaraman , z. e .
surujon , t. m. p. tait , arxiv:1104.0947 ; k. blum _ et al .
_ , arxiv:1107.4350 ; m. i. gresham _ et al .
_ , arxiv:1107.4364 ; y. cui _ et al .
_ , arxiv:1106.3086 ; m. duraisamy , a. rashed , a. datta , arxiv:1106.5982 ; b. grinstein , a. l. kagan , j. zupan and m. trott , arxiv:1108.4027 ; d. kahawala , d. krohn , m. j. strassler , arxiv:1108.3301 ; p. ko , y. omura and c. yu , arxiv:1108.4005 ; m. frank , a. hayreter and i. turan , arxiv:1108.0998 ; j. y. liu , y. tang and y. l. wu , arxiv:1108.5012 .
d. w. jung _ et al .
_ , 691 , 238 ( 2010 ) ; arxiv:1012.0102 ; c. zhang and s. willenbrock , arxiv:1008.3869 ; j. a. aguilar - saavedra , 843 , 638 ( 2011 ) ; nucl . phys .
b * 812 * , 181 ( 2009 ) ; c. degrande _ et al . _ , arxiv:1010.6304 ; k. blum _ et al .
_ , arxiv:1102.3133 ; c. delaunay _ et al . _ ,
arxiv:1103.2297 ; c. degrande _ et al .
_ , arxiv:1104.1798 ; j. a. aguilar - saavedra and m. perez - victoria , phys .
b * 701 * , 93 ( 2011 ) ; d. y. shao _ et al .
_ , arxiv:1107.4012
. for top quark reviews , see , e.g. , w. bernreuther , 35 , 083001 ( 2008 ) d. chakraborty , j. konigsberg , and d. rainwater , ann .
nucl . part .
* 53 * , 301 ( 2003 ) ; e. h. simmons , hep - ph/0211335 ; c .- p .
yuan , hep - ph/0203088 ; s. willenbrock , hep - ph/0211067 ; m. beneke _
et al . _ , hep - ph/0003033 ; t. han , arxiv:0804.3178 ; for model - independent new physics study , see , e.g. , c. t. hill and s. j. parke , 49 , 4454 ( 1994 ) ; k. whisnant _ et al .
_ , 56 , 467 ( 1997 ) ; j. m. yang and b .-
young , 56 , 5907 ( 1997 ) ; k. hikasa _
_ , 58 , 114003 ( 1998 ) ; r.a .
_ , arxiv:0811.1743 . http://cdsweb.cern.ch/record/1369186/files/atlas-conf-2011-095 ; cms collaboration , arxiv:1107.4771 [ hep - ex ] .
s. k. gupta , arxiv:1011.4960 [ hep - ph ] ; j. cao _ et al .
_ , arxiv:1101.4456 ; e. l. berger _
et al . _ ,
lett . * 106 * , 201801 ( 2011 ) ; arxiv:1109.3202 ; e. l. berger , arxiv:1109.3202 [ hep - ph ] . t. aaltonen _ et al . _
[ cdf collaboration ] , arxiv:1108.0101 [ hep - ex ] ; s. chatrchyan _ et al . _ [ cms collaboration],jhep 1108 , 005 ( 2011 ) ; http://cdsweb.cern.ch/record/1385032/files/atlas-conf-2011-139 .
j. l. hewett _ et al .
_ , arxiv:1103.4618 ; j. a. aguilar - saavedra and m. perez - victoria , arxiv:1105.4606 [ hep - ph ] ; arxiv:1107.0841 [ hep - ph ] ; arxiv:1107.2120 [ hep - ph ] ; j. a. aguilar - saavedra , a. juste and f. rubbo , arxiv:1109.3710 [ hep - ph ] ; j. f. arguin , m. freytsis and z. ligeti , arxiv:1107.4090 [ hep - ph ] . c. kao , 348 , 155 ( 1995 ) ; c. kao , g. a. ladinsky and c. p. yuan , int . j. mod .
a * 12 * , 1341 ( 1997 ) . c. kao and d. wackeroth , 61 , 055009 ( 2000 ) . c. s. li _ et al .
_ , 398 , 298 ( 1997 ) ; k. hikasa _ et al .
_ , 60 , 114041 ( 1999 ) ; p. y. li _ et al .
_ , 51,163 ( 2007 ) ; s. gopalakrishna _ et al .
_ , phys .
d * 82 * , 115020 ( 2010 ) ; n. liu and l. wu , commun .
phys . * 55 * , 296 ( 2011 ) .
r. m. godbole _ et al .
_ , jhep * 1011 * , 144 ( 2010 ) ; 84 , 014023 ( 2011 ) ; j. cao , l. wu , j. m. yang , 83 , 034024 ( 2011 ) ; d. w. jung _ et al .
_ , arxiv:1011.5976 ; e. l. berger _
_ , phys .
d * 83 * , 114026 ( 2011 ) d. krohn _ et al .
_ , arxiv:1105.3743 ; v. barger , w. y. keung and c. t. yu , arxiv:1108.2275 [ hep - ph ] . j. a. aguilar - saavedra and m. perez - victoria , jhep * 1105 * , 034 ( 2011 ) .
n. craig _ et al . _ , arxiv:1103.2127 ; m. t. bowen _ et al .
_ , phys .
d * 72 * , 074016 ( 2005 ) ; m. t. bowen , phys .
d * 73 * , 097501 ( 2006 ) ; s. heim _ et al .
_ , phys .
d * 81 * , 034005 ( 2010 ) ; f. penunuri , f. larios , a. o. bouzas , phys .
rev . * d83 * , 077501 ( 2011 ) ; s. jung _ et al .
_ , arxiv:1108.1802 ; c. h. chen _ et al . _ ,
e. l. berger _
_ , arxiv:1108.3613 . j. shelton and k. m. zurek , phys .
d * 83 * , 091701 ( 2011 ) h. tanaka and i. watanabe , int .
j. mod .
phys . a * 7 * , 2679 ( 1992 ) ; s. atag _ et al .
_ , phys .
d * 59 * , 015008 ( 1999 ) ; t. plehn , phys .
b * 488 * , 359 ( 2000 ) ; e. arik _ et al .
_ , jhep * 0209 * , 024 ( 2002 ) ; o. cakir and m. sahin , phys . rev .
d * 72 * , 115011 ( 2005 ) ; r. n. mohapatra _ et al .
_ , phys .
d * 77 * , 011701 ( 2008 ) ; c. r. chen _ et al .
_ , phys .
d * 79 * , 054002 ( 2009 ) ; e. del nobile _
et al . _ ,
b * 826 * , 217 ( 2010 ) ; e. l. berger _
et al . _ ,
lett . * 105 * , 181802 ( 2010 ) ; h. zhang _
et al . _ ,
b * 696 * , 68 ( 2011 ) ; t. han _ et al .
_ , jhep * 1001 * , 123 ( 2010 ) ; j. pumplin _ et al . _ ,
jhep * 0602 * , 032 ( 2006 ) . t. aaltonen
_ et al . _ [ the cdf collaboration ] , phys .
d * 82 * , 052002 ( 2010 ) .
cms collaboration , note cms - pas - top-11 - 001 ; cms - pas - top-10 - 007 ; g. aad
_ et al . _ [ the atlas collaboration ] , note atlas - conf-2011 - 070 n. kidonakis , phys .
d * 82 * , 114030 ( 2010 ) .
n. kidonakis and r. vogt , 78 , 074005 ( 2008 ) ; m. cacciari _ et .
_ , jhep * 0809 * , 127 ( 2008 ) ; s. moch and p. uwer , 78 , 034003 ( 2008 ) . f. maltoni and t. stelzer , jhep * 0302 * , 027 ( 2003 ) ; j. alwall
_ et al . _ ,
jhep * 0709 * , 028 ( 2007 ) ; j. alwall , m. herquet , f. maltoni , o. mattelaer and t. stelzer , jhep * 1106 * , 128 ( 2011 ) . | as a top quark factory , the lhc can test new physics models used to explain the top quark forward - backward asymmetry @xmath0 measured at the tevatron . in this work
we perform a comparative study for two such models : the @xmath1 model and the color triplet diquark ( @xmath2 ) model . requiring these models to explain @xmath0 and
also satisfy the top pair production rate measured at the tevatron , we examine their contributions to the lhc observables such as the polarizations and charge asymmetries in top quark productions and the charge asymmetry in @xmath1 ( or @xmath2 ) pair production .
we find that these observables can be enhanced to their observable levels and current lhc measurement on the top charge asymmetry has already tightly constrained the @xmath1 model .
we also find that each observable shows different characteristics in different models , which can be utilized to discriminate the models . |
quantum optomechanics has recently witnessed a rapid succession of key advances , with important milestones including the cooling of several systems to the quantum ground state of vibration @xcite , the demonstration of strong optomechanical coupling @xcite and of quantum coherent coupling between optical and phonon fields @xcite , the realization of optomechanical systems from ultracold atomic ensembles @xcite , and impressive advances toward quantum state transfer @xcite , to mention just a few examples .
these developments hint at the promise for optomechanics applications in quantum metrology @xcite , quantum information science @xcite , and also in possible tests of fundamental physics questions @xcite .
so far , though , the focus of most advances has been on single mode dynamics of the oscillator @xcite , and only few authors have started to investigate the effects of multiple mechanical modes @xcite . in complete analogy with the situation in quantum optics , however , single - mode quantum acoustics fails to capture many of its most important aspects , including most obviously perhaps propagation phenomena , nonlinear acoustical effects , focusing and defocusing of acoustic waves , etc .
this paper represents a step toward the development of multimode quantum acoustics .
the specific situation we consider is a multimode mechanical field perhaps provided by several mechanical oscillators or atomic samples trapped near extrema of an optical lattice , the mode coupling being provided by a common optical field . in some ways , this can be thought of as describing a situation analogous to collisions in atomic systems .
very much like collisions are effective interactions between atoms resulting from the elimination of the electromagnetic field , the optically mediated effective coupling between phononic modes can also be thought of in these terms .
generally speaking , collisions tend to be largely incoherent processes resulting e.g. in decoherence and dephasing in atomic ensembles , but it is known from ultracold atomic systems that this needs not be the case . in ultracold systems , such as quantum degenerate atomic gases , collisions can be a coherent process , and have been demonstrated to lead to nonlinear atom optics effects @xcite such as e.g matter - wave four - wave mixing , soliton generation , the creation of entangled atom pairs , and more . in analogy to that situation , one can expect similar effects to occur in quantum acoustics .
our main goal is to develop a simple theory that shows that this is indeed the case , and that incoherent effects due to quantum fluctuations associated primarily with the elimination of the electromagnetic field need not overwhelm coherent effects .
this paves the way to investigate nonlinear effects in phonon physics reminiscent of those encountered in nonlinear optics and nonlinear atom optics ; with the promise of similarly exciting applications .
our starting point is a relatively general discussion of optically mediated effective interactions between many mechanical modes , discussing in some detail the quantum noise and damping associated with that interaction , both for linear and quadratic optomechanical coupling .
we identify a number of specific forms of the effective interaction that lead to qualitatively different dynamics , in particular a regime where the coupling of the quantum fluctuations of the mechanical modes is intrinsically nonlinear .
we then illustrate the general analysis with an example involving just two mechanical modes , with emphasis on two limiting cases , the resolved sideband and the doppler regimes .
a specific example is given by the state transfer between two mechanical modes , for example a bose - einstein condensate coupled to the center - of - mass mode of a moving end mirror of a fabry - prot resonator through a cavity field .
we comment on the optimization of the transfer fidelity using squeezed input light as the optically mediating light field and conclude with some general remarks and outlook .
we consider a multimode optomechanical system described in terms of a set of @xmath0 modes of effective masses @xmath1 , bare frequencies @xmath2 and damping rates @xmath3 . the interaction between these modes is assumed to be mediated by a single mode of the optical resonator , of nominal frequency @xmath4 and driven by a monochromatic external field of frequency @xmath5 .
a realization of this system could be a high - finesse fabry - prot resonator with a series of internal non - absorptive dielectric membranes such as depicted in fig .
[ fig : system ] , a collection of ultracold atomic samples trapped at or near the minima of an optical lattice inside such a resonator , one or several bose - einstein condensates trapped inside a fabry - prot cavity , perhaps with a moving end - mirror , or a number of similar setups operating either in the optical or in the microwave regime .
the hamiltonian governing the system is @xmath6 where @xmath7 describes the cavity field mode , pumped externally by a field of frequency @xmath5 at a rate @xmath8 , @xmath9 describes the mechanical modes , with @xmath10 = \delta_{jk}$ ] .
the optomechanical interaction is @xmath11 in the linear case and @xmath12 in the quadratic case , with @xmath13 and @xmath14 linear and quadratic single - photon optomechanical coupling coefficients , respectively .
finally , @xmath15 describes the interaction of the cavity field and the mechanics with their respective reservoirs and accounts for dissipation . for simplicity
, we will assume that all mechanical modes either couple purely linearly or quadratically .
this section discusses the case of linear coupling including modifications to cold damping and optical spring effects .
with @xmath16 and for the optomechanical coupling ( [ h linear ] ) the heisenberg - langevin equations of motion in the standard input - output formalism @xcite are @xmath17 \hat{a}+\eta + i\hat{a}\sum_{j=1}^{\cal n } g_{0,j}(\hat{b}_{j}+\hat{b}_{j}^{\dag } ) + \sqrt{\kappa } \hat { a}_{\rm in } , \nonumber \\ \dot{\hat{b}}_{j } & = & -i\omega_{j}\hat{b}_{j } -\frac{\gamma_{j}}{2}(\hat{b}_{j}-\hat{b}_{j}^{\dag } ) + ig_{0,j}\hat{a}^{\dag}\hat{a } + i\tilde { \xi}_j , \label{hl eq}\end{aligned}\ ] ] where @xmath18 @xmath19 and @xmath3 are the decay rates of the cavity field and mechanical modes , respectively , and @xmath20 and @xmath21 the associated markovian quantum noise operators .
their two - time correlation functions will be given explicitly later on .
if the system is driven by a classical field , it is useful to decompose the various field operators as the sum of their expectation values and small quantum fluctuations , whose dynamics are treated to lowest order only .
specifically we expand the operators @xmath22 and @xmath23 as [ mean field ] @xmath24 where @xmath25=1 $ ] , @xmath26=\delta_{jk}$ ] and we assume that their expectation values @xmath27 and @xmath28 are much larger than the amplitudes of the fluctuations , for example , @xmath29 .
( note that rigorously speaking this assumes all quantum contributions to have zero expectation value .
however , the treatment remains valid for fluctuations with non - zero , but small expectation values @xcite . )
this yields the mean field equations of motion @xmath30 with steady - state values @xmath31 where @xmath32 here we have chosen the phase of @xmath8 such that @xmath33 is real without loss of generality for the case of cw pumping .
as is well known , dynamical back - action gives rise to frequency shifts the optical spring effect and radiative damping or amplification for each mechanical oscillator @xcite . in order to capture these effects explicitly , we switch to an interaction picture with the transformations @xmath34 where @xmath35 and the expressions for the dynamical shifts @xmath36 will be determined self - consistently later on @xcite . the equations of motion for the fluctuation operators then read @xmath37\hat{a } + i\sum_{j=1}^{\cal n}g_{j}(\hat{b}_{j}e^{-i\nu_jt}+\hat{b}_{j}^{\dag}e^{i\nu_jt } ) + \sqrt{\kappa}\hat{a}_{\rm in } , \label{deltadt}\ ] ] @xmath38\hat{b}_{j}+\frac{\gamma_{j}}{2}\hat{b}_{j}^{\dag}e^{2i\nu_jt } + ig_{j}e^{i\nu_jt}(\hat{a}^{\dag}+\hat{a } ) + i\hat{\xi}_j , \label{deltabdt}\ ] ] where we have introduced the amplified optomechanical coupling strengths , @xmath39 in the physically relevant regime where the optical loss is much larger than the mechanical decay rates , @xmath40 , the cavity field follows the dynamics of the mechanical oscillators adiabatically . for time
scales long compared to @xmath41 , the formal solution of eq .
( [ deltadt ] ) reads @xmath42 \nonumber \\ & & + \hat{f}_{\rm in}(t),\end{aligned}\ ] ] where the noise operator @xmath43 is given by @xmath44 substituting the formal solution into the equations of motion for the mechanical oscillators , eq .
( [ deltabdt ] ) , gives @xmath45\hat{b}_{j}+\frac{\gamma_{j}}{2}\hat{b}_{j}^{\dag}e^{2i\nu_jt } \nonumber \\ & - & \sum_{k=1}^{\cal n}g_{jk}\left[i\omega_{k}+\frac{\gamma_{k}}{2}\right]e^{i(\nu_j-\nu_k)t}\hat{b}_{k } \nonumber \\ & - & \sum_{k=1}^{\cal n}g_{jk}\left[i\omega_{k}-\frac{\gamma_{k}}{2}\right]e^{i(\nu_j+\nu_k)t}\hat{b}_{k}^{\dag } \nonumber \\ & + & ig_{j}e^{i\nu_jt}(\hat{f}_{\rm in}+\hat{f}_{\rm in}^{\dag } ) + i \hat{\xi}_j \label{eombeforerwa},\end{aligned}\ ] ] where @xmath46 and the radiation - induced frequency shifts @xmath47 and damping coefficients @xmath48 are given by [ shiftsdamping ] @xmath49 , \label{omegaj}\end{aligned}\ ] ] @xmath50.\label{gammaj}\end{aligned}\ ] ] due to the appearance of the shifted frequencies @xmath51 in the denominators on its right hand side , eq .
( [ omegaj ] ) is a fifth - order polynomial equation for @xmath52 , so that it is not lorentzian in general . in the strong coupling regime @xmath53 ,
it becomes multivalued for a range of effective detunings , a feature that can be interpreted in terms of normal mode splitting @xcite .
in this situation the mechanical and optical modes are strongly hybridized and it is not meaningful to treat them separately . also , the counter - rotating terms in eqs .
( [ eombeforerwa ] ) , combined in particular with a driving field blue detuned from the resonator resonance , can result in instabilities in the strong coupling regime and a concomitant would lead to the break - down of the linearization process .
the strong coupling and blue detuned regimes will be the object of future work . here
we focus on the simpler case of weak coupling , @xmath54 . in this limit
the frequency shifts @xmath47 remain small compared to the bare frequencies of the oscillators , so that we can approximate @xmath55 by @xmath56 on the right hand side of eqs .
( [ shiftsdamping ] ) , see eq .
( [ shifted freq ] ) .
the frequency shifts @xmath47 are then the sum of two dispersion curves while the decay rates @xmath48 consist of two lorentzians with opposite signs , all of which are centered around @xmath57 .
the weak coupling regime also allows us to drop the counter - rotating terms in eqs .
( [ eombeforerwa ] ) .
the resulting equations of motion are @xmath58\hat{b}_{j } -\sum_{k=1}^{\cal n}g_{jk}\left[i\omega_{k}+\frac{\gamma_{k}}{2}\right]e^{i(\nu_j-\nu_k)t}\hat{b}_{k } \nonumber \\ & + & ig_{j}e^{i\nu_jt}(\hat{f}_{\rm in}+\hat{f}_{\rm in}^{\dag } ) + i \hat{\xi}_j , \label{rwa } \label{weakcoupling linear}\end{aligned}\ ] ] these equations describe a system of @xmath59 pairwise , linearly coupled oscillators , with coherent contributions proportional to the optically induced frequency shifts @xmath47 , cold damping coefficients @xmath48 familiar from the single - mode situation , and associated noise operators @xmath60 and @xmath61 . the fact that each mechanical element experiences a different frequency shift and radiative decay , depending on its optomechanical coupling strength and the pump - cavity detuning , is the key ingredient that allows one to couple oscillators with different bare frequencies on - demand , by choosing appropriate pump - powers and detunings .
we discuss selected consequences of this coupled dynamics in section v for the case of a two - mode system . but
first , the next section extends our discussion to the case of quadratic optomechanical coupling .
we now consider the quadratic optomechanical coupling @xmath62 see eq .
( [ h quadratic ] ) , where @xmath14 are single - photon coupling coefficients . in that case
the heisenberg - langevin equations of motion are , in the frame of the laser frequency @xmath5 , @xmath63\hat{a}+\eta -i\hat{a}\sum_{j=1}^{\cal n}g_{0,j}^{(2)}(\hat{b}_j+\hat{b}_j^{\dag})^2 + \sqrt{\kappa}\hat{a}_{\rm in } , \nonumber \\ \dot{\hat{b}}_j & = -i\omega_j\hat{b}_j-\frac{\gamma_j}{2}(\hat{b}_j-\hat{b}_j^{\dag } ) -2ig_{0,j}^{(2)}\hat{a}^{\dag}\hat{a}(\hat{b}_j+\hat{b}_j^{\dag } ) + i\tilde{\xi}_j \nonumber\end{aligned}\ ] ] with @xmath64 as before . with the expansion ( [ mean field ] ) we have @xmath65 where the effective detuning is now @xmath66 and consistently with the linearization procedure we have neglected in the second line the vacuum expectation value contribution @xmath67 .
the steady - state expectation value of the cavity field amplitude is @xmath68 and the steady - state displacements of the mechanical oscillators are found from the @xmath59 coupled fifth - order equations @xmath69 zero displacement @xmath70 is always a solution to these equations , and for @xmath71 , which is the case for membranes located at minima of the intracavity field , it is the only stable solution .
however that solution may become unstable for negative couplings , @xmath72 , occurring when oscillators are located at cavity - field maxima . in this case , each oscillator settles in one of the stable positions , @xmath73 located symmetrically around the intracavity field maximum . as we will see in the following sections , these two geometries result in qualitative consequences for the behavior of the system . to simplify the discussion
we assume that all coupling constants @xmath74 have the same sign , and consider first the case where they are negative .
the stable steady state is degenerate , as each oscillator can have a negative or positive mean displacement .
a particular choice of steady state will break the symmetry of the sign of the displacements and influence the physics of the system .
experimentally , it could be controlled by addressing the individual oscillators through an additional field or preparing them slightly displaced towards the desired steady - state position .
the heisenberg - langevin equations of motion for the fluctuations can be approximated to lowest non - vanishing order , as @xmath75\hat{a}-4i\sum_{j=1}^{\cal n}g_j^{(2)}\bar{x}_j(\hat{b}_j+\hat{b}_j^{\dag } ) \nonumber \\ & + & \sqrt{\kappa}\hat{a}_{\rm in},\\ \dot{\hat{b}}_j & \simeq & -\left[i\varpi_j+\frac{\gamma_j}{2}\right]\hat{b}_j -\left[2ig_{0,j}^{(2)}\bar{n}_{\rm c}-\frac{\gamma_j}{2}\right]\hat{b}_j^{\dag } \nonumber \\ & - & 4ig_j^{(2)}\bar{x}_j(\hat{a}+\hat{a}^{\dag})+i\tilde{\xi}_j , \end{aligned}\ ] ] where we have introduced the amplified quadratic optomechanical strengths @xmath76 and the shifted detuning of the cavity field and mechanical frequencies are @xmath77 note that in contrast with the situation for linear optomechanical coupling the correction @xmath78 to the detuning @xmath79 is now twice as large as the correction to the mean - filed dynamics , compare eq . ( [ cor delta ] ) to eq .
( [ shifted delta ] ) .
this second - harmonic generation process , a direct acoustic analog of optical second harmonic generation , is a direct consequence of the nonlinear nature of the quadratic optomechanical coupling and is driven by the mean - field mechanical oscillations . assuming that the resulting frequencies of the mechanical oscillators are @xmath80 and following the same approach as before the equations of motion for the mechanical modes become @xmath81\hat{b}_j -\left[2ig_{0,j}^{(2)}\bar{n}_{\rm c}-\frac{\gamma_j}{2}\right]\hat{b}_j^{\dag}e^{2i\nu_jt } \nonumber \\ & -&\sum_{k=1}^{\cal n}g_{jk}^{(2)}\left[i\omega_{k}+\frac{\gamma_{k}}{2}\right]e^{i(\nu_j-\nu_k)t}\hat{b}_{k } \nonumber \\ & -&\sum_{k=1}^{\cal n}g_{jk}^{(2)}\left[i\omega_{k}-\frac{\gamma_{k}}{2}\right]e^{i(\nu_j+\nu_k)t}\hat{b}_{k}^{\dag } \nonumber \\ & -&4ig_{j}^{(2)}\bar{x}_je^{i\nu_jt}(\hat{f}_{\rm in}+\hat{f}_{\rm in}^{\dag } ) + i\hat{\xi}_j , \label{localmaximabeforerwa}\end{aligned}\ ] ] where the noise operator associated with the adiabatically eliminated optical field is @xmath82 and @xmath83 we then find for the radiation - induced frequency shifts @xmath84 and damping coefficients @xmath85 @xmath86,\\ \frac{\gamma_{k}}{2 } & = \left(4g_{k}^{(2)}\bar{x}_k\right)^2 \nonumber\\ & \times \left[\frac{\kappa/2}{(\delta^{(2)}+\nu_k)^2+\kappa^2/4 } -\frac{\kappa/2}{(\delta^{(2)}-\nu_k)^2+\kappa^2/4 } \right].\end{aligned}\ ] ] in the weak coupling regime , @xmath87
, we can neglect the fast oscillating terms on the right - hand side of eq .
( [ localmaximabeforerwa ] ) so that the equations of motion for the mechanics can be further simplified to @xmath81\hat{b}_j -\sum_{k=1}^{\cal n}g_{jk}^{(2)}\left[i\omega_{k}+\frac{\gamma_{k}}{2}\right]e^{i(\nu_j-\nu_k)t}\hat{b}_{k } \nonumber \\ & & -4ig_{j}^{(2)}\bar{x}_je^{i\nu_jt}(\hat{f}_{\rm in}+\hat{f}_{\rm in}^{\dag } ) + i \hat{\xi}_j . \end{aligned}\ ] ] these equations have the same form as in the case of linear optomechanical coupling of the mechanical oscillators , the main new features in this regime are second - harmonic generation and the flexibility to choose the sign and strength of the coupling coefficients @xmath88 through the equilibrium positions @xmath89 and @xmath90 , see eq . (
[ quad coupling 1 ] ) .
the situation is changed qualitatively in the case where the mechanical oscillators are located at minima of the cavity field .
the single - photon , single mode quadratic optomechanical coupling coefficients are then positive and the cavity field results in a tighter confinement through the static optical spring effect . the only stable steady - state displacement is given by @xmath91 so that the effects described in the previous section are highly suppressed .
on the other hand , a number of novel features appear as a result of the fact that the first non - vanishing couplings are of higher order in the quantum fluctuations .
in particular , instead of the second - harmonic generation discussed in the previous section , the dominant nonlinear effects are akin to four - wave mixing in optics .
to lowest order in the fluctuations , the heisenberg - langevin equations of motion become @xmath92\hat{a } -i\sum_{j=1}^{\cal n}g_{j}^{(2)}(\hat{b}_j^2+\hat{b}_j^{\dag 2}+2\hat{b}_j^{\dag}\hat{b}_j ) \nonumber \\ & & + \sqrt{\kappa}\hat{a}_{\rm in},\\ \dot{\hat{b}}_j & \simeq & -\left[i\varpi_j+\frac{\gamma_j}{2}\right]\hat{b}_j -\left[2ig_{0,j}^{(2)}\bar{n}_{\rm c}-\frac{\gamma_j}{2}\right]\hat{b}_j^{\dag } \nonumber \\ & & -2ig_{j}^{(2)}(\hat{a}+\hat{a}^{\dag})(\hat{b}_j+\hat{b}_j^{\dag})+i\tilde{\xi}_j , \end{aligned}\ ] ] where the effective detuning @xmath93 is now @xmath94 applying the same arguments and approximations as before to adiabatically eliminate the optical field dynamics , we find the coupled equations of motion for the mechanical modes @xmath81\hat{b}_j -\left[2ig_{0,j}^{(2)}\bar{n}_{\rm c}-\frac{\gamma_j}{2}\right]\hat{b}_j^{\dag}e^{2i\nu_jt } -\frac{1}{2}\sum_{k=1}^{\cal n}g_{jk}^{(2 ) } \left[i\omega_k+\frac{\gamma_k}{2}\right](e^{-2i\nu_kt}\hat{b}_k^{2}\hat{b}_j + e^{2i(\nu_j-\nu_k)t}\hat{b}_k^{2}\hat{b}_j^{\dag } ) \nonumber \\ & -&\frac{1}{2}\sum_{k=1}^{\cal n}g_{jk}^{(2 ) } \left[i\omega_k-\frac{\gamma_k}{2}\right](e^{2i\nu_kt}\hat{b}_k^{\dag2}\hat{b}_j + e^{2i(\nu_j+\nu_k)t}\hat{b}_k^{\dag2}\hat{b}_j^{\dag } ) -\frac{i}{2}\sum_{k=1}^{\cal n}g_{jk}^{(2)}\lambda_k(\hat{b}_k^{\dag}\hat{b}_k\hat{b}_j + \hat{b}_k^{\dag}\hat{b}_k\hat{b}_j^{\dag}e^{2i\nu_jt } ) \nonumber \\ & -&2ig_{j}^{(2)}(\hat{f}_{\rm in}+\hat{f}_{\rm in}^{\dag})(\hat{b}_j+\hat{b}_j^{\dag}e^{2i\nu_jt } ) + i\hat{\xi}_j , \label{bdot minima}\end{aligned}\ ] ] where the mode - coupling is now given by @xmath95 the frequency shifts and radiative damping coefficients are @xmath96 , \nonumber \\ \frac{\gamma_{k}}{2 } & = & ( 2g_{k}^{(2 ) } ) ^2\\ & \times & \left[\frac{\kappa/2}{(\bar{\delta}^{(2)}+2\nu_k)^2+\kappa^2/4 } -\frac{\kappa/2}{(\bar{\delta}^{(2)}-2\nu_k)^2+\kappa^2/4 } \right],\nonumber\end{aligned}\ ] ] @xmath97 and @xmath98 note that in eq .
( [ bdot minima ] ) the quantum noise stemming from the elimination of the optical field is now multiplicative . since we can neglect fast rotating terms in the weak coupling regime , eqs
( [ bdot minima ] ) further simplify to @xmath81\hat{b}_j -\frac{i}{2}\sum_{k=1}^{\cal n}g_{0,jk}^{(2)}\lambda_k\hat{b}_k^{\dag}\hat{b}_k\hat{b}_j\nonumber \\ & -&\frac{1}{2}\sum_{k=1}^{\cal n}g_{0,jk}^{(2 ) } \left[i\omega_k+\frac{\gamma_k}{2}\right]e^{2i(\nu_j-\nu_k)t}\hat{b}_k^{2}\hat{b}_j^{\dag } \nonumber \\ & -&2ig_{j}^{(2)}(\hat{f}_{\rm in}+\hat{f}_{\rm in}^{\dag})(\hat{b}_j+\hat{b}_j^{\dag}e^{2i\nu_jt } ) + i\hat{\xi}_j . \label{localminimarwa}\end{aligned}\ ] ] these coupled nonlinear equations are indicative of four - wave mixing processes , and are formally similar to situations driven by @xmath99 nonlinear susceptibilities in nonlinear classical and quantum optics @xcite . in that context , four - wave mixing phenomena are known to lead to a rich spectrum of effects that have found numerous applications in applied and fundamental studies .
they include for example sum frequency generation , high harmonic generation , optical parametric amplification , self - focusing and defocusing , the generation and propagation of optical solitons , phase conjugation , the generation of quantum mechanical squeezing and entangled photon pairs , and more @xcite .
the analysis of this section paves the way to investigate a similar range of studies in quantum optomechanics and quantum acoustics , and as such , opens the way to an intriguing and rich direction of investigation that will be carried out in follow - up studies .
one particularly attractive feature of optomechanics in this context is the ease with which the sign and strength of the nonlinear interactions can be adjusted in combination with the functionalization of these devices .
the following sections discuss a few illustrative examples , concentrating for now on the simple situation of a two - mode system with linear optomechanical coupling only , and demonstrating a example of functionalization that involves the optically mediated coupling of a nanomechanical system and a quantum degenerate atomic system .
elementary multimode effects can be studied in the case where two mechanical modes interact linearly with a common cavity field , i.e. @xmath100 . in the weak coupling regime
the heisenberg - langevin eqs .
( [ weakcoupling linear ] ) become @xmath101\hat{b}_2 e^{i(\nu_1-\nu_2)t } \nonumber \\ & + & ig_2e^{i\nu_1 t}(\hat{f}_{\rm in}+\hat{f}_{\rm in}^{\dag } ) + i \hat{\xi}_1\nonumber \\ \dot{\hat{b}}_2 & = & -\frac12 \gamma_{e,2}\hat{b}_2 -\left[i\omega_{c,2}+\frac12\gamma_{c,2}\right]\hat{b}_1e^{-i(\nu_1-\nu_2)t } \nonumber \\ & + & ig_2e^{i\nu_2t}(\hat{f}_{\rm in}+\hat{f}_{\rm in}^{\dag } ) + i \hat{\xi}_2 \label{2modes}\end{aligned}\ ] ] where we have introduced the effective decay rate @xmath102 the cross - damping rate @xmath103 and the cross - coupling frequency @xmath104 the cross - damping rate is associated with effects such as sympathetic cooling or heating of coupled oscillators , while the cross - coupling frequency is associated with the coherent aspects of multimode coupling , such as e.g. quantum state transfer . note that when the two effective mechanical frequencies are matched , @xmath105 , the cross - coupling frequencies and damping coefficients become equal @xmath106 . in terms of the motional quadratures @xmath107 eqs .
( [ 2modes ] ) become @xmath108 where @xmath109 , the drift matrix @xmath110 is given by @xmath111 and the noise operator matrix @xmath112 is @xmath113 with the effective optical noise operator @xmath114 in the following we limit ourselves to the case of gaussian states and assume that the mechanical elements are coupled to statistically independent markovian heat baths at low temperature @xmath115 characterized by the two - time correlations functions @xcite @xmath116 where @xmath117^{-1}$ ] . from eqs .
( [ mean field ] ) the first moments of all position and momentum quadratures are always zero , and one readily derives a closed set of differential equations for the second moments of the oscillator quadratures .
the equation of motion for the covariance matrix @xmath118 can be written as @xcite @xmath119 where @xmath120 is the drift matrix , and the inhomogeneous term @xmath121 is given by @xmath122 .
this system of equations can be solved exactly in principle but the resulting solutions lack transparency . more physical insight into the coupled dynamics of the oscillator can be gained in the resolved sideband limit and in the doppler limit , two cases to which we now turn .
the resolved sideband regime is characterized by slow optical resonator damping compared to the mechanical frequencies @xmath123 .
( we assume without loss of generality that @xmath124 in the following . )
it is known from the standard theory of single - mode cavity optomechanics that in the absence of coupling each oscillator experiences a frequency shift @xmath52 comprised of the sum of two dispersion curves centered around @xmath125 with characteristic widths @xmath19 , see fig . [
fig : frequency_matching ] , while their damping coefficients @xmath126 are the sum of a positive and negative lorentzian , also of width @xmath19 , and centered at these same detunings @xcite .
the red and blue detunings for which the two oscillators are brought into resonance with each other are given by the intersections of the two curves in fig .
[ fig : frequency_matching ] .
there are two such intersections on the red - detuned side of the optical resonance , one near @xmath127 and one on the wing of the dispersion curve .
two more intersections are located similarly on the blue - detuned side . at these points
we have @xmath128 and therefore @xmath129 and @xmath130 from eq .
( [ cross coupling ] ) .
it is for those detunings that the coupling between the two mechanical oscillators is most effective .
the detailed dynamics of resonant mode coupling depends on whether the resonance condition @xmath131 occurs near the center or on the wing of the dispersion curves that comprise @xmath52 . near the resonance
@xmath132 optical damping ( or amplification ) dominates over the optical spring effect , see eqs .
( [ omegaj ] ) and ( [ gammaj ] ) , and to first approximation one can safely neglect that effect compared to those of @xmath133 by setting @xmath134 , see eqs .
( [ cross damping ] ) and ( [ cross coupling ] ) .
the drift matrix @xmath135 reduces to @xmath136 and the dominant consequence of the effective optomechanical coupling between the oscillators is the modification of their rate of cold damping or amplification compared to their uncoupled values .
one such example is shown in fig .
[ fig : heating ] , which plots the evolution of the variances of the motional quadratures of the mechanical modes when their effective frequencies are matched around the center of the lorentzian curve on the red side of the cavity field .
the mechanical modes are both assumed to be initially in thermal equilibrium . in the absence of cross interaction , both oscillators are subject to cold damping ( dotted and dot - dashed lines ) with cooling rates that differ since @xmath137 .
the coupling changes the situation significantly , with both cooling rates now markedly slower ( solid and dashed lines ) .
this is a result of the exchange of thermal excitations between the two oscillators that inhibits their direct optical cooling .
( note that our numerical calculations suggest that the long - time limit and thus the cooling limit does not change . ) performing a normal - mode analysis on the two modes reveals that the two out - of - phase oscillations decouple from the cavity field , while the two in - phase modes are cooled with an increased damping rate .
we now turn to the resonances @xmath131 that occur on the wings of the dispersion curves . in these cases
the optical spring effect dominates over cold damping ( or amplification ) , a consequence of the faster decay of the lorentzian @xmath126 ( with @xmath138 ) compared to the dispersive nature of @xmath52 , which decreases as @xmath139 .
that situation is therefore characterized by the dominance of coherent exchange between the two oscillators , and we approximate the corresponding drift matrix by @xmath140 in this regime , and provided that the noise sources associated with the optical field can be managed , one can realize coherent effects such as quantum state transfer between the two mechanical modes .
we return to this effect in the doppler regime , which we consider in the following . in the doppler regime , @xmath141 ,
the width of the optical cavity is so large that it washes out the two distinct features of @xmath52 and @xmath126 , characteristic of the resolved sideband regime , and one can approximate @xmath142 the qualitative behavior of the effective frequencies @xmath143 is plotted in fig .
[ fig : doppler_frequency_matching ] for @xmath144 and shows the intersections of the two curves on the red - detuned side of the cavity resonance .
( they would lie on the blue - detuned side for @xmath145 ) . since @xmath146 decays with the third power of the cavity linewidth
while @xmath147 scales as @xmath148 , it follows that for the large cavity damping rates , characteristic of the doppler regime , we can neglect in first approximation the effects of cold damping ( or amplification ) and at resonance we find again @xmath149 note that the argument leading to this result is different in the resolved sideband limit , where the radiation induced damping can be neglected if the effective detuning is outside of the mechanical sideband , and in the doppler regime , in which case it is negligible due to the large cavity linewidth .
the previous discussion showed that both , the resolved sideband and the doppler regimes , provide situations where cold damping can be neglected and the two oscillators are resonantly coupled by the effective coherent interaction @xmath150 this is the familiar beam splitter hamiltonian , which is known in particular to lead to quantum state swapping between the two oscillators .
however , an important issue is that the optical field mediating the interaction also results in the appearance of quantum noise , further limiting the fidelity of state transfer in addition to the clamping noise associated with the mechanical coupling to thermal reservoirs . clamping noise can be reduced in principle in cryogenic environments or perhaps in levitated structures @xcite .
what we show in this brief section is that the fundamental quantum noise associated with the optical field can also be reduced in principle by the use of squeezed input fields with correlation functions @xcite @xmath151 where @xmath5 is the central frequency of a squeezing device , and positive - valued @xmath152 and complex - valued @xmath153 are the squeezing parameters which define the strength of the squeezing as well as its direction in phase space , given by the complex phase of @xmath153 . in the following ,
we assume an ideal squeezed state which satisfies @xmath154 the equations of motion for vacuum input noise are recovered by substituting @xmath155 .
for concreteness we consider the explicit situation of a hybrid two - mode mechanical system where one of the modes is a recoil - induced side - mode of a bose - einstein condensate @xcite trapped inside an optical resonator with an oscillating end - mirror , and the other is the center - of - mass mode of oscillation of that mirror . as discussed in ref .
@xcite , in the case where a system is prepared in such a way that the shifted optomechanical frequencies of the two oscillators @xmath156 and @xmath157 are equal , @xmath158 , the equations of motion for the quadratures of the coupled mechanical oscillators reduce to @xmath159 where the subscripts 1 and 2 refer to the oscillating mirror and the condensate side - mode respectively , and the cross - coupling frequency is @xmath160 for a squeezed input field the two - time correlation function of the effective noise can be approximated as @xmath161+n+\frac{1}{2}\right)\delta(t - t'),\end{aligned}\ ] ] with @xmath162 .
here we have assumed that intrinsic mechanical decoherence is negligible for the time scales of interest .
figure [ fig : thermal_bec ] shows the co - evolution of the variance of one of the motional quadratures of the two oscillators . in this example the resonator is driven by a coherent field and the initial states of the bec and the moving mirror are the ground state and a thermal state with mean phonon number @xmath163 , respectively .
as expected , the fidelity of the coherent quantum state transfer , which would be unity at times @xmath164 , with @xmath165 integer , in the absence of noise , is increasingly reduced by the quantum fluctuations of the light field , even without intrinsic mechanical decoherence mechanisms . the use of a squeezed optical field to reduce the detrimental effect of quantum noise
is illustrated in fig .
[ fig : squeezed_bec ] , where the improved fidelity is readily apparent in particular at time @xmath166 . as illustrated in figs .
[ fig : fidelity_bec ] and [ fig : fidelity_bec_squeezed ] , the state transfer fidelity depends strongly on the phase difference between the cavity field and a squeezing device . for @xmath167 , the fluctuations in the position quadrature of the cavity field are minimized so that very high fidelity state transfer is achieved , @xmath168 for @xmath169 in our example .
as would be intuitively expected , we also find that strong squeezing leads to an increased sensitivity of the state transfer fidelity on that relative phase . as a comparison to fig .
[ fig : fidelity_bec ] , which is for a bec side - mode initially in the ground state and the oscillating mirror in a thermal state with @xmath170 , fig .
[ fig : fidelity_bec_squeezed ] displays the fidelity of state transfer where the bec is initially in a squeezed state with @xmath171 and @xmath172 . the transfer fidelity in that case exhibits an increased dependence of the relative phase between the cavity field and a squeezing device , as would be expected . in the optimal case @xmath167 and @xmath169 ,
the fidelity of squeezed state transfer is @xmath173 , which is somewhat lower than for a vacuum state , a consequence of the higher sensitivity of squeezed states to decoherence .
in summary , we have analyzed the optically mediated interaction between multimode phonon fields in cavity optomechanical systems where it is possible to adiabatically eliminate the electromagnetic field .
a number of possible forms of this interaction were identified , including multimode beam - splitter interactions and , in the case of quadratic optomechanical coupling , a plethora of possible nonlinear effective interactions , such as analogs of second - harmonic generation and four - wave mixing in nonlinear optics .
the specific forms of the effective interaction depend sensitively on the sign of the couplings .
importantly , we found that incoherent effects due to quantum fluctuations associated primarily with the elimination of the electromagnetic field need not overwhelm coherent mode - coupling effects . as a result
one can expect that analogs of nonlinear effects such as four - wave mixing , soliton generation , the creation of entangled phonon pairs , and others should be achievable in quantum acoustics .
our treatment was kept sufficiently generic that it can be applied to a wide variety of systems and parameter regimes , and as a concrete example we have discussed quantum state transfer between a momentum side - mode of a bose - einstein condensate and the center - of - mass mode of oscillation of the end mirror of a fabry - prot interferometer .
we showed how decoherence effects can be reduced significantly in that case by the use of squeezed optical fields .
future work will expand these results to more general situations and apply it to the study of practical situations where the effective mode coupling is nonlinear and investigate particular effects such as phase conjugation , entanglement , and the creation of entangled phonon pairs , for example .
we will also exploit pulsed optomechanics ideas to further expand the toolbox of quantum optomechanics . | we consider theoretically the optomechanical interaction of several mechanical modes with a single quantized cavity field mode for linear and quadratic coupling .
we focus specifically on situations where the optical dissipation is the dominant source of damping , in which case the optical field can be adiabatically eliminated , resulting in effective multimode interactions between the mechanical modes . in the case of linear coupling , the coherent contribution to the interaction can be exploited e.g. in quantum state swapping protocols , while the incoherent part leads to significant modifications of cold damping or amplification from the single - mode situation .
quadratic coupling can result in a wealth of possible effective interactions including the analogs of second - harmonic generation and four - wave mixing in nonlinear optics , with specific forms depending sensitively on the sign of the coupling .
the cavity - mediated mechanical interaction of two modes is investigated in two limiting cases , the resolved sideband and the doppler regime . as an illustrative application of the formal analysis we discuss in some detail a two - mode system where a bose - einstein condensate is optomechanically linearly coupled to the moving end mirror of a fabry - prot cavity . |
one of us , l.r . , acknowledges support from the ministerio de educacin y ciencia . we would like to thank t. nakano for useful discussions .
this work is partly supported by the spanish csic and jsps collaboration , the dgicyt contract number bfm2003 - 00856 , and the e.u .
euridice network contract no .
hprn - ct-2002 - 00311 .
n. kaiser , p. b. siegel and w. weise , nucl .
a * 594 * ( 1995 ) 325 [ arxiv : nucl - th/9505043 ] .
n. kaiser , p. b. siegel and w. weise , phys .
b * 362 * ( 1995 ) 23 [ arxiv : nucl - th/9507036 ] .
e. oset and a. ramos , nucl .
a * 635 * ( 1998 ) 99 [ arxiv : nucl - th/9711022 ] .
u. g. meissner and j. a. oller , nucl .
phys . a * 673 * ( 2000 ) 311 [ arxiv : nucl - th/9912026 ] .
u. g. meissner and j. a. oller , nucl .
a * 673 * ( 2000 ) 311 [ arxiv : nucl - th/9912026 ] .
e. oset , a. ramos and c. bennhold , phys .
b * 527 * ( 2002 ) 99 [ erratum - ibid .
b * 530 * ( 2002 ) 260 ] [ arxiv : nucl - th/0109006 ] .
a. ramos , e. oset and c. bennhold , phys .
* 89 * ( 2002 ) 252001 [ arxiv : nucl - th/0204044 ] .
d. jido , j. a. oller , e. oset , a. ramos and u. g. meissner , nucl .
a * 725 * ( 2003 ) 181 [ arxiv : nucl - th/0303062 ] .
t. hyodo , s. i. nam , d. jido and a. hosaka , phys .
c * 68 * ( 2003 ) 018201 [ arxiv : nucl - th/0212026 ] .
e. e. kolomeitsev and m. f. m. lutz , phys .
b * 585 * ( 2004 ) 243 [ arxiv : nucl - th/0305101 ] . c. garcia - recio , m. f. m. lutz and j. nieves , phys .
b * 582 * ( 2004 ) 49 [ arxiv : nucl - th/0305100 ] .
s. eidelman _ et al . _
[ particle data group collaboration ] , phys .
b * 592 * ( 2004 ) 1 .
s. sarkar , e. oset and m. j. vicente vacas , arxiv : nucl - th/0407025 . s. sarkar , e. oset and m. j. vicente vacas , arxiv : nucl - th/0404023 . t. nakano _ et al . _
[ leps collaboration ] , phys .
* 91 * ( 2003 ) 012002 [ arxiv : hep - ex/0301020 ] .
d. p. barber _ et al .
_ , z. phys .
c * 7 * ( 1980 ) 17 .
s. b. chan , j. button - shafer , s. s. hertzbach , r. r. kofler and m. schiff , phys .
* 28 * ( 1972 ) 256 .
j. c. nacher , e. oset , h. toki and a. ramos , phys .
b * 455 * ( 1999 ) 55 [ arxiv : nucl - th/9812055 ] . | recent studies within the framework of chiral unitary theory produce the @xmath0 resonance , among several others , as a dynamically generated resonance from the interaction of the baryon decuplet and the meson octet .
the @xmath0 qualifies as a quasibound state of @xmath1 and this has repercussion in some observable quantities . in the present work
we show that the @xmath2 reaction has a sizeable cross section for invariant @xmath3 masses above the @xmath4 mass . on the other hand
, we also find that the @xmath5 reaction has a sizeable cross section in that energy region as a consequence of the strong coupling of the @xmath0 to @xmath6 , and then we make predictions for the ratio of this cross section to that of the @xmath7 reaction .
+ _ @xmath8 departamento de fsica terica and ific centro mixto universidad de valencia - csic + institutos de investigacin de paterna , apdo .
correos 22085 , 46071 , valencia , spain _ the introduction of unitary techniques in the study of meson baryon interaction with chiral lagrangians has allowed to show that many of the low lying baryonic resonances are dynamically generated in the implicit process of multiple scattering , or equivalently , they qualify as quasibound meson baryon states .
early studies in this direction pointed at the @xmath9 @xcite and the @xmath10 @xcite as dynamically generated resonances .
the unitary methods to deal with the meson baryon interaction within the chiral framework have been made more systematic and a variety of them are now available , leading to remarkably similar results @xcite .
one of the consequences of these studies is that the interaction of the baryon octet and meson octet leads to two octets and one singlet of dynamically generated resonances with @xmath11 @xcite , to which the @xmath9 and the @xmath10 belong .
an interesting follow up of these developments has been recently done in @xcite with the interaction of the baryon decuplet with the meson octet .
also in this case a number of @xmath12 resonances are dynamically generated , which are easily identified with known resonances in the pdg@xcite , and some other ones are predicted .
these results have been confirmed in another study where poles and residues in the complex plane are searched for @xcite , and which allows one to get the couplings of the resonance to the different coupled channels . among these states
there is one suggested in @xcite , and studied in detail in @xcite , and is formed from @xmath13 interaction with quantum numbers @xmath14 , @xmath15 , and is an exotic baryon impossible to construct with three constituent quarks , hence , a resonance as exotic as the @xmath16 discovered in spring8/osaka @xcite ( see also proceedings of the pentaquark04 workshop for an update of the experimental and theoretical status of this issue @xcite ) .
one of the side effects of the study of the @xmath16 state is the test for the production of standard resonances , which is conducted both as a proof that the methods used to identify resonances do indeed work , and also to determine regions of phase space which are ideal to reduce backgrounds and , hence , see a clearer signal of the @xmath16 state .
one of the resonances thoroughly studied at spring8/osaka is the @xmath4 in the @xmath17 reaction using photons of @xmath18 to @xmath19 @xcite .
the @xmath4 peak is clearly seen in the @xmath20 missing mass spectrum , which also shows a sizeable background at energies above the peak , even when the background from kaons coming from @xmath21 decay is eliminated .
another experimental work on this reaction was done in @xcite using photons from @xmath22 , and similar features as in @xcite were observed .
we shall see that the interaction of coupled channels which leads to the @xmath4 pole , together with the d - wave character of the @xmath4 resonance in its @xmath23 decay , can explain this background .
the @xmath4 is one of the dynamically generated resonances in @xcite .
it appears from the interaction of the coupled channels @xmath1 and @xmath24 , mostly the first one .
in addition , the @xmath4 mass is just about 5 mev below the @xmath25 threshold .
all these things make the @xmath4 qualify as a quasibound @xmath1 state .
however , the small branching ratio of the @xmath4 to @xmath1 of about 4 percent in the only experiment available @xcite ( about 10 percent assuming , as done in the pdg , that the @xmath26 channel is mostly @xmath1 ) does not seem to indicate such a large coupling of the @xmath4 resonance to the @xmath1 state . actually , for the nominal values of the masses , the decay of the @xmath4 into this channel is forbidden since 1520 mev is about 5 mev below threshold of @xmath1 .
hence , it is the width of the @xmath4 and the @xmath27 , which are both rather narrow , what makes the decay possible , however , relatively small . also , when the channels are so close to threshold the branching ratio to these channels is always partially a matter of choice since it depends on the energy cuts one is taking .
a cleaner observable to find the coupling of the @xmath4 resonance to @xmath1 is hence called for .
a consequence of the nature of the @xmath4 as an approximately quasibound state of @xmath6 is a relatively large coupling of the resonance to this channel , which the chiral unitary approach provides @xcite .
one of the issues we address in this paper is how this coupling could be determined experimentally , which could shed light on the nature of the @xmath4 resonance .
for this purpose we suggest the measurement of the ratio of cross sections for the reactions @xmath28 and we evaluate this ratio within the framework of the chiral unitary approach .
preliminary results for the first reaction of eq .
( [ eq : reac ] ) have been obtained at spring8/osaka @xcite , and experimental results are also available in @xcite .
one observes there a clear peak of the @xmath4 in the @xmath3 invariant mass distribution on top of a moderate background at masses beyond the @xmath4 peak .
another aim of the present work is to show that such a background appears naturally within the chiral unitary approach , which provides scattering amplitudes and not just poles of resonances .
since the @xmath4 is dynamically generated from the @xmath1 ( and to a much lesser extend the @xmath29 coupled channel @xcite ) , the microscopic description of the @xmath2 process would be given diagrammatically by the mechanism in fig .
[ fig : unit ] , in analogy whith what was done in @xcite for the photoproduction of the @xmath9 .
we shall try to make our results the least model dependent possible , hence we avoid making an explicit model for the @xmath30 amplitude at tree level ( @xmath31 ) and similarly we also avoid making a model for the final @xmath1 state decaying into @xmath32 . the only important thing to keep in mind
is that the transition from the @xmath4 to @xmath32 proceeds in d - wave and this implies a factor @xmath33 in the transition amplitude , with @xmath34 the @xmath35 momentum in the @xmath32 center of mass frame .
the mechanism also involves the @xmath36 amplitude which implicitly contains the @xmath4 pole . for the rest of the amplitudes we shall assume a smooth energy dependence ,
although this assumption will be unnecessary when we study the ratio of the cross sections of the two reactions in eq .
( [ eq : reac ] ) , where the same vertex appears in both cases .
the amplitude for the @xmath37 reaction will be given by @xmath38 where @xmath39 , @xmath40 , are the vertices of fig .
[ fig : unit ] including , respectively , the first and last loop functions of fig .
[ fig : unit ] , involving the @xmath41 and @xmath27 propagators .
we shall assume @xmath39 and @xmath40 smooth compared with the @xmath42 amplitude which incorporates the @xmath4 structure .
the details on how to evaluate this latter amplitude can be seen in @xcite . in eq .
( [ eq : t ] ) , @xmath43 stands for the scattering amplitude of @xmath44 in isospin @xmath45 which is obtained in @xcite within the framework of a chiral unitary approach using the @xmath46 unitarization method in coupled channels @xcite ( or equivalently the bethe - salpeter equation @xcite ) and contains the resummation of the dyson series involving @xmath6 and @xmath47 loops .
the @xmath48 invariant mass distribution for the reaction @xmath49 is given by @xmath50 where @xmath51 is supposed to be a constant and @xmath52 stands for the cosinus of the @xmath53 and @xmath54 angle which is fixed by the other variables and given by @xmath55 application of the results of @xcite to the present work requires some fine tunning which we describe here .
in @xcite the position of the @xmath4 appears around 1560 mev when using a global subtraction constant in the dispersion relation formula of @xcite to reproduce the bulk of the @xmath56 resonances , and the width is larger than the nominal one .
this is because the @xmath57 channel is open at these energies .
we can do fine tunning , changing the subtraction constant from @xmath58 to @xmath59 , which brings the position of the resonance down to 1520 mev and zero width , since it is below the @xmath6 threshold and we are ignoring the @xmath27 width and the @xmath32 and @xmath60 decay channels .
we calculate the coupling , @xmath61 , of the @xmath4 resonance to the @xmath1 channel by means of the residue of the @xmath62 amplitude , which close to the pole behaves as @xmath63 we obtain @xmath64 with that procedure and assume a conservative error for @xmath61 of @xmath65 percent .
half of this uncertainty comes from varying the pole position of the @xmath4 within the experimental errors in the mass .
the rest of the error would come by assuming an uncertainty of abot @xmath66 in the @xmath27 mass to partly account for its width and from our neglect of the @xmath67 and @xmath68 channels in the build of the @xmath4 resonance . as mentioned above
, we obtain a @xmath4 pole in the real axis with this prescription , and we have used the procedure to obtain the coupling @xmath61 which is rather stable with respect to small variations of the parameters . on the other hand , in order to have a shape for the @xmath4 excitation similar to the one found in @xcite , we change the subtraction constant to @xmath69 to @xmath70 and simultaneously the mass of the @xmath27 by about @xmath71 ( to simulate contributions coming from the consideration of the @xmath27 width ) . by means of this ,
we obtain a finite width , which allows for a realistic distribution of the strength of the resonance . in fig .
[ fig : dares ] we show the experimental results for @xmath72 in the @xmath2 reaction from daresbury @xcite , together with results of our model for the two different values of the subtraction constants which have been fine tuned to get an approximate agreement with the experimental data from the @xmath4 peak onward .
we have not made any attempt to reproduce the data below the @xmath4 peak since the deficiencies of our model ( not including the @xmath67 and @xmath68 channels , the only ones open below the @xmath6 threshold ) do not allow for a realistic description of the data in that region .
however , our model contains the coupling of the @xmath4 to @xmath6 which is largely dominant , and as soon as there is phase space for @xmath6 it becomes mostly responsible for the strength of the distribution .
it is interesting to note that the fine tuning of the subtraction constant changes moderately the strength at the peak but barely changes the strength in the region of @xmath73 .
the comparatively large strength of the distribution at energies higher than the peak is due to the large @xmath74 amplitude in this region , together with the @xmath33 character of the @xmath75wave transition amplitude @xmath76 . in order to give an idea of the relative size of this strength in this region , we show in the same figure the distribution produced by a naive breit - wigner resonance around the peak ( including also the @xmath33 factor in the amplitude ) .
we see a sizeable difference which is tied to the large @xmath74 amplitude .
this is why we consider this mechanism mostly responsible for the strength in this region , particularly at energies close to the tail of the @xmath4 resonance , in spite that there are other terms of non resonant nature that can produce a background there , which we are not considering .
we would also like to note that most models for the @xmath4 ( like quark models ) that just provide a mass and a width for the resonance , would lead to a distribution ( in the absence of the background terms neglected by us ) given approximately by the breit - wigner distribution shown in the figure .
the difference between this breit - wigner form and the distribution of our model is a genuine consequence of the unitary chiral dynamics assumed in our approach .
it is interesting to see what our model gives for @xmath72 at the photon energies of spring8/osaka @xmath77 .
we show the results in fig .
[ fig : spring8 ] where we have made the average for various energies of the beam in this interval and binned the results with a resolution of @xmath78 of the experiment @xcite .
we show the results for the two different values of the subtraction constant which we used before to account for uncertainties .
we see from the results that there is also a sizeable background at energies above the peak .
the agreement with the preliminary data of @xcite , not shown in the figure , is rather fair .
this exercise has served to show consistency of the preliminary data obtained in spring8/osaka with the old data of daresbury @xcite , particularly concerning the background beyond the @xmath4 peak , which is a matter of concern when testing the ability of experimental methods to deal with backgrounds and identify peaks on top of them .
+ the other issue we address now is the evaluation of the cross section for @xmath79 with our assumption that the @xmath4 is a dynamically generated resonance , mostly from @xmath25 interaction , the mechanism for the reaction of eq .
( [ eq : kk1 ] ) is given in fig .
[ fig : unitb ] , where we can see that it shares with that of the @xmath37 reaction the primary production of @xmath1 and the full @xmath80 scattering matrix .
only the final vertices , leading to different final states , are different between these reactions
. we can easily establish a link between these two vertices using simultaneously empirical information of the @xmath4 decay into @xmath48 and the theoretical value for the coupling of the @xmath4 to @xmath1 . in order to make the comparison between the two processes clearer , we draw schematically in fig . [
fig : unitc ] the previous figures [ fig : unit ] and [ fig : unitb ] , which shows more transparently that what we need is the coupling of the @xmath4 resonance to the two decay channels .
the coupling to the @xmath1 we already found theoretically , see eq .
( [ eq : pole ] ) .
the one to the @xmath48 we find now empirically . from the pdg
we know that the partial decay width of the @xmath4 to @xmath32 is 7.02 mev .
since the vertex is of d - wave type we can take for it @xmath81 with the same normalization as the coupling of the @xmath4 to @xmath1 , by means of which the @xmath4 partial decay width into @xmath32 is given by @xmath82 this gives @xmath83 . all this said , the ratio of cross sections for the two reactions is given by @xmath84 with @xmath85 the corresponding @xmath52 variable , eq .
( [ eq : a ] ) , for the kinematics of the @xmath5 channel , and @xmath86 is the common invariant mass of the @xmath87 and @xmath3 system .
the @xmath1 scattering matrix cancels in the numerator and denominator in eq .
( [ eq : r ] ) .
thus the ratio of mass distributions for the two processes is given by the ratio of the couplings squared and the phase space , including the factor @xmath88 in the @xmath89 channel .
so far we have not made any distinction for the charge of the final states since we have being using the isospin basis and have taken the @xmath1 amplitudes in @xmath90 , the channel of the @xmath4 .
the isospin decomposition of the @xmath23 and @xmath1 states is given in eqs .
( [ eq : iso1 ] ) and ( [ eq : iso2 ] ) @xmath91 @xmath92 we see in the equations that , neglecting the @xmath93 , for which the amplitudes in the chiral unitary approach are very small , the @xmath94 channel is purely of @xmath90 .
this channel is hence ideal to isolate the @xmath90 term . on the other hand , similarly to what was done in @xcite in the photoproduction of the @xmath9
, one can see that the cross sections are proportional to @xmath95 thus , both the @xmath94 channel and the average of the @xmath96 and @xmath97 cross sections can be used to isolate the @xmath90 cross section , provided that the @xmath15 cross section is relatively small compared to the @xmath90 one .
this exercise of removing the interference term between @xmath90 and @xmath15 , may be important in case the @xmath15 component were not much smaller that the @xmath90 one , since @xmath98 can be small compared to @xmath99 but not the interference term
. this could be the case since in the region close to @xmath100 , there is another @xmath56 dynamically generated resonance , the @xmath101 , constructed from the same building blocks than the @xmath4 , @xmath25 , together with @xmath102 to which the resonance couples with largest strength .
actually , the @xmath101 should in principle be already seen in the experiment of spring8 , @xcite , but there is no trace of this resonance in this experiment .
this could be understood in terms of the small branching ratio of that resonance to the @xmath23 system , which is about @xmath103 percent according to the pdg by contrast to the @xmath104 percent of the @xmath4 . on the other hand
, we also have a small branching ratio of the @xmath101 to @xmath25 which is also of the order of 10 percent as reconstructed from the @xmath105 ratio of the pdg .
this indicates that the background of @xmath15 in the spectrum of @xmath23 , or in the one of @xmath25 , should be relatively small , and we can rely upon the @xmath90 dominance of the amplitudes of eq .
( [ eq : t ] ) , particularly in the region between the @xmath4 and @xmath101 , this is , around 1550 - 1630 mev .
given the clebsch - gordan coefficients for the isospin decomposition of the @xmath48 and @xmath94 states in eqs .
( [ eq : iso1 ] ) and ( [ eq : iso2 ] ) , we then conclude that the ratio of mass distributions for these observable channels is given by @xmath106 for the reasons given above , the average of the @xmath96 and @xmath107 cross sections could be similarly used instead of the @xmath94 one , which is not observable at present in some labs like spring8 . in fig .
[ fig : ratio ] we plot the ratio of eq .
( [ eq : rb ] ) as a function of the @xmath48 and @xmath94 invariant mass for @xmath64 and @xmath108 . in our model
, this ratio is independent of the photon energy . however , the photon energy is relevant to establish the maximum invariant mass where the ratio can be defined .
we can see in fig .
[ fig : ratio ] that , in the region we suggest , @xmath109 , the ratio @xmath110 decreases from values around @xmath111 to @xmath112 .
note that the bump in fig .
[ fig : ratio ] has nothing to do with the @xmath4 resonance , since the amplitudes producing it have canceled in the ratio .
it comes essentially from the phase space of the two reactions . with assumed uncertainties of about @xmath65 percent in @xmath113 , which would lead to about @xmath114 percent uncertainties in the ratio of fig .
[ fig : ratio ] and extra uncertainties from approximations done , we assume that an error of about @xmath115 percent is a conservative estimate of the uncertainties in the calculations .
the results obtained are essentially related to the theoretical coupling , @xmath61 , of the @xmath4 to the @xmath25 channel which is predicted by the theory .
hence the actual measurement of the ratio discussed , when compared with theoretical predictions , would produce an experimental measurement of that coupling which could substantiate the claim that the @xmath4 is a dynamically generated resonance from the interaction of the @xmath1 and @xmath24 coupled channels , and particularly from the first one . at the same time it would be interesting to evaluate the same coupling with present quark models to see if there are substantial differences , such that the experimental determination of the coupling could clearly favor one picture over the other one .
+ summarizing the results , we have done a study of some implications of the nature of the @xmath4 resonance from a perspective of chiral unitary dynamics in which framework the resonance appears as a dynamically generated state , mostly from the @xmath6 interaction in @xmath116 and @xmath90 .
first we have addressed the origin of the cross section for the @xmath2 reaction at invariant masses close and above the @xmath4 mass .
the dynamical origin of the @xmath4 within the chiral unitary approach , together with the @xmath75wave character of the @xmath117 decay , are responsible in our model for this relatively large strength , which is tied to a large coupling of the @xmath4 to the @xmath6 channel and a fairly large @xmath74 amplitude in @xmath90 .
second we have made predictions of the ratio of the @xmath5 to the @xmath2 cross sections , which is closely tied to the value of the @xmath4 coupling to @xmath6 , and which is provided by the chiral unitary approach . therefore , an experimental measurement of such ratio would provide a test of the claimed nature of this resonance . |
semiconductor superlattices ( sls ) are nanostructures formed from several alternating layers of different semiconductor materials @xcite .
this periodic structure leads to the formation of energy minibands that enable electrons , in the presence of an electric field , to demonstrate a number of interesting quantum - mechanical phenomena , which include the formation of wannier - stark ladders , sequential and resonant tunneling , bragg reflections , and bloch oscillations .
consequently , sls are of a great interest for both fundamental and applied science @xcite . due to the high mobility of miniband electrons and the very high frequency of the bloch and charge - domain oscillations ,
sls have prospective applications in sub - thz and thz electronic devices @xcite .
recently , it has been shown that a tilted magnetic field applied to a sl can strongly affect , and hence control , the electrical properties of the structure .
nonlinear interaction between the electronic bloch oscillations along the superlattice and cyclotron motion in the plane of the layers induces chaotic semiclassical electron dynamics , which , depending on the ratio between the bloch and cyclotron frequencies , either accelerate or decelerate charge transport through the sl @xcite .
coupling of cyclotron and bloch motion by a tilted magnetic field has also been related to the fiske effect in superconducting josephson junctions @xcite . on resonance , when the bloch and cyclotron frequencies are commensurate , the electrons
exhibit a unique type of quantum chaos , which does not obey kolmogorov - arnold - moser theory@xcite .
this type of chaos is characterised by the formation of intricate `` web - like '' structures , known in the literature as `` stochastic webs ''
@xcite , which extend throughout the phase space of the miniband electrons .
the appearance of these webs abruptly delocalises the electrons in real space , thus significantly increasing their drift velocity @xcite .
similar dynamics can occur in other spatially periodic systems in which wave interference gives rise to band transport phenomena , including ultracold atoms in an optical lattice @xcite , graphene @xcite , and light propagating through spatially - modulated photonic crystals @xcite .
on the quantum level , such resonant delocalisation of the electrons manifests itself in the formation of additional magnetic - field - induced miniband structure corresponding to extended electron states , which extend across many periods of the sl @xcite .
this delocalisation dramatically affects the collective electron behavior by inducing multiple propagating charge domains and ghz - thz current oscillations with frequencies and amplitudes much higher than with no tilted field @xcite .
it has also been shown that in the vicinity of bloch - cyclotron resonances the usually unstable bloch gain profile becomes stable @xcite , which can be used for the amplification of thz signals .
although the above effects of the bloch - cyclotron resonances on electron dynamics have been reported for different finite temperatures , it is still unclear how the character of charge transport in the sl changes with temperature . plane represent quantum wells ( tunnel barriers ) within the superlattice .
electric field , @xmath0 , is applied anti - parallel to the @xmath1-axis and the magnetic field , @xmath2 , lies in the @xmath3-plane at an angle @xmath4 to the @xmath1-axis . ] in this paper , we study how increasing the electron and lattice temperature affects the drift velocity of the electrons in an electric and tilted magnetic field and , consequently , the electric current through the sl .
we find that the thermal distribution of the electrons can , counter - intuitively , enhance the effect of bloch - cyclotron resonances on the drift velocity .
in particular , increasing the temperature makes the resonant maxima in the drift velocity versus electric field characteristics more prominent .
this shifts the frequency and amplitude of the collective charge - domain oscillations , associated with each resonant peak , together with the threshold voltage at which the oscillations switch on .
we consider the field configuration shown schematically in fig .
[ fig : conf ] .
the electric field @xmath5 is applied perpendicular to the plane of the sl layers and anti - parallel to the @xmath1-axis .
the magnetic field @xmath6 lies in the @xmath3-plane at an angle @xmath4 to the @xmath1-axis .
the semiclassical equation of motion for a miniband electron is @xmath7 \label{veceq}\ ] ] where @xmath8 is the magnitude of the electron charge , @xmath9 is the electron s crystal momentum at time @xmath10 and @xmath11 is the dispersion relation for the lowest miniband within the tight - binding approximation . here , @xmath12 is the miniband width , @xmath13 is the sl period , and @xmath14 is the electronic effective mass for motion in the @xmath15-plane . equation ( [ veceq ] ) can be expressed in its constituent components @xcite @xmath16 where @xmath17=@xmath18 and @xmath19=@xmath20 are the cyclotron frequencies corresponding to the magnetic field components along the @xmath1- and @xmath21-axes , respectively . the electron velocity along the @xmath1-direction
is given by @xmath22 where the peak velocity @xmath23 . to determine the drift velocity , @xmath24 , of electrons with initial momentum @xmath25 we use the esaki - tsu formalism @xcite : @xmath26 where @xmath27 is the electron scattering rate @xcite .
in the case of non - zero temperature , @xmath28 , one should take into account the thermal distribution of electron momenta @xmath29 , which we assume to obey the boltzmann statistics @xcite @xmath30 in eq .
( [ eq : boltzmann ] ) , @xmath31 can be found from the normalisation condition @xmath32 which yields @xmath33 where @xmath34 is a modified bessel function of the first kind .
then , the drift velocity of miniband electrons at temperature @xmath28 is @xmath35 in the appendix we show , following previous work @xcite , that for small angles @xmath4 , for which @xmath36 and @xmath37 , where @xmath38 is the frequency of the bloch oscillations , the drift velocity can be approximated by @xmath39 \sum^{\infty}_{n = - \infty } i_n \left [ m^{\ast } k_b t \left ( \frac{\omega_{\perp}}{\text{$\omega_{\|}$ } } \frac{d}{\hbar } \right)^2 \right ] \frac{\nu ( \omega_b - n \omega_{\|})}{\nu^2 + ( \omega_b - n \omega_{\|})^2 } , \label{eq : bass}\end{aligned}\ ] ] where @xmath40 , @xmath41 are the modified bessel functions of the first kind . when @xmath42 , eq .
( [ eq : bass ] ) reduces to @xmath43 note that eq .
( [ eq : romanov ] ) can be derived exactly @xcite by integrating eqs .
( [ compeq1])-([compeq3 ] ) with @xmath42 , and substituting the resulting expressions for @xmath44 , @xmath45 and @xmath46 into eqs .
( [ eq : vx ] ) and ( [ eq : vdt ] ) . for @xmath47
the equation ( [ eq : romanov ] ) yields the famous esaki - tsu relation for the electron drift velocity @xcite
. in the general case of arbitrary parameters @xmath48 , @xmath49 and @xmath4 , eqs .
( [ compeq1])-([compeq3 ] ) , which , in principle , can exhibit deterministic chaos @xcite , can not be solved analytically , and therefore the electron drift velocity must be calculated numerically . in our computations
we use the following sl parameters , taken from recent experiments @xcite : @xmath50 nm , @xmath51 mev , @xmath52 s@xmath53 and @xmath54 , where @xmath55 is the mass of a free electron .
in our numerical simulations , we integrate eqs . ( [ compeq1])-([compeq3 ] ) for a number of different initial conditions @xmath56 , which were randomly generated according to the probability density function ( [ eq : boltzmann ] ) .
this function can be decomposed into momentum components @xmath57 as a consequence of this decomposition , one can use independent random number generators for each of the initial momentum components @xmath58 , @xmath59 , @xmath60 .
the values of @xmath59 and @xmath60 can be obtained from standard routines @xcite for the generation of uncorrelated random numbers obeying a gaussian distribution with zero mean and variance @xmath61 .
however , the generation of random values of @xmath58 requires a more sophisticated procedure , which we now explain .
let us find a continuous function , @xmath62 , which transforms a random variable , @xmath63 , uniformly distributed within the interval @xmath64 , into a random variable , @xmath65 , having a distribution @xmath66 \right).\ ] ] to find @xmath62 , we use the probability conservation relation @xmath67 , which gives us a differential equation for @xmath62 : @xmath68}. \label{eq : eq_dis}\ ] ] due to the symmetry of the distribution in eq .
( [ eq : pxdistr ] ) , @xmath69 , which can be used as an initial condition for eq .
( [ eq : eq_dis ] ) .
moreover , it is possible to show that @xmath70 for any values of temperature @xmath28 .
however , in the general case of arbitrary argument , @xmath63 , @xmath62 can not be expressed analytically .
we therefore evaluate it by numerical integration of eq .
( [ eq : eq_dis ] ) . for three different temperatures : @xmath71 k ( curve 1 ) , @xmath72 k ( curve 2 ) , @xmath73 k ( curve 3 ) .
@xmath50 nm , @xmath51 mev , @xmath52 s@xmath53 .
[ fig : transf ] ] figure [ fig : transf ] illustrates the function @xmath62 for three different temperatures @xmath71 k ( curve 1 ) , @xmath72 k ( curve 2 ) , @xmath73 k ( curve 3 ) . as the figure shows , @xmath62 is an odd function , which becomes almost linear as @xmath28 increases . in practice , numerical integration of eq .
( [ eq : eq_dis ] ) significantly slows the simulations .
therefore , we used approximate analytical solutions .
we found that for small temperatures , @xmath74 k , the distribution ( [ eq : pxdistr ] ) can be well approximated by a gaussian with zero mean and variance @xmath75 .
hence , in this case @xmath58 can be also defined using the gaussian random number generator . for larger temperatures @xmath76
k we approximate the solution of eq .
( [ eq : eq_dis ] ) by the following polynomial expression @xmath77 where the coefficients @xmath78 for different temperatures are given in tab .
[ tab_1 ] .
.coefficients @xmath79 for the polynomial expansion in eq .
( [ eq : poly ] ) at different temperatures @xmath28[tab_1 ] [ cols="^,^,^,^,^,^",options="header " , ] note that with increasing @xmath28 , the coefficient @xmath80 , corresponding to the linear term , grows , whereas the other ( nonlinear ) coefficients , @xmath81 , @xmath82 , @xmath83 , decrease .
also , the higher - order terms can be neglected .
thus , to calculate the drift velocity , @xmath84 , for given temperature , @xmath28 , we ( i ) randomly generate the components of the initial momentum vector @xmath85 , ( ii ) use them as an initial condition to integrate eqs .
( [ compeq1])-([compeq3 ] ) , ( iii ) substitute the results of this integration into eqs .
( [ eq : vx ] ) and ( [ eq : ud ] ) to determine the drift velocity , @xmath84 , for the given initial momentum , and ( iv ) average @xmath84 over all randomly - chosen initial momenta .
to ensure that the above procedure for simulating the thermal distribution of electron momenta is accurate enough , we check it for the case of no applied magnetic field , for which @xmath84 can be calculated analytically @xcite using eq .
( [ eq : romanov ] ) . ,
with electric field , @xmath48 , determined analytically from eq .
( [ eq : romanov ] ) ( solid curves ) and calculated numerically ( symbols ) for @xmath86 k ( curve 1 and `` @xmath87 '' ) and @xmath73 k ( curve 2 and `` @xmath88 '' ) . ( b ) variation of maximal drift velocity , @xmath89 , with @xmath28 calculated analytically ( solid line ) and numerically ( @xmath87 ) . numerical simulations involved averaging over 250000 different initial momenta , @xmath85 , chosen randomly .
[ fig : vdb0 ] ] the solid curves in fig .
[ fig : vdb0](a ) shows @xmath90 plots calculated using eq .
( [ eq : romanov ] ) for @xmath86 k ( upper curve ) and @xmath73 k ( lower curve ) .
the data points marked by symbols `` @xmath87 '' and `` @xmath88 '' show values of @xmath84 obtained numerically for @xmath86 k and @xmath73 k , respectively .
the figure demonstrates excellent agreement between the exact eq .
( [ eq : romanov ] ) and our numerical procedure .
more precisely , we find that by averaging over @xmath91 trajectories , the relative error does not exceed @xmath92 .
consequently , in all our simulations we used @xmath93 .
( [ eq : romanov ] ) shows that the maximum drift velocity , @xmath89 , is achieved when @xmath94 , i.e. when a significant fraction of the electrons complete whole bloch oscillations before scattering .
as @xmath95 increases beyond @xmath27 , @xmath89 decreases with increasing @xmath28 according to the equation @xmath96 the variation of @xmath89 with @xmath28 predicted by eq .
( [ eq : maxdriftvelocityt>0theta=0 ] ) is shown by the solid curve in fig .
[ fig : vdb0](b ) . for comparison , the result of our numerical simulation is shown by the symbols `` @xmath87 '' .
the figure shows excellent quantitative agreement between our analytical and numerical results .
we have also applied the numerical procedure described in the previous section to the case of a tilted magnetic field , for which the drift velocity @xmath84 can not be determined analytically .
, with ( lower scale ) electric field , @xmath48 , ( upper scale ) @xmath97 calculated for @xmath98 t and @xmath99 and different temperatures .
curve 1 corresponds to @xmath47 k ; curve 2 to @xmath71 k ; curve 3 to @xmath100 k ; curve 4 to @xmath73 k ; curve 5 to @xmath101 k. [ fig : vdb15t40 ] ] figure [ fig : vdb15t40 ] shows @xmath84 versus @xmath48 curves calculated numerically for @xmath98 t and @xmath99 at several different temperatures @xmath28 .
in contrast to the case of @xmath42 [ fig . [
fig : vdb0](a ) ] , all of the @xmath90 curves in fig .
[ fig : vdb15t40 ] exhibit multiple maxima .
the first maximum , for the lowest value of @xmath48 , also exists when @xmath42 and , as noted in the previous section , is associated with the onset of bloch oscillations .
all other maxima occur because of the enhanced acceleration of the electrons whenever the ratio of the bloch and cyclotron frequencies @xmath102 , or @xmath103 ( upper scale in fig .
[ fig : vdb15t40 ] ) @xcite .
remarkably , as @xmath28 increases , the first peak weakens significantly , whereas the amplitude of the cyclotron - bloch resonances ( i.e. the peak - to - valley ratio ) increases sharply even though the peak @xmath84 values decrease .
moreover , for high enough temperatures , new peaks , reflecting higher order resonances , appear in the @xmath90 dependencies .
plane taken with strobe period @xmath104 for the @xmath105 resonance ( @xmath98 t , @xmath99 ) .
( a ) `` stochastic web '' and neighboring quasi - periodic orbits ;
( b ) , ( c ) enlarged part of `` stochastic web '' ( black dots ) and probability ( grey - scale map in which lighter shades correspond to higher probability ) that the initial momentum ( @xmath106 ) lies in the given area of the @xmath107 plane calculated for @xmath71 k ( b ) , and @xmath101 k ( c ) .
[ fig : webz ] ] to qualitatively understand the effects of temperature on drift velocity , we analysed single electron orbits described by the eqs .
( [ compeq1])-([compeq3 ] ) .
figure [ fig : webz](a ) shows a stroboscopic poincar section phase portrait for the @xmath105 resonance ( @xmath98 t and @xmath99 ) , taking the strobe period to be @xmath104 .
in addition to localised quasi - periodic orbits , which appear in the stroboscopic section as dotted closed curves , the phase space also contains a resonant unbounded structure formed by chaotic layers and known in the literature as a `` stochastic web '' @xcite .
it has been shown previously @xcite that electrons having initial momenta in the stochastic web move rapidly through the sl , so producing a large resonant peak in the @xmath90 curve , as shown in fig .
[ fig : vdb15t40 ] for @xmath108 .
the height of each resonant peak increases as more electrons have momenta lying within the stochastic web . in figs .
[ fig : webz](b),(c ) we show enlarged fragments of the stochastic web near the origin together with a grey - scale map indicating the probability of finding an electron with given initial momentum components @xmath109 at @xmath71 k [ fig . [
fig : webz ] ( b ) ] and @xmath101 k [ fig .
[ fig : webz](c ) ] .
comparison of these two figures reveals that , with increasing temperature , the probability of the electron s initial momentum lying within the stochastic web grows . at the same time
, the number of electrons with higher initial momenta , and thus higher velocity , also increases .
together , these two effects increase @xmath84 and thus enhance the bloch - cyclotron resonant peaks in fig .
[ fig : vdb15t40 ] .
interestingly , although eq .
( [ eq : bass ] ) is unable correctly to predict the exact electron drift velocity for an arbitrary magnetic field configuration , it still captures certain qualitative trends in the variation of @xmath84 with the field parameters and @xmath28 .
first , we note that according to eq .
( [ eq : bass ] ) , the expression for @xmath84 consists of a series of resonant terms proportional to @xmath110 $ ] .
the term with @xmath111 reflects the esaki - tsu @xmath90 curve , whereas other terms with @xmath112 correspond to the bloch - cyclotron resonances , for which @xmath113 is an integer .
moreover , since , for very large @xmath1 , @xmath40 has the asymptotic form @xmath114 , as @xmath28 increases in eq .
( [ eq : bass ] ) the higher - order resonant terms strengthen . under certain conditions
this leads to the appearance of additional peaks in @xmath90 , which we see in our numerical simulations shown in fig .
[ fig : vdb15t40 ] .
in order to study the effects of temperature on the collective dynamics of electrons in the sl , we self - consistently solve discrete versions of the current continuity and poisson equations , splitting the miniband transport region into @xmath115 layers , each of width @xmath116 nm , which is small enough to approximate a continuum @xcite .
the evolution of the electron density , @xmath117 , in the @xmath118 layer , whose right - hand edge is at @xmath119 , is given by the discretised current continuity equation @xmath120 where @xmath121 [ @xmath122 are the areal current densities at the left- [ right- ] hand edges of the @xmath118 layer . within the drift - diffusion approximation ,
the current density is @xmath123 where @xmath124 is the mean electric field in the @xmath118 layer@xcite , @xmath125 is the diffusion coefficient @xcite , and the drift velocity @xmath126 is obtained by using the approach described in section [ sec : num_proc ] . since @xmath127 depends on the local drift velocity , @xmath126 , the collective electron dynamics depend directly on the single - electron orbits .
the electric field @xmath128 [ @xmath129 at the left- [ right- ] hand edges of the @xmath118 layer can be described by the discretised poisson equation @xmath130 where @xmath131 and @xmath132 are , respectively , the absolute and relative permittivities and @xmath133 m@xmath134 is the n - type doping density in the sl layers@xcite . in the emitter and collector ohmic contacts , @xmath135 .
characteristics calculated for @xmath136 t ( a ) and @xmath137 t , @xmath138 ( b ) at ( from top to bottom ) @xmath47 k , 50 k , 100 k , 200 k and 300 k. in ( a ) , data points on dashed line show values of @xmath139 calculated from eqs .
( [ eq : criticalcurrent ] ) and ( [ eq : criticalvoltage ] ) and discussed in the text .
[ fig : ivsv ] ] we use ohmic boundary conditions to determine the current , @xmath140 , in the heavily doped emitter contact whose electrical conductivity @xmath141 sm@xmath53.@xcite the voltage , @xmath142 , applied to the device is a global constraint given by @xmath143 where the voltage , @xmath144 , dropped across the contacts includes the effect of charge accumulation and depletion in the emitter and collector regions and a contact resistance @xmath145@xcite .
we calculate the current as @xmath146 where @xmath147 m@xmath148 is the cross - sectional area of the sl.@xcite the model described by eqs .
( [ eq : continuity ] ) - ( [ eq : tot_cur ] ) exhibits both constant and oscillating electric current , depending on the voltage , @xmath142 , applied to the device .
typical current - voltage [ @xmath149 characteristics are shown in fig .
[ fig : ivsv ] both for @xmath42 and when a tilted magnetic field ( @xmath98 t , @xmath150 ) is applied .
note that for @xmath142 values at which current oscillations occur , the dc current was calculated by averaging @xmath151 over time . when @xmath42 [ fig .
[ fig : ivsv ] ( a ) ] , the @xmath152 curves reveal the usual esaki - tsu - like behavior , characterised by a single maximum , which is associated with the onset of single - electron bloch oscillations throughout much of the sl transport region .
figure [ fig : ivsv ] ( a ) shows that as @xmath28 increases , the peak current decreases , which agrees well with a number of experimental observations @xcite .
this decrease of the peak current reflects that of the maximal electron drift velocity , predicted by eq .
( [ eq : maxdriftvelocityt>0theta=0 ] ) , as @xmath28 increases . in particular , the factor @xmath153 , characterising the effect of the temperature on the drift velocity in eq .
( [ eq : maxdriftvelocityt>0theta=0 ] ) , also accurately describes the drop of the peak current both in our simulations and in earlier experiments discussed in refs .
@xcite the electron dynamics changes significantly when a tilted magnetic field is applied to the sl .
figure [ fig : ivsv ] ( b ) shows typical @xmath152 curves calculated , in the presence of a tilted magnetic field , for a range of temperatures .
all of the @xmath152 curves reveal clear bloch - cyclotron resonances , which manifest themselves through the appearance of additional features in the curves @xcite . for low temperatures ,
the bloch - cyclotron resonances produce sudden changes in the slope of the @xmath152 curves [ for example , the kink near @xmath154 v in the ( upper ) @xmath152 curve calculated for @xmath47 in figure [ fig : ivsv ] ( b ) ] and also shift the position of the current peak . as @xmath28 increases ,
these effects become more prominent and , eventually , give rise to additional maxima in the @xmath152 curves , e.g. the double peaks at @xmath100 k. this evolution of the @xmath152 characteristics originates from the variation of the @xmath90 curves with changing temperature , shown in fig .
[ fig : vdb15t40 ] .
comparison of figs .
[ fig : ivsv ] ( a ) and ( b ) shows that although the maximum current decreases with increasing temperature in both cases , the peak dc current is always larger in the presence of a tilted magnetic field .
this can be explained by the strong resonant enhancement of @xmath84 produced by the bloch - cyclotron resonances . moreover , as fig .
[ fig : vdb15t40 ] indicates , with increasing temperature the amplitude of these resonances increases and so the difference between the peak currents for @xmath42 and for @xmath155 becomes larger .
, with voltage , @xmath142 , applied to the sl calculated for @xmath42 t ( a ) and @xmath98 t , @xmath150 ( b ) at the temperatures indicated .
[ fig : fvsv ] ] when the applied voltage , @xmath142 , exceeds some critical value , @xmath156 , which depends on @xmath28 , @xmath49 , and @xmath4 , the stationary state of the system loses its stability , and the electric current starts to oscillate at a frequency , @xmath157 , which , for the given parameters , is in the microwave range .
these current oscillations are associated with the formation of traveling charge domains @xcite .
, with voltage , @xmath142 , applied to the sl calculated for @xmath42 t ( a ) and @xmath98 t , @xmath150 ( b ) at the temperatures indicated .
[ fig : iavsv ] ] the variation of the oscillation frequency , @xmath157 , with @xmath142 is shown in fig .
[ fig : fvsv ] for a range of @xmath28 .
note that for each @xmath28 value , the corresponding @xmath158 curve starts at a particular threshold voltage , @xmath156 , above which @xmath151 oscillations occur .
in general , as @xmath142 increases , @xmath157 decreases for both @xmath42 [ fig .
[ fig : fvsv ] ( a ) ] and @xmath98 t , @xmath150 [ fig .
[ fig : fvsv ] ( b ) ] .
this can be understood in terms of the @xmath90 curves shown in figs .
[ fig : vdb0 ] and [ fig : vdb15t40 ] .
these figures suggest that when @xmath48 is sufficiently large , @xmath84 decreases with increasing @xmath48 for both @xmath42 and @xmath98 t. moreover , further increasing @xmath48 in this regime slows the electrons even more .
the applied voltage produces an inhomogeneous electric field distribution throughout the sl layers , which is determined by eqs .
( [ eq : continuity])-([eq : volt ] ) . as @xmath142 increases , the region of high electric field inside the sl , where @xmath84 is small , also increases and so the speed of the charge domains is low
. the larger the region of high electric field , the smaller the average velocity of the charge domains and , hence , the lower the frequency of current oscillations .
when @xmath98 t and @xmath150 , the multiple peaks in the @xmath90 curve tend to keep the average @xmath84 value high , which means that the charge domains move faster than for @xmath42 .
in addition , the onset of bloch - cyclotron resonances increases the _ maximal _ @xmath84 value that the electrons attain for given @xmath28 [ see fig . [
fig : vdb15t40 ] ] .
all of these factors improve the propagation speed of charge domains in the presence of a tilted magnetic field .
thus , for given @xmath142 , the frequency of charge domain oscillations ( if they occur at that @xmath142 ) is higher when a tilted magnetic field is applied . the evolution of the amplitude of the current oscillations , @xmath159 , with @xmath142 is shown in fig . [ fig : iavsv ] .
comparison of figs .
[ fig : iavsv ] ( a ) and [ fig : iavsv ] ( b ) , reveals that applying a tilted magnetic field at any temperature allows one to generate current oscillations of higher amplitude and power .
the mechanisms that produce this power enhancement were discussed in detail in ref.@xcite . in particular , it was shown that multiple peaks in the @xmath90 curves induce multiple propagating charge domains , which strengthen the current oscillations . since the multi - peak character of the @xmath90 curves persists over a wide range of @xmath28 ( see fig .
[ fig : vdb15t40 ] ) , the power enhancement induced by a tilted magnetic field occurs even at room temperature .
remarkably , the modification of the @xmath152 curves by a tilted magnetic field can lead to the appearance of new regions of negative differential conductance , for example associated with the two peaks in the @xmath152 curve at @xmath100 k ( purple curve in fig .
[ fig : ivsv ] ) . in this case
, there are several distinct regions of @xmath142 within which the sl exhibits current oscillations [ see fig . [
fig : iavsv ] ( b ) ] . figures [ fig : fvsv ] and [ fig : iavsv ] indicate that for both @xmath42 and for @xmath98 t , the threshold voltage for current oscillations , @xmath156 , depends non - trivially on @xmath28 . for the low temperature range @xmath160 k , increasing @xmath28 decreases @xmath156 . but
as @xmath28 increases further , @xmath156 starts to increase .
this is due to the effect of temperature on the current density , @xmath161 , which is described by eqs .
( [ eq : j_eq ] ) and ( [ eq : diff_eq ] ) .
as noted above , the generation of oscillating current in the sl is associated with the formation of traveling change domains , which occurs when the voltage , @xmath162 , dropped across the active layers of the sl becomes sufficiently large @xcite .
( [ eq : volt ] ) shows that this voltage is the difference between the total voltage , @xmath142 , applied to the sl and the voltage , @xmath144 , dropped across the contacts .
the voltage , @xmath144 , in turn , may be expressed as @xmath163 , where @xmath164 describes the effect of charge accumulation and depletion in the emitter and collector regions .
thus , any changes in the maximal value of @xmath161 and , therefore , of @xmath165 , will affect @xmath144 , and thus shift the threshold voltage @xmath156 . for low temperatures ,
when electron diffusion can be neglected , increasing @xmath28 decreases the maximal electron drift velocity and , thus , the maximal value of @xmath165 .
this reduces the voltage , @xmath144 , dropped across the contacts , since this is proportional to the current through those contacts . when @xmath144 decreases but @xmath142 remains constant , the voltage , @xmath162 , dropped across the active layers of the sl increases .
thus , within some temperature range , increasing @xmath28 decreases @xmath156 .
the value of @xmath156 and the corresponding critical dc current @xmath166 can be estimated analytically at low @xmath28 and when @xmath42 . if one neglects electron diffusion and uses the drift velocity given by eq .
( [ eq : maxdriftvelocityt>0theta=0 ] ) , the critical current , @xmath167 , at temperature @xmath28 can be estimated as @xmath168 where @xmath169 is the critical current for @xmath47 k. according to eq .
( [ eq : volt ] ) , the critical voltage , @xmath170 , required for current oscillations is @xmath171 where @xmath172 is the voltage dropped across the sl layers .
assuming that @xmath164 and @xmath162 are almost independent of @xmath28 , one obtains @xmath173 where @xmath174 is the critical voltage at @xmath47 . from eqs .
( [ eq : criticalcurrent ] ) and ( [ eq : criticalvoltage ] ) it follows that @xmath175 the @xmath176 curve defined by eq .
( [ eq : criticalline ] ) is shown by the dashed line in fig .
[ fig : ivsv ] ( a ) .
the open circles mark the coordinates @xmath177 for the five different @xmath28 values corresponding to the five @xmath152 plots shown in the figure .
the values of @xmath174 and @xmath169 were taken from the @xmath152 curve for @xmath47 . from figs .
[ fig : ivsv ] ( a ) and [ fig : fvsv](a ) it is easy to see that for the low @xmath28 values considered ( @xmath178 k ) , the analytical values of @xmath156 and @xmath167 obtained from eq .
( [ eq : criticalline ] ) are in excellent quantitative agreement with those inferred from the numerical @xmath152 curves . for large @xmath28 ,
the diffusion contribution to the current becomes important and eq .
( [ eq : criticalline ] ) fails .
as @xmath28 increases , the current through the sl and the voltage dropped across the contacts also increase , thereby raising @xmath156 .
the same mechanism operates when @xmath155 . in this case
, @xmath156 is larger than for @xmath42 ( see fig . [
fig : fvsv ] ) .
this is because in the presence of a tilted magnetic field , the maximal drift velocity is larger due to the onset of bloch - cyclotron resonances ( compare figs . 3 and 4 ) .
in conclusion , we have shown that a tilted magnetic field strongly affects , and can significantly enhance , the transport characteristics of sls even at room temperature .
the temperature dependence of the electron drift velocity in the presence of a tilted magnetic field is quite different from the @xmath42 case . in particular
, increasing @xmath28 quickly suppresses the @xmath42 esaki - tsu peak in the @xmath90 curve but has a much smaller effect when a tilted magnetic field is applied . in this geometry ,
increasing @xmath28 sometimes even enhances the drift velocity peaks , relative to the background , caused by the bloch - cyclotron resonances . as a result ,
when @xmath49 and @xmath4 are both non - zero , the electrons move faster , which increases both the dc- and ac - components of the current through the sl and also improves its high - frequency performance .
the effects of temperature on the @xmath90 curves also transform the sl s @xmath152 curves by inducing new instabilities and transport regimes , which we hope will stimulate future experimental and theoretical studies .
since the effects that we have demonstrated are generic features of semiclassical energy band transport , we expect similar phenomena to occur in , for example , cold atoms @xcite and photonic crystals @xcite .
the work was supported by the federal special purpose programme `` scientific and educational personnel of innovation russia ( 2009 - 2013 ) '' , the uk engineering and physical sciences research council and the `` dynasty '' foundation .
we solved eqs .
( [ compeq1])-([compeq3 ] ) assuming @xmath179 . under this condition eqs .
( [ compeq1])-([compeq3 ] ) can be linearised as @xmath180 which for initial conditions @xmath181 , @xmath182 , @xmath183 yields @xmath184 where @xmath185 , and @xmath186 .
@xmath187 = & & \nonumber\\ \frac{v_0}{2 i } \left [ \exp \left ( i \frac{p_{\| } d}{\hbar } + i \omega_b
t - i \frac{\omega_{\perp}}{\text{$\omega_{\|}$ } } \frac{d}{\hbar } p_{\perp } \sin ( \text{$\text{$\omega_{\|}$ } t$ } + \varphi_0 ) + i \frac{\omega_{\perp}}{\text{$\omega_{\|}$ } } \frac{d}{\hbar } p_{\perp } \sin \varphi_0 \right ) - \right . & & \\ \left .
\exp \left ( - i \frac{p_{\| } d}{\hbar } - i \omega_b t + i \frac{\omega_{\perp}}{\text{$\omega_{\|}$ } } \frac{d}{\hbar } p_{\perp } \sin ( \text{$\text{$\omega_{\|}$ } t$ } + \varphi_0 ) - i \frac{\omega_{\perp}}{\text{$\omega_{\|}$ } } \frac{d}{\hbar } p_{\perp } \sin \varphi_0 \right ) \right ] , & & \nonumber\end{aligned}\ ] ] the drift velocity , @xmath191 , of electrons with initial momentum components @xmath192 and @xmath193 can be calculated using the esaki - tsu formula ( [ eq : ud ] ) . substituting ( [ eq : app_vx ] ) in ( [ eq : ud ] ) we get to find the drift velocity , @xmath84 , at finite temperature @xmath28 we average @xmath196 according to the thermal distribution of electron momenta ( [ eq : boltzmann ] ) . in the current notations
the definition of @xmath84 becomes @xmath199 d p_{\| } - & & \\ - \frac{1}{z } \frac{v_0}{2 i } b ( p_{\perp } ,
\varphi_0 ) e^{- \frac{\delta}{2 k_b t } } \int^{\pi \hbar / d}_{- \pi \hbar / d } \exp \left ( - i \frac{p_{\| } d}{\hbar } \right ) \exp \left [ \frac{\delta \cos \frac{p_{\| } d}{\hbar}}{2 k_b t } \right ] d p_{\| } & & \end{aligned}\ ] ] @xmath203 \exp \left [ - \frac{p^2_{\perp}}{2 m^{\ast } k_b t } \right ] p_{\perp } d p_{\perp } d \varphi_0=\\ \nonumber = & & \frac{(2 \pi)^2 v_0}{z } \frac{\hbar}{d } e^{- \frac{\delta}{2 k_b t } } i_1 \left ( \frac{\delta}{2 k_b t } \right ) \int^{\infty}_0 \sum^{\infty}_{n = - \infty } j^2_n \left ( \frac{\omega_{\perp}}{\text{$\omega_{\|}$ } } \frac{d}{\hbar } p_{\perp } \right ) \frac{\nu ( \omega_b - n \omega_{\|})}{\nu^2 + ( \omega_b - n \omega_{\|})^2 } \times \exp \left [ - \frac{p^2_{\perp}}{2 m^{\ast } k_b t } \right ] p_{\perp } d p_{\perp},\\ \label{vdprel}\end{aligned}\ ] ] @xmath205 p_{\perp } d p_{\perp } + \\
\nonumber \left .
+ \int^{\infty}_0 j^2_n \left ( \frac{\omega_{\perp}}{\text{$\omega_{\|}$ } } \frac{d}{\hbar } p_{\perp } \right ) \frac{\nu ( \omega_b + n \omega_{\|})}{\nu^2 + ( \omega_b - n \omega_{\|})^2 } \times \exp \left [ - \frac{p^2_{\perp}}{2 m^{\ast } k_b t } \right ] p_{\perp } d p_{\perp } \right].\end{aligned}\ ] ] @xmath206 p_{\perp } d p_{\perp } = m^{\ast } k_b t \exp
\left [ - m^{\ast } k_b t \left ( \frac{\omega_{\perp}}{\text{$\omega_{\|}$ } } \frac{d}{\hbar } \right)^2 \right ] i_n \left [ m^{\ast } k_b t \left ( \frac{\omega_{\perp}}{\text{$\omega_{\|}$ } } \frac{d}{\hbar } \right)^2 \right ] & & \end{aligned}\ ] ] is , for @xmath207 , a special from of the second weber exponential integral @xcite . after
integration and re - indexing the sum in ( [ eq : app_vd2 ] ) , and substitution of the expression for @xmath31 from ( [ eq : znorm ] ) we obtain the final formula for the drift velocity @xmath208 \sum^{\infty}_{n = - \infty } i_n \left [ m^{\ast } k_b t \left ( \frac{\omega_{\perp}}{\text{$\omega_{\|}$ } } \frac{d}{\hbar } \right)^2 \right ] \frac{\nu ( \omega_b - n \omega_{\|})}{\nu^2 + ( \omega_b - n \omega_{\|})^2}.\ ] ] e. schomburg , t. blomeier , k. hofbeck , j. grenzer , s. brandl , i. lingott , a. a. ignatov , k. f. renk , d. g. pavelev , yu . koschurinov , b. ya .
melzer , v. m. ustinov , s. v. ivanov , a. zhukov , and p. s. kopev , phys .
b * 58 * , 4035 ( 1998 ) .
p. khosropanah , a. baryshev , w. zhang , w. jellema , j. n. hovenier , j. r. gao , t. m. klapwijk , d. g. paveliev , b. s. williams , s. kumar , q. hu , j. l. reno , b. klein , and j. l. hesler , optics letters , * 34 * , 2958 ( 2009 ) .
in order to simplify the analysis , we neglect the temperature dependence of the contact resistance and electron scattering rate , which is an acceptable assumption for the sls used in several experiments @xcite . | we show that resonant electron transport in semiconductor superlattices with an applied electric and tilted magnetic field can , surprisingly , become more pronounced as the lattice and conduction electron temperature increases from 4.2 k to room temperature and beyond .
it has previously been demonstrated that at certain critical field parameters , the semiclassical trajectories of electrons in the lowest miniband of the superlattice change abruptly from fully localised to completely unbounded .
the unbounded electron orbits propagate through intricate web patterns , known as stochastic webs , in phase space , which act as conduction channels for the electrons and produce a series of resonant peaks in the electron drift velocity versus electric field curves . here
, we show that increasing the lattice temperature strengthens these resonant peaks due to a subtle interplay between thermal population of the conduction channels and transport along them .
this enhances both the electron drift velocity and the influence of the stochastic webs on the current - voltage characteristics , which we calculate by making self - consistent solutions of the coupled electron transport and poisson equations throughout the superlattice .
these solutions reveal that increasing the temperature also transforms the collective electron dynamics by changing both the threshold voltage required for the onset of self - sustained current oscillations , produced by propagating charge domains , and the oscillation frequency . |
polycyclic aromatic hydrocarbons ( pah s ) are hydrocarbon compounds with multiple six - membered aromatic rings . such species are thought to be a common component of the interstellar medium in the milky way and in external galaxies .
pah s are typically observed in infra - red emission bands following excitation by either ultra - violet or visible radiation in assorted regions of the interstellar medium exposed to radiation . despite all attempts ,
no specific types of pah s responsible for these features have been identified , and the form and intensity of the spectra vary strongly from one region to another @xcite , probably reflecting a variety of sizes and abundances .
the origin of pah s also remains an unanswered question .
most believe that pah s are formed in the atmospheres of carbon stars or in supernovae @xcite .
various observations show that the abundance of small free - flying pah s have a tendency to decrease in dense clouds where uv photons do not penetrate .
proposed that pah s form larger pah s or pah - clusters in such dense regions . as a consequence ,
free - flying pah s observed at the border of illuminated clouds may be the result of photodissociation of these larger species .
although they determined a minimal size of 400 carbons for these species using mid - infrared isocam maps of photodissociation regions ( pdr s ) , this number is very uncertain .
if free - flying pah s are not present in the dense interstellar medium , an efficient process is needed to aggregate them .
aggregation is poorly understood , and the time in which this process occurs and the fraction of small pah s remaining in dense clouds unexposed to intense radiation are undetermined .
currently , constraints on the pah size and abundance as a function of the physical conditions in the dense interstellar medium remain weak . in this article , we focus not on the size of pah s but on the chemical impact of their presence in the dense cold interstellar cores .
nevertheless , the sensitivity of our results to size will be examined .
we follow past authors by lumping all pah s into three groups of species : neutral , positively charged , and negatively charged pah s , designated pah , pah@xmath1 , and pah@xmath0 respectively . of these three groupings ,
pah@xmath1 species were found to be unimportant and were excluded from the detailed calculations .
the chemical effect of pah s at steady - state has been studied previously . in the diffuse
interstellar medium , the inclusion of pah s , among other effects , enhances the neutralization of atomic ions such as carbon , silicon and sulfur , reducing their abundances @xcite . in their study of dense cloud chemistry
, @xcite concluded that for pah s consisting of molecules with 30 - 50 carbon atoms , with a fractional abundances greater than @xmath2 , pah@xmath3 is the main source of negative charge , which considerably changes the abundances of some neutral species , enhancing , for example , the abundances of carbon - bearing molecules .
finally , @xcite studied the importance of pah s for c - shock models . in this article
, we report a time - dependent study of the role of pah s in the gas - phase chemistry of cold dense clouds and compare our results with observations .
one focus of our study will be how the presence of pah s affects the choice of elemental abundances .
the elemental abundances typically used for chemical models of dense clouds are referred to by the appellation `` low - metal '' .
originally defined by @xcite , these abundances are based on the _ copernicus _ satellite observations of the @xmath4 ophiuchi diffuse region @xcite .
but , in order to reproduce the observation of molecules in dense clouds better , the @xmath4 oph abundances of all the elements but , he , c , n and o are decreased by two orders of magnitude .
this assumption implies that a strong depletion of the heavier elements occurs between the diffuse and dense cloud phases @xcite .
although the details of this depletion are not understood , the depleted abundances are necessary to reproduce the abundances of molecular ions and , to a lesser extent , carbon - bearing species .
since the presence of pah s in dense clouds should modify the ionization fraction , it may help in solving the problem of the choice of elemental abundances .
this article is organized as follows . in sect .
[ pah_model ] , we describe the parameters of the pah s used for the analysis and the chemical processes involving them . in sect .
[ model_other ] , other details of the chemical model are given .
the influence of pah s on the abundances of smaller species is reported in the next section , while the sensitivity of the results to the pah parameters is discussed in sect .
[ sensitivity ] . in sect . 6
, we compare observations in molecular clouds with grids of models in which we vary several parameters , including pah size and abundance .
the paper ends with our conclusions .
@xcite reproduced the extinction curves of various diffuse regions of the milky way from _ spitzer _ observations and constrained the size distribution of carbonaceous grains including pah s down to 3@xmath5 in size ( see fig .
6 of * ? ? ?
using the model @xmath6 of @xcite , which has the largest abundance of small pah s , we integrated over a radius range of 3.8 to 6.0@xmath5 on radius @xmath7 for pah s is not well - known since it depends on the structure of the molecule .
we used the relation given by @xcite : @xmath8 . the radius range here then corresponds to pah s with between 25 and 100 carbon atoms . ] .
the integration yields a mean radius of 4 @xmath5 ( @xmath9 ) and an integrated fractional abundance with respect to @xmath10 of @xmath11 . multiplying by 30 leads to a fractional abundance of c atoms in pah s of @xmath12 , or roughly 3% of the elemental carbon abundance .
because pah s may not be free in dense clouds , we decided to also consider pah - clusters of radius 10 and 100 @xmath5 . with the formula of @xcite
, these limits correspond to pah s with 468 and @xmath13 carbon atoms , leading to fractional abundances of @xmath14 and @xmath15 , respectively , if one assumes ( a ) the same atomic carbon abundance in pah s as for the smaller species , and ( b ) that weakly - bound clusters can also be approximated by the same relationship between number of carbon atoms and size .
@xcite and found that the abundance of pah@xmath16 decreases strongly in clouds with an a@xmath17 larger than 2 and @xcite found that the abundance of pah@xmath16 is smaller than the abundance of the other forms of pah s by about one order of magnitude in their calculations .
for this reason , in our study of dense cloud chemistry , we did not originally include cations but only neutral and negatively - charged pah s .
prodded by the referee , however , we looked more carefully at the role of pah@xmath16 , but concluded again that their exclusion is justified .
the reconsideration will be discussed in the next section .
we start the chemistry with neutral pah s only , based on the results of @xcite and .
neutral pah s become negatively charged very rapidly and our results after a very small time interval do not depend on whether the chemistry commences with pah or pah@xmath3 .
the main problem of including pah s in chemical models is that few experimental measurements of reaction rates have been performed on these molecules , and these few tend to involve rather small cationic pah s only .
for example , @xcite studied the rates of the reactions between a variety of positively - charged pah s with six , ten , and sixteen carbon atoms and the neutrals h , o , n , and h@xmath18 , while @xcite studied reactions of coronene cation ( c@xmath19h@xmath20 ) with the same reactants .
although association reactions appear to be dominant , most the experiments were done at sufficiently high densities that the mechanism , three - body association , is not what occurs in the low - density interstellar medium . moreover , cationic pah s are not important in our study .
the processes in our model include electron attachment , recombination of negative pah ions with smaller positive ions , and photodetachment of negative pah ions by uv radiation .
we consider these in turn .
negative pah ions are formed by radiative electron attachment : @xmath21 the rate coefficient for this reaction is thought to depend on the size of the pah .
and @xcite proposed two different relations : @xmath22 @xmath23 s@xmath24 and @xmath25 @xmath23 s@xmath24 .
these two formulae lead to differences of about a factor of 2 for pah s with @xmath26 .
we arbitrarily decided to follow @xcite and use the first formula .
the rate coefficient is @xmath27 @xmath23 s@xmath24 for the attachment of electrons to neutral pah s with 30 carbons .
the results depend on this value only if @xmath28 is decreased by a factor of 100 . in other words ,
an increase of this rate coefficient does not change the results .
calculations by herbst ( in preparation ) based on the statistical approach of @xcite indicate that with typical electron affinities in the range of 0.5 - 1.0 ev , the large rate coefficients deduced by previous authors are valid for neutral pah s with 30 carbon atoms , but not for those significantly smaller , in agreement with the experimental results of @xcite .
negatively - charged pah s can recombine with positive ions , designated x@xmath16 .
these recombination reactions differ from dissociative recombination of positive ions with electrons in two important ways @xcite .
first , they are less exothermic because of the electron affinity of the pah .
secondly , some of the reaction exothermicity can be deposited in internal energy of the pah so that less is available to dissociate the positive ion upon neutralization .
we assume that two resulting channels are possible for these recombinations , simple charge exchange : @xmath29 and , if the positive ion contains hydrogen : @xmath30 where x designates a neutral with one fewer h atom than the ion .
the terms `` neutralization '' and `` dissociative neutralization '' are used for these two processes to distinguish them from recombination with electrons .
we assume equal branching fractions for the two sets of products , if they exist .
if the neutralization leads to a neutral species that does not exist or is not included in the model , only the second channel is considered .
the products of the dissociation are very uncertain , and laboratory experiments are required .
the rate coefficients for these reactions have been computed by taking into account the coulomb attraction between the charged species via the expression @xcite : @xmath31 where @xmath32 is the electronic charge , @xmath33 is the boltzmann constant , @xmath7 is the radius of the pah ( or pah cluster ) , and @xmath34 is the mean relative velocity between pah@xmath3 and ions , which is given by the expression @xmath35 with @xmath36 the temperature and @xmath37 the reduced mass , which is mainly determined by the mass of x@xmath16 . in expression ( [ k_grain_mol ] ) , we have assumed that the radius of the pah is the critical radius for which reaction can occur .
the radius at which the second term equals the first term on the right - hand - side is approximately 1 @xmath37 m at 10 k. this radius is much larger than the pah radii considered here , so that @xmath38 .
the rate coefficient at 10 k is approximately @xmath39 @xmath40 s@xmath24 for an ion of 30 amu and a pah of 4 @xmath5 .
relation ( [ k_grain_mol ] ) is based on the fact that , due to the delocalization of the negative charge on the pah , the trajectory of the positive ion as it approaches the pah will always be affected by this charge .
taking into account the complex shape and large number of possible pah clusters in dense clouds , several negative charges may stick , and the attraction between such a system and a positive ion may depend on the localization of the electrons in the cluster . as a consequence ,
expression ( [ k_grain_mol ] ) may be more likely to be correct for small pah s of 4@xmath5 radius than for larger ones .
for this reason , we consider the alternative possibility that rate coefficients for the reactions between @xmath41 and x@xmath16 are given by the simple ballistic relation @xmath42 for small pah s , the rate coefficient varies as @xmath7 whereas for large pah s , if the process is ballistic , it varies as @xmath43 .
for a 100 @xmath5 pah and an ion of 30 amu , the rate coefficient at 10 k is then approximately @xmath44 @xmath40 s@xmath24 , or 0.1 of the value for the smaller pah s .
in addition to these reactions , pah@xmath3 is neutralized by photodetachment of electrons due to ultraviolet , visible , and even near infra - red photons : @xmath45 following @xcite , we assume a cross section of 10@xmath46 @xmath47 significantly above threshold , which is the electron affinity ( see also * ? ? ? * ) . for a standard interstellar radiation field
, this leads to a conservative estimate for the rate coefficient in the absence of shielding of approximately 10@xmath48 s@xmath24 , neglecting differences in electron affinities among pah s . a crude estimate of the dependence on @xmath49 is given by the simple exponent @xmath50 .
even with an @xmath49 of 10 , the rate is not negligible in dense clouds . indeed , under most conditions , photodetachment is more rapid than neutralization , but it must be remembered also that the electrons produced in photodetachment then attach to neutral pah .
we do not consider photodetachment by cosmic ray - induced photons .
if we assume a rate coefficient of @xmath51 for this process @xcite , the photodetachment by direct uv photons is then more efficient by one order of magnitude even at an @xmath52 of 10 . in our original calculations
, we did not include pah@xmath1 .
in dense clouds , positively charged pah s can be produced by secondary uv photons induced by cosmic - rays and by charge transfer with small positive ions x@xmath1 .
the latter process can neutralize the x@xmath1 abundance and so affect the small - molecule chemistry
. then pah@xmath16 can be destroyed by recombination with free electrons . as suggested by the referee , we tested the importance of positively - charged pah species using rate coefficients from @xcite for pah@xmath16 production by photoionization and for recombination with electrons . for the charge - transfer reactions , we used rate coefficients computed with the langevin theory and the polarizability from .
as an example , the couloub - based rate coefficient for the recombination reaction between pah@xmath3 and x@xmath53 is @xmath54 @xmath40 s@xmath24 for pah s of 4 @xmath5 and an ion of 30 amu whereas the rate coefficient for the charge - transfer reaction between pah and x@xmath16 for the same ion is @xmath55 @xmath40 s@xmath24 . in the various models that we ran , we found that the abundance of pah@xmath16 is at least two orders of magnitude smaller than the pah and pah@xmath3 abundances .
moreover , its inclusion does not affect any of the results discussed below .
this conclusion holds even when the recombination rate coefficient is the ballistic one .
we utilize a pseudo - time - dependent gas - phase model , which computes the chemical evolution at a fixed temperature and density .
a version of this model ( nahoon ) is now available online at + ` http://www.obs.u-bordeaux1.fr/radio/vwakelam/valentine\%20wakelam/downloads.html ` .
the chemical database is a recent version of the osu network : ` osu_01_2007 ` , which can be found online at ` http://www.physics.ohio-state.edu/~eric/research.html ` .
the reactions involving pah s will also be available at the same address .
we use three different sets of elemental abundances ( ea1 , ea2 and ea3 ) , which are listed in table 1 with respect to the total nuclear hydrogen abundance along with the initial choice of ionization .
ea1 represents the low - metal abundances discussed earlier .
ea2 is based on the high - metal abundances observed in the @xmath4 ophiuchi diffuse cloud , but is modified based on recent observations .
the abundance of helium has not been observed in cold cores and we use a value of 0.09 for ea2 with respect to hydrogen , based on observations in the orion nebula @xcite . to account for the oxygen and carbon that are depleted on grain mantles in the form of h@xmath56o and
co , we removed 2.4% of c and 13% of o .
it is possible that the metals are more depleted in dense clouds than in diffuse regions .
thus , for ea3 , we assumed that the silicon and magnesium are totally depleted on grains and we assumed an abundance of @xmath57 for iron as suggested by @xcite .
the complete depletion of si is one explanation for the non - detection of sio in dense clouds ( see * ? ? ?
we also considered an additional depletion of a factor of 10 for chlorine as suggested by the observation of hcl in cold material around omc-1 @xcite .
note that in all the elemental abundances that we considered , the elemental c / o ratio is 0.4 .
the species are assumed to be initially in atomic form except for hydrogen , which is assumed to be entirely in its molecular form .
we consider typical physical conditions for dense clouds : a temperature of 10 k , an h@xmath56 density of @xmath58 @xmath59 , a cosmic - ray ionization rate @xmath4 of @xmath60 s@xmath24 and a visual extinction @xmath49 of 10 .
in this section , we present results for the three sets of elemental abundances ( ea1 , ea2 and ea3 ) with and without pah s . in
the models with pah s , referred to as ea / pah , we assume their fractional abundance to be @xmath11 , our standard value for small pah s .
variations in this value will be considered in sect .
[ sensitivity ] and [ comp_obs ] .
figure [ fract_ion ] shows the ionization fraction ( sum of the negatively or positively charged species ) as a function of time for the three sets of elemental abundances , each with and without pah s .
the initial ionization is typical of a diffuse cloud since we start with ionized carbon in all models . with low - metal elemental abundances ( ea1 ) and no pah s
, the ionization fraction stays high for only @xmath61 yr and then decreases sharply , reaching a steady - state value of @xmath62 by 10@xmath63 yr . with both ea2 and ea3 , and no pah s ,
the ionization fraction decreases slowly , remaining larger than @xmath64 for about @xmath58 yr and eventually reaching a steady - state value of a few @xmath65 .
when pah s are added , the ionization fraction does not depend significantly on the elemental abundances .
the main effect is to sharply decrease the earlier ionization fraction of the gas : within the first 300 yr , the ionization fraction goes from @xmath66 to @xmath67 , its steady - state value , which is only slightly larger than that achieved with the ea1 model .
thus , the inclusion of pah s reduces the ionization fraction for ea2 and ea3 to a value close to that achieved with low - metal abundances and no pah s . the steady - state ionization fraction for model ea1/pah is larger by a factor of 2 compared with model ea1 without pah s .
the ionization fraction of model ea2/pah at steady - state can be compared with the very similar value of @xmath68 reported by @xcite for a ratio of total pah to hydrogen of @xmath65 and high - metal abundances .
figure [ e_pah ] depicts the fractional abundances of neutral and negatively - charged pah s as well as the electron abundance as functions of time for model ea2/pah .
the other pah models are not shown because these abundances do not depend significantly on the chosen elemental abundances . by 300 yr , the fractional abundances of electrons and pah@xmath3 drop to @xmath69 and @xmath70 respectively , so that pah@xmath3 becomes the dominant negatively - charged species .
the conversion of neutral to negatively - charged pah s occurs via electron attachment , which is very efficient at early times because of the high electron abundance .
but , once formed , pah@xmath3 is efficiently destroyed by reactions with atomic ions , which are also very abundant at the beginning of the chemical evolution .
note that the recombination of atomic ions is much more efficient with pah s than with electrons since the latter must occur radiatively .
for instance , the neutralization of c@xmath16 by pah@xmath3 ( with @xmath71 ) is four orders of magnitude larger than by electrons at 10 k. because of this efficient neutralization , the ionization fraction of the gas decreases much more quickly when pah s are present ( see fig .
[ fract_ion ] ) .
we also find that the abundance of neutral pah s exceeds that of the anionic pah s by a factor of 5 - 10 after 300 yr , which is also in agreement with the steady - state results of @xcite . in the absence of pah s , the higher - metal elemental abundances ea2 and ea3 lead to a very different chemistry from ea1 , and tend to worsen the agreement between the modeling results and the observations for selected classes of molecules @xcite . with models ea2/pah and ea3/pah , however , the chemistry is much closer to that computed with the low metal elemental abundances in the absence of pah s .
the major change in the chemistry is caused by the high abundance of pah@xmath0 , which neutralizes atomic positive ions efficiently and , upon reaction with positive molecular ions , tends to neutralize them without significant dissociation , contrary to dissociative recombination reactions with electrons .
in addition , the decrease of the ionization fraction at initial stages of the chemistry increases the abundance of ionic species such as h@xmath72 and h@xmath73o@xmath16 because they are depleted more slowly . at later times , some of the c- , s- , n - bearing species are enhanced in the models ea2/pah and ea3/pah because of i ) larger elemental abundances and ii ) a smaller electron abundance .
we now present the results for the most important species and for selected molecules that show significant differences among models . in figure [ c_fam ] , we show our results for c and c@xmath16 as functions of time for the three elemental abundances with and without pah s . as expected , the fractional abundance of ionized carbon is decreased most rapidly when the pah s are included .
the agreement at steady - state for the model ea2/pah is in reasonable agreement with the result of @xcite .
a consequence of the c@xmath16 neutralization by pah@xmath3 is that atomic carbon becomes the main reservoir of carbon earlier than with the models without pah s . at the age of dense clouds , between @xmath74 and @xmath75 yr
, atomic carbon is more abundant in ea2 and ea3 compared with ea1 because of a higher elemental abundance of carbon .
our steady - state fractional abundance of neutral carbon ( @xmath76 ) in model ea2/pah is three orders of magnitude below that of the analogous model in @xcite .
after early time , co becomes the main repository of carbon in all models ; the co abundance ( not shown in the figure ) is insensitive to the model used .
the rapid decrease of the electron abundance in the models with pah s leads to strong increases in the abundances of molecular ions such as h@xmath72 , h@xmath73o@xmath16 and hco@xmath16 and neutral species formed from them at the initial stage of the chemistry , as can be seen in fig .
[ oh_fam ] . at longer times
, the ionic abundances show much less time dependence , with two general but not universal patterns : ( i ) they are more abundant in the models with pah s than in the analogous models without them , and ( ii ) they are most abundant in models with ea1 abundances and least abundant in models with ea2 abundances , with and without pah s .
general reasons for the trends include the higher electron abundances in models ea2 and ea3 , the slower depletion rates by negatively charged species ( mainly pah@xmath0 ) in the pah - containing models , and differences in abundances of important neutral species involved in depletion processes among the pah - containing models . the ion h@xmath77 is a partial exception to the trends in that its abundance is not enhanced by pah s at later times , mainly because it is destroyed principally by heavy neutral species such as co and h@xmath18o rather than by charged species .
the abundance in ea2 and ea3 is smaller than in ea1 because of dissociative recombination by more abundant electrons whereas it is smaller in ea2/pah and ea3/pah than in ea1/pah because of reaction with h@xmath56o , which is more abundant in the high - metal models .
the enhancement in the ion hco@xmath1 for models with pah s in the initial stages of the chemistry is even more dramatic than that of h@xmath77 because of its formation from abundant carbon atoms , by reactions leading first to ch@xmath78 . at later times
, hco@xmath16 is formed by the reaction between h@xmath72 and co in all models . without pah s ,
hco@xmath16 is destroyed by electrons whereas it is destroyed by h@xmath56o in models with pah s .
the abundance of h@xmath79o@xmath1 is also dramatically enhanced in the initial stages of the chemistry in the presence of pah s . without pah s ,
h@xmath73o@xmath16 is destroyed by electrons whereas it is destroyed by pah@xmath3 in ea1/pah and ea3/pah . in ea2/pah ,
due to a large abundance of elemental silicon , h@xmath73o@xmath16 reacts with gas phase sio , which is as abundant as @xmath80 at @xmath75 yr .
such a large abundance of sio is in disagreement with observations in dense clouds @xcite .
one consequence of the large h@xmath73o@xmath16 abundance , at initial stages of the chemistry , is that the abundances of oh , h@xmath56o and o@xmath56 , which stem from the recombination and neutralization of h@xmath79o@xmath1 , are also enhanced in pah - containing models .
the situation is somewhat more complex at later times with only water showing a much higher abundance in pah - containing models .
the predicted enhancement of h@xmath56o in pah - containing models is a problem because the calculated abundances far exceed the observational upper limits of @xmath81 in l134n and @xmath82 in tmc-1 @xcite .
the result for molecular oxygen is , on the contrary , not affected by pah s at times after 10@xmath63 yr .
our theoretical o@xmath56 abundance is about @xmath69 at @xmath74 yr using ea2/pah or ea3/pah , which compares favorably with the observational upper limits of @xmath83 in l134n and @xmath84 in tmc-1 .
the computed abundances of the hydrocarbons c - c@xmath73h@xmath56 ( cyclopropenylidene ) , c@xmath73h@xmath85 ( methyl acetylene ) , c@xmath86h@xmath56 ( triacetylene ) and ch@xmath73c@xmath85h ( methyl diacetylene ) are shown in fig .
[ hydro_fam ] as functions of time . without pah s , the abundances peak at so - called early - times ( 10@xmath63 yr ) for low - metal abundances ( ea1 ) , a well - known effect . for ea2 and ea3 , on the other hand , the early - time peaks are weak to non - existent and the peak abundances very low , an effect caused by the high electron abundances , which disrupt the growth of molecular ions . when pah s are added , the abundances of the four hydrocarbons depicted are enhanced for all three models .
although the enhancement is significantly greater for ea2 and ea3 , the peak ea1 values are still the largest .
the enhancements are caused mainly by the lesser destructive effect of pah@xmath0 compared with electrons when it reacts with intermediate ions in the chain of synthetic reactions , although in individual cases secondary effects can be isolated .
the major effect , which comprises both lower rates and larger product neutral channels , is stronger than the enhancement in destruction rates of hydrocarbons due to often larger abundances of molecular ions .
that differences remain among the pah models despite the fact that pah@xmath0 has the same abundance shows , once again , that other processes are also important .
the computed abundances of hydrocarbons at steady - state can be very different from the results of @xcite .
for example , the c@xmath73h@xmath56 abundance is @xmath87 in model ea2/pah whereas it is @xmath88 in @xcite
. this difference may be due to discrepancies in the network for carbon chemistry . as can be seen in fig .
[ ch_fam ] , the abundances of the oxygen - containing organic neutral molecules hcooh , ch@xmath56co , ch@xmath73oh and ch@xmath73cho in models containing pah s are enhanced at almost all times by at least one order of magnitude over the abundances in models without pah s . once again , the reason is generally the same as for the hydrocarbons : neutralization reactions between precursor ions with pah@xmath0 are not as destructive as dissociative recombination reactions with electrons .
even if a species , such as ketene , is formed by a neutral - neutral reaction , its precursor is likely formed at least partially through ion - molecule channels .
an as example , consider the case of methanol .
methanol ( ch@xmath79oh ) is produced through the ion ch@xmath79oh@xmath89 , which itself is formed via the radiative association between ch@xmath77 and water . in models without pah s ,
the protonated ion can recombine dissociatively with electrons to form ch@xmath79oh + h , although this channel is now known to be a minor one @xcite .
with pah@xmath0 , on the other hand , we assume that this channel is the only one , since the neutral form ch@xmath79oh@xmath18 does not exist and we only allow one hydrogen to be ejected from the species ( see section [ pah_chem ] ) .
another reason for the enhanced abundance of methanol lies in the enhanced abundance of water .
we note that with pah s , the peak predicted fractional abundance of methanol is still 1 - 2 orders of magnitude below the typical value in cold dense clouds of 10@xmath90 @xcite . unlike the situation with the hydrocarbons , the abundances in models ea2/pah and ea3/pah are not significantly less and sometimes even greater than those in model ea1/pah .
the small differences suggest that pah@xmath0 plays an even more dominant role than in the case of the hydrocarbons . in the non - pah models , although the ea1 abundances tend to be greater than those in ea2 and ea3 , the effect is smaller than for hydrocarbons except for acetaldehyde .
cyanopolyynes are very abundant towards the famous cyanopolyyne peak ( cp ) of the cold core tmc1 , yet their abundances are under produced by gas - phase models unless a high c / o elemental abundance ratio is assumed @xcite . as shown in fig .
[ hcn_fam ] , the pah s have a strong enhancing effect on the formation of these molecules , especially on the more complex cyanopolyynes . once again , there is little difference among the results for the three pah models .
in general , the dominant formation mechanism for each cyanopolyyne more complex than hc@xmath79n involves the next smaller member of the series . at @xmath74
yr , the hc@xmath91n , hc@xmath92n and hc@xmath93n abundances are enhanced by more than three orders of magnitude over their values without pah s .
the peak fractional abundances are indeed larger than the values observed in tmc1 ( cp ) of @xmath94 , @xmath95 and @xmath96 for hc@xmath91n , hc@xmath92n and hc@xmath93n respectively ( see * ? ? ?
* and references therein ) .
similar enhancements are seen for other nitrogen - containing molecules such as ch@xmath18chcn , ch@xmath73cn , ch@xmath73c@xmath73n and c@xmath73h@xmath56n@xmath16 .
the elemental abundance of sulfur is smaller by more than 2 orders of magnitude in ea1 compared with the two other sets of elemental abundances . as a consequence ,
most of the sulfur - bearing species have larger abundances with high - metal abundances at all times , with and without pah s . in fig . [ s_fam ]
, we follow the temporal evolution of four species : ocs , h@xmath56s , cs and c@xmath73s .
three of these molecules tend to have enhanced peak abundances in the presence of pah s , although the situation is more complex towards steady - state .
the exceptional cases are cs for all elemental abundances and h@xmath18s for the low - metal case .
the cs molecule is produced by the dissociative recombination of hcs@xmath16 and hc@xmath56h@xmath16 by electrons or dissociative neutralization of hcs@xmath16 by pah@xmath3 .
the larger ion hc@xmath56h@xmath16 does not react with pah@xmath0 to produce cs with our assumptions . as a consequence ,
the abundance of cs is smaller in the models including the pah s .
the late - time situation regarding h@xmath18s is more complex .
the primary formation mechanism for h@xmath18s at most times is through the protonated precursor h@xmath79s@xmath1 , which is formed by a chain of reactions starting with s + h@xmath77 and proceeding through hs@xmath1 and h@xmath18s@xmath1 . without pah s ,
h@xmath56s is also produced by the charge exchange between h@xmath56s@xmath16 and s. the ion h@xmath56s@xmath16 is more abundant in the low - metal case without pah s because it is produced by the reaction @xmath97 where s@xmath16 is more abundant than the neutral form in the absence of pah s .
there are some observational and theoretical indications that pah s in dense clouds may be aggregated into clusters with radius up to 100 @xmath5 .
we have performed additional calculations using the three elemental abundances and pah - clusters of radii 10 @xmath5 and 100 @xmath5 with the reduced abundances discussed in sect .
[ pah_model ] . in our discussion of the results , we use the word pah
whatever the size . for the larger pah s ( 10 and 100 @xmath5 )
, we consider two expressions for the rate coefficients of the neutralization reactions between pah@xmath0 and positive ions : one involving charge attraction , as expressed in equation ( [ k_grain_mol ] ) , and the other a simple ballistic one , as expressed in equation ( [ k_ball ] ) .
the ballistic rate is considerably lower .
for radii of 10 and 100 @xmath5 and a temperature of 10 k , the ballistic rate is , respectively , about 1000 @xmath98 and 10 @xmath98 smaller than the rate involving charge attraction .
another process with a rate coefficient that depends on the pah size is the radiative attachment of electrons to neutral pah s , which increases with increasing radius . in fig .
[ fract_othermod ] , we show the ionization fraction as a function of time with each set of elemental abundances and pah s of 4 , 10 , and 100 @xmath5 . as can be seen , the ionization fraction depends highly on the size of the pah , and on whether or not charge attraction is considered .
we do not show the dependence on the total abundance of c atoms incorporated into pah s , but a larger abundance implies a more rapid decrease of the ionization fraction to a smaller value at late times . in the left panel of fig .
[ fract_othermod ] , neutralization proceeds with charge attraction for all three sizes of pah s .
for the two smaller sizes of pah s , the main negatively - charged species in the model remains pah@xmath3 .
for the pah s of 100 @xmath5 radius , the total abundance of pah s is @xmath15 , much smaller than the abundance of electrons .
the evolution of the ionization fraction is then closer to the case without pah s , but the steady - state ionization fraction can still be somewhat lower . for the case with model ea1 and the largest pah s , the ionization fraction is the same with or without charge attraction and is not different from the model in which there are no pah s at all . as a consequence ,
the 10 @xmath5 and ( for ea2 and ea3 ) the 100 @xmath5 cases represent intermediate stages between the models with 4 @xmath5 pah s and those without pah s .
the right - hand panel of fig .
[ fract_othermod ] shows the results when positive ion - pah@xmath3 neutralization occurs ballistically for the larger pah s . in this case , the 10 @xmath5 and 100 @xmath5 cases give very similar results to the model without pah s so that the effect of large pah s at the chosen abundances is negligible in dense cloud chemistry .
this effect occurs because the major depletion mechanism for pah@xmath0 is photodetachment rather than recombination with positive ions , so that lowered neutralization rates have the sole effect of decoupling pah@xmath0 from the ion - molecule chemistry .
the modification of the ionization fraction has consequences for the abundances of species . in general , pah s of 100 @xmath5 ( considering charge attraction or not ) and of 10 @xmath5 without charge attraction lead to abundances similar to the models without pah s . for most of the species , pah s of 10 @xmath5 with charge attraction give intermediate abundances between the case of very small 4 @xmath5 pah s and the case without pah s
many species such as c , c@xmath16 h@xmath72 , hco@xmath16 and oh show a difference smaller than a factor of two while using larger pah s of 10 @xmath5 instead of smaller pah s of 4 @xmath5 .
some species are largely underproduced using the elemental abundances ea2 and ea3 without pahs .
this is the case for c@xmath73h@xmath56 , c@xmath73h@xmath85 , c@xmath86h@xmath56 , ch@xmath73c@xmath85h , ch@xmath73cho and cyanoppolyynes . in the presence of 10 @xmath5 pah s , these species are almost as abundant as with smaller pah s .
the objective of this section is to study the impact of pah s on the comparison with observations in the two cold cores l134n ( north peak ) and tmc-1 ( cyanopolyyne peak ) .
these two clouds are known to show differences in the abundances of c - rich molecules that are sometimes explained by differences in age or c / o ratio . here
we explore the possibility that they are also explained by the amount and size of pah s present .
the observed abundances in l134n ( north peak ) and tmc-1 ( cyanopolyyne peak ) are listed in tables 3 and 4 of @xcite .
in addition to these species , comprising 42 molecules for l134n and 53 for tmc-1 , we considered sio , for which only upper limits of @xmath99 for l134n and @xmath100 for tmc1 have been determined @xcite . in this comparison , we utilize the three different sets of elemental abundances listed in table 1 .
of all these elements , sulfur is the least constrained by abundances of gas - phase molecules gas - phase abundances so that the depleted fraction is uncertain @xcite .
pah size and abundance are also poorly determined parameters .
instead of fixing these parameters to previously used values , we try to give some trends by comparing observations with grids of models in which these three parameters are varied in addition to time .
we computed 441 models by varying the abundance of elemental sulfur between @xmath2 and @xmath101 , the abundance of pah s between @xmath102 and @xmath65 and the size of pah s between 4 and 400 @xmath5 .
note that the largest values of pah sizes and abundances used here are not in agreement with the relation @xmath8 used in the first part of the article .
considering the little information available on the relation of size to number of carbons for pahs , in particular for pah clusters , we consider the pah size and abundance as independent .
the agreement between present models and observations quantified using statistical comparisons , @xmath103 minimization for instance , is not good enough to give reasonable constraints on model parameters .
for this reason , we did not use such methods but instead for each model and each time , we count the number of species with fractional abundance x reproduced by the model within one order of magnitude ( @xmath104 ) where @xmath105 is the observed abundance and @xmath106 the computed abundance . for those species for which only upper limits to their abundances are known and for the case of atomic carbon , for which only a lower limit is known , we check that the predicted abundance is in agreement with the observed limit
when this is the case , we consider that the species is reproduced . figures [ l134n ] and [ tmc1 ] show the maximum percentage of species reproduced in each cloud as a function of the parameters chosen .
column ( a ) shows the level of agreement as a function of pah size and abundance , with each point optimized for the best elemental abundance of sulfur and the best time . column ( b ) shows the level of agreement as a function of the elemental abundance of sulfur and the time optimized at each point for the best pah parameters .
finally , column ( c ) shows the agreement as a function of the elemental abundance of sulfur and the time for models without pah s .
these results are shown for each set of elemental abundances ( 1 : ea1 , 2 : ea2 and 3 : ea3 ) .
best agreement , with or without pah s , is most often achieved at so - called early times of @xmath107 yr . moreover , high levels of agreement can not be achieved with high abundances of pah clusters , as shown in the upper - right corners in column 1 for both figures .
the maximum percentage of molecules reproduced depends on all the parameters , and several solutions exist .
pah s do not improve significantly the best agreement for l134n ( already above 70% without pah s ) but they do expand the range of parameters ( time and elemental abundance of sulfur ) for which we have an agreement larger than 70% , especially if elemental abundances 1 or 2 are utilized .
note particularly the expanded range of agreement with the use of elemental abundances 2 ( the high - metal ones ) .
the optimum pah parameters corresponding to the yellow zones in column b do not depend on the optimum sulfur abundance . as an example
, high agreement is obtained with @xmath108 abundance of 4@xmath5 pah s both for low ( @xmath109 ) and high ( @xmath110 ) elemental abundance of sulfur .
the chemistry in tmc-1 on the contrary is better reproduced if pah s are present , as can be seen immediately by comparison of columns b and c in figure [ tmc1 ] . without pah s , agreement rarely extends to more than 60% , while with pah s , agreement in excess of 70% exists for all three sets of elemental abundances . for elemental abundances 1 ,
the pah parameters are relatively unconstrained ; while for the high - metal abundances ( ea2 ) , high abundances of small pah s are needed .
the situation is less clear for ea3 .
for both clouds , we find isolated areas of agreement larger than 70% at very early times , @xmath111 yr , for large abundances of sulfur ( @xmath112 ) and big ( @xmath113 ) pah s ( @xmath114 ) if ea2 and/or ea3 is utilized .
such agreement is of mathematical interest only .
most chemical studies of the dense interstellar medium rely purely on gas - phase processes . a few contain , in addition ,
a varying amount of surface chemistry on interstellar grains but the inclusion of pah s , which occupy a middle ground in size between small gas - phase species and dust particles , is quite rare .
part of the reason for this neglect is that to include pah s properly , one should view them as individual species .
but the actual individual species that make up the pah content of interstellar clouds are not well constrained from spectral observations , and even their chemistry has not been studied in great detail .
so , inclusion of pah s means that one must treat them as small grains , and follow just their charge state . in this study , as in previous ones , we lump all pah s into just three groups , depending on whether they are neutral , positively - charged , or negatively - charged .
positively - charged pah s were initially assumed to be unimportant in cold dense sources , an assumption that was confirmed upon the request of the referee .
they were not included in the chemical network used for the results presented here . by including an assortment of chemical processes involving negatively - charged and neutral pah s into our standard gas - phase network of reactions , we are able to determine the effect of pah s on the time - dependent gas - phase chemistry of cold and dense interstellar cloud cores
. the gas - phase model employed is known as ` osu_01_2007 ` , and can be downloaded from the url http://www.physics.ohio-state.edu/@xmath115eric/research.html . to this model
, we add electron attachment to neutral pah s , neutralization between negatively - charged pah s and positively - charged smaller ions , and photodetachment of negatively - charged pah s by external photons .
our standard model contains a fractional abundance of @xmath116 pah s of size 4 @xmath5 , corresponding to 30 carbon atoms .
in addition to this model , we vary the size and abundance of the pah s to study the impact of these parameters on the chemistry .
we also use three different sets of elemental abundances with different abundances of heavy atoms . in general , the dominant effect in our calculations
is that electrons are replaced by pah@xmath0 as the dominant negative charge carrier , an effect which weakens and eventually vanishes as the size of the pah s , due probably to clustering , and their fractional abundance decrease .
when the abundance of pah@xmath0 exceeds that of electrons , the overall ionization fraction is reduced because atomic positive ions are far more rapidly neutralized by reactions with pah@xmath0 than by radiative recombination with electrons .
these changes lead to many secondary effects .
our calculations show that many of the molecular species have abundances that are strongly affected by the presence of pah s .
in particular , molecular ions tend to have increased abundances at most times and large unsaturated neutrals , such as the polyynes , as well as saturated species such as methanol , show very large increases in abundance .
the sensitivity of our results to variations in the size of pah s depends upon assumptions made in the rate of neutralization of pah@xmath0 and positive ions .
the differences among results obtained with different elemental abundances are much smaller in models containing pah s than in models without them . by comparing observations in the two dense cloud cores l134n and
tmc-1 with grids of models , we showed only that the chemistry in such sources is not consistent with large amounts of large pah clusters .
we are not able to give more severe unambiguous constraints on the pah parameters , since high levels of agreement can exist for large parameter variations .
the situation is similar for the elemental abundance of sulfur , where values over several orders of magnitude are consistent with maximum agreement with observations .
as regards time , both the standard early - time and very early times can lead to optimum agreement , although the latter must be rejected on astronomical grounds .
the situations for l134n and for tmc-1 show differences . for l134n
, the best agreement is not improved when pah s are included , although the range of optimum sulfur elemental abundances and times are extended , most significantly so for the high - metal abundances ( ea 2 ) . for tmc-1 ,
on the other hand , the best level of agreement is significantly better when pah s are included , using any of the three elemental abundances .
this difference may indicate a difference in pah properties between the two regions .
in particular , for ea 2 , optimum agreement in tmc-1 is obtained with smaller and more numerous pah s , while for l134n , such parameters make matters worse . finally , unlike the case of gas - phase models without pah s , models including pah s are able to accommodate high - metal elemental abundances , similar to those observed in the diffuse medium , if the effects of pah s are included and optimized . ,
l. , olofsson , a. o. h. , bergman , p. , bernath , p. , black , j. h. , booth , r. s. , buat , v. , crovisier , j. , curry , c. l. , encrenaz , p. j. , falgarone , e. , feldman , p. a. , fich , m. , floren , h. g. , frisk , u. , gerin , m. , gregersen , e. m. , harju , j. , hasegawa , t. , hjalmarson , . , johansson , l. e. b. , kwok , s. , larsson , b. , lecacheux , a. , liljestr " om , t. , lindqvist , m. , liseau , r. , mattila , k. , mitchell , g. f. , nordh , l. h. , olberg , m. , olofsson , g. , ristorcelli , i. , sandqvist , a. , von scheele , f. , serra , g. , tothill , n. f. , volk , k. , wiklind , t. , & wilson , c. d. 2003 , a&a , 402 , l77 , r. l. , howe , j. e. , ashby , m. l. n. , bergin , e. a. , chin , g. , erickson , n. r. , goldsmith , p. f. , harwit , m. , kleiner , s. c. , koch , d. g. , neufeld , d. a. , patten , b. m. , plume , r. , schieder , r. , stauffer , j. r. , tolls , v. , wang , z. , winnewisser , g. , zhang , y. f. , & melnick , g. j. 2000 , apj , 539 , l101 l|l|l|l|l species & ea1 & ea2 & ea3 & reference for ea2 + he & 1.40(-1 ) & 9.00(-2 ) & 9.00(-2 ) & see text + n & 2.14(-5 ) & 7.60(-5 ) & 7.60(-5 ) & @xcite + o & 1.76(-4 ) & 2.56(-4 ) & 2.56(-4 ) & @xcite + c@xmath16 & 7.30(-5 ) & 1.20(-4 ) & 1.20(-4 ) & @xcite + s@xmath16 & 8.00(-8 ) & 1.50(-5 ) & 1.50(-5 ) & @xcite + si@xmath16 & 8.00(-9 ) & 1.70(-6 ) & 0.00 & @xcite + fe@xmath16 & 3.00(-9 ) & 2.00(-7 ) & 1.50(-8 ) & @xcite + na@xmath16 & 2.00(-9 ) & 2.00(-7 ) & 2.00(-7 ) & @xcite + mg@xmath16 & 7.00(-9 ) & 2.40(-6 ) & 0.00 & @xcite + cl@xmath16 & 1.00(-9 ) & 1.8(-7 ) & 1.80(-8 ) & @xcite + p@xmath16 & 2.00(-10 ) & 1.17(-7 ) & 1.17(-7 ) & @xcite + f@xmath16 & 6.68(-9 ) & 1.8(-8 ) & 6.68(-9 ) & @xcite + | virtually all detailed gas - phase models of the chemistry of dense interstellar clouds exclude polycyclic aromatic hydrocarbons ( pah s ) .
this omission is unfortunate because from the few studies that have been done on the subject , it is known that the inclusion of pah s can affect the gas - phase chemistry strongly .
we have added pah s to our network to determine the role they play in the chemistry of cold dense cores .
initially , only the chemistry of neutral and negatively - charged pah species was considered , since it was assumed that positively - charged pah s are of little importance .
subsequently , this assumption was checked and confirmed . in the models presented here ,
we include radiative attachment to form pah@xmath0 , mutual neutralization between pah anions and small positively - charged ions , and photodetachment .
we also test the sensitivity of our results to changes in the size and abundance of the pah s .
our results confirm that the inclusion of pah s changes many of the calculated abundances of smaller species considerably . in tmc-1 ,
the general agreement with observations is significantly improved contrary to l134n .
this may indicate a difference in pah properties between the two regions . with the inclusion of pah s in dense cloud chemistry ,
high - metal elemental abundances give a satisfactory agreement with observations . as a result
, we do not need to decrease the observed elemental abundances of all metals and we do not need to vary the elemental c / o ratio in order to produce large abundances of carbon species in tmc-1 ( cp ) . |
all graphs in this paper are simple , finite and undirected . for notation and terminology that are not defined here
, we refer to west @xcite .
let @xmath13 , @xmath14 , @xmath15 denote the induced path , induced cycle and complete graph on @xmath16 vertices respectively .
let @xmath17 be the complete bipartite graph with classes of size @xmath18 and @xmath19 .
if @xmath20 is a family of graphs , a graph @xmath7 is said to be _
@xmath20-free _ if it contains no induced subgraph isomorphic to any member of @xmath20 .
if @xmath21 and @xmath22 are two vertex disjoint graphs , then their _ union _
@xmath23 is the graph with @xmath24 @xmath25 and @xmath26 @xmath27 @xmath28 .
similarly , their _ join _
@xmath29 is the graph with @xmath30 @xmath27 @xmath31 and @xmath32 @xmath27 @xmath33@xmath34 . for any positive integer @xmath35
, @xmath36 denotes the union of @xmath35 graphs each isomorphic to @xmath7 . for a graph @xmath7 , the complement of @xmath7
is denoted by @xmath37 . a _ proper coloring _ ( or simply _ coloring _ ) of a graph @xmath7 is an assignment of colors to the vertices of @xmath7 such that no two adjacent vertices receive the same color .
the minimum number of colors required to color @xmath7 is called the _ chromatic number _ of @xmath7 , and is denoted by @xmath5 .
a _ clique _ in a graph @xmath7 is a set of vertices that are pairwise adjacent in @xmath7 .
the _ clique number _ of @xmath7 , denoted by @xmath6 , is the size of a maximum clique in @xmath7 . obviously , for any graph @xmath7 , we have @xmath38 .
the existence of triangle - free graphs with large chromatic number ( see @xcite for a construction of such graphs ) shows that for a general class of graphs , there is no upper bound on the chromatic number as a function of clique number . a graph @xmath7 is called _ perfect _ if @xmath39 , for every induced subgraph @xmath40 of @xmath7 ; otherwise it is called
_ imperfect_. a hereditary class @xmath0 of graphs is said to be @xmath1-_bounded _
@xcite if there exists a function @xmath2 ( called a @xmath1-_binding function _ of @xmath41 ) such that @xmath3 , for every @xmath4
. if @xmath41 is the class of @xmath40-free graphs for some graph @xmath40 , then @xmath2 is denoted by @xmath42 .
we refer to @xcite for an extensive survey of @xmath1-bounds for various classes of graphs .
the class of @xmath8-free graphs and its related classes have been well studied in various contexts in the literature ; see @xcite . here , we would like to focus on showing @xmath1-binding functions for some classes of graphs related to @xmath8-free graphs .
wagon @xcite showed that the class of @xmath43-free graphs admits an @xmath44 @xmath1-binding function for all @xmath45 .
in particular , he showed that @xmath46 , and the best known lower bound is @xmath47 , where @xmath48 denotes the smallest @xmath35 such that every graph on @xmath35 vertices contains either a clique of size @xmath49 or the complement of the graph contains a @xmath50 @xcite .
this lower bound is non - linear because chung @xcite showed that @xmath51 is at least @xmath52 for some @xmath53 .
it is interesting to note that brause et al .
@xcite showed that the class of ( @xmath10)-free graphs does not admit a linear @xmath1-binding function .
it follows that the class of ( @xmath11)-free graphs , where @xmath40 is any @xmath8-free graph with independence number @xmath54 , does not admit a linear @xmath1-binding function .
here we are interested in classes of @xmath8-free graphs that admit a linear @xmath1-binding function , in particular , some classes of @xmath8-free graphs that admit a ` special ' linear @xmath1-binding function @xmath55 , where @xmath56 is an integer , that is , @xmath8-free graphs @xmath57 such that @xmath58 .
if @xmath59 , then this special upper bound is called the _ vizing bound _ for the chromatic number , and is well studied in the literature ; see @xcite and the references therein .
brause et al .
@xcite showed that if @xmath7 is a connected ( @xmath60)-free graph with independence number @xmath61 , then @xmath7 is perfect .
it follows from a result of @xcite that if @xmath7 is a ( @xmath8 , paw)-free graph , then either @xmath7 is perfect or @xmath62 and @xmath63 ( see also @xcite ) .
nagy and szentmiklssy ( see @xcite ) showed that if @xmath7 is a ( @xmath64)-free graph , then @xmath65 .
blaszik et al .
@xcite and independently gyrfs @xcite showed that if @xmath7 is ( @xmath66)-free graph , then @xmath67 , and the equality holds if and only if @xmath7 is a split - graph .
it follows from a result of @xcite that if @xmath7 is a ( @xmath68)-free graph , then @xmath67 .
fouquet et al .
@xcite showed that if @xmath7 is a ( @xmath69)-free graph , then @xmath70 , and the bound is tight .
brause et al .
@xcite showed that if @xmath7 is a ( @xmath71)-free graph , then @xmath72 . .known
chromatic bounds for ( @xmath11)-free graphs , where @xmath40 is any @xmath8-free graph on @xmath73 vertices with @xmath74 , and the graph @xmath75 . [ cols="<,<,<,<",options="header " , ] in this paper , by using structural results ,
we show that the class of ( @xmath11)-free graphs , where @xmath76 admits a special linear @xmath1-binding function @xmath55 , where @xmath56 is an integer ; see figure [ fig:1 ] .
we also show that the class of ( @xmath77)-free graphs admits a linear @xmath1-binding function .
table 1 shows the known chromatic bounds for a ( @xmath11)-free graph @xmath7 , where @xmath40 is any @xmath8-free graph on @xmath73 vertices with @xmath78 .
some of the cited bounds are consequences of much stronger results available in the literature . finally , we show @xmath1-binding functions for some superclasses of @xmath8-free graphs .
let @xmath7 be a graph , with vertex - set @xmath79 and edge - set @xmath80 . for @xmath81
, @xmath82 denotes the set of all neighbors of @xmath83 in @xmath7 .
for any two disjoint subsets @xmath84 , @xmath85 $ ] denotes the set of edges @xmath86 in @xmath87 .
also , for @xmath88 , let @xmath89 $ ] denotes the subgraph induced by @xmath90 in @xmath7 , and for convenience we simply write @xmath91 $ ] instead of @xmath89 $ ] . note that if @xmath92 and @xmath93 are any two graphs , and if @xmath7 is ( @xmath94)-free , then @xmath37 is ( @xmath95)-free . for any integer @xmath35
, we write @xmath96 $ ] to denote the set @xmath97 . a _ diamond _ or a @xmath98 is the graph with vertex set @xmath99 and edge set @xmath100 .
paw _ is the graph with vertex set @xmath99 and edge set @xmath101 .
see figure [ fig:1 ] for some of the other special graphs used in this paper .
a graph @xmath7 is a _ split graph _ if its vertex set @xmath79 can be partitioned into two sets @xmath102 and @xmath103 such that @xmath102 is a clique and @xmath103 is an independent set . in @xcite , fldes and
hammer showed that a graph @xmath7 is a split graph if and only if @xmath7 is ( @xmath104)-free . a graph @xmath7 is a _ pseudo - split graph _
@xcite if @xmath7 is ( @xmath66)-free .
the class of pseudo - split graphs generalizes the class of split graphs .
a _ @xmath35-clique covering _ of a graph @xmath7 is a partition @xmath105 of @xmath79 such that @xmath106 is a clique , for each @xmath107 .
the _ clique covering number _ of the graph @xmath7 , denoted by @xmath108 , is the minimum integer @xmath35 such that @xmath7 admits a @xmath35-clique covering .
independent / stable _ set in a graph @xmath7 is a set of vertices that are pairwise non - adjacent in @xmath7
. the _ independence number _ of @xmath7 , denoted by @xmath109 , is the size of a maximum independent set in @xmath7 . clearly , for any graph @xmath7 , we have @xmath110 and @xmath111 .
we also use the following known results : 1 .
* seinsche ( @xcite ) * : _ if @xmath112 and @xmath113 are @xmath114-free , then @xmath115 and @xmath116 are @xmath114-free .
* seinsche ( @xcite ) * : _ every @xmath114-free graph is perfect . _
* chudnovsky et al . ( @xcite ) * ( the strong perfect graph theorem * ( spgt ) * ) : _ a graph is perfect if and only if it contains no odd hole ( chordless cycle ) of length at least @xmath73 and no odd antihole ( complement graph of a hole ) of length at least @xmath73 .
* choudum et al . ( @xcite ) * : _ let @xmath41 and @xmath20 be hereditary classes of graphs where @xmath20 admits a linear @xmath1-binding function .
if there exists a constant @xmath35 such that for any @xmath117 , @xmath79 can be partitioned into @xmath35 subsets @xmath118 , where @xmath119\in \cal{f}$ ] for each @xmath120 , then @xmath41 has a linear @xmath1-binding function . _ 5 .
* blazsik et al . ( @xcite ) * : _ for every pseudo - split graph @xmath7 , @xmath121 .
* brause et al . ( @xcite ) * : _ for every ( @xmath8 , paw)-free graph @xmath7 , @xmath121 .
* karthick and maffray ( @xcite ) * : _ for every ( @xmath8 , diamond)-free graph @xmath7 , @xmath121 . _
in this section , we show that the class of ( @xmath11)-free graphs , where @xmath122 is linearly @xmath1-bounded . note that the class of ( @xmath123)-free graphs and the class of ( @xmath124)-free graphs generalize the class of ( @xmath66)-free graphs or pseudo - split graphs . also the class of ( @xmath125)-free graphs , the class of ( @xmath124)-free graphs , and the class of ( @xmath126)-free graphs generalize the class of ( @xmath68)-free graphs .
first we prove a structure theorem for the complement graph of a ( @xmath71)-free graph .
[ 2k2-gem - comp - struc ] let @xmath7 be an imperfect @xmath127-free graph .
then @xmath7 is connected and there exists a partition @xmath128 of @xmath79 such that 1 .
@xmath102 induces a @xmath114-free subgraph of @xmath7 , and 2 .
@xmath103 and @xmath129 are cliques .
let @xmath7 be an imperfect @xmath127-free graph .
since @xmath7 is ( @xmath130)-free , @xmath7 contains no hole of length at least @xmath131 , and since @xmath7 is @xmath50-free , @xmath7 contain no anti - hole of length at least @xmath131 .
thus , it follows from spgt @xcite that @xmath7 contains a @xmath73-hole ( hole of length @xmath73 ) , say @xmath132 with vertex - set @xmath133 , and edge - set @xmath134 .
throughout this proof , we take all the subscripts of @xmath135 to be modulo 5 .
[ cl:1 ] any vertex @xmath136 is adjacent to at least two vertices in @xmath132 .
suppose not .
if @xmath83 is not adjacent to any of the vertices in @xmath132 , or if @xmath83 is adjacent to exactly one vertex in @xmath132 , say to @xmath137 , then @xmath138 induces a @xmath130 in @xmath7 , which is a contradiction .
@xmath139 by claim [ cl:1 ] , @xmath7 is connected .
[ cl:2 ] if @xmath136 , then @xmath140 $ ] is isomorphic to a member of @xmath141 .
suppose not .
then by claim [ cl:1 ] , @xmath142 is either @xmath143 or @xmath144 or @xmath145 , for some @xmath146 .
but then in all the cases , @xmath147 induces a @xmath50 in @xmath7 , which is a contradiction .
@xmath139 for @xmath148 $ ] , let : @xmath149 moreover , let @xmath150 , and @xmath151 .
then by claims [ cl:1 ] and [ cl:2 ] , we have @xmath152 .
[ cl : ai ] for each @xmath148 $ ] ( @xmath146 mod 5 ) , the following hold : * @xmath153 induces a @xmath114-free subgraph of @xmath7 . *
@xmath154 $ ] is complete . *
if @xmath153 is not a clique , then @xmath155 and @xmath156 are cliques in @xmath7 . * if @xmath157 , then @xmath158 .
\(i ) suppose to the contrary that @xmath159 $ ] contains an induced @xmath114 , say @xmath160 .
then by the definition of @xmath153 , @xmath161 induces a @xmath130 in @xmath7 , a contradiction .
so ( i ) holds .
suppose that ( ii ) does not hold .
then there exist vertices @xmath162 and @xmath163 such that @xmath164 .
but , then @xmath165 induces a @xmath130 in @xmath7 , a contradiction .
so ( ii ) holds .
suppose that ( iii ) does not hold .
then there exist two non - adjacent vertices @xmath166 and @xmath167 in @xmath153 , and up to symmetry , two non - adjacent vertices @xmath83 and @xmath168 in @xmath155 .
but then since @xmath154 $ ] is complete ( by ( ii ) ) , @xmath169 induces a @xmath50 in @xmath7 , a contradiction .
so ( iii ) holds .
suppose that ( iv ) does not hold .
then there exist vertices @xmath162 and @xmath170 . by symmetry
, we may assume that @xmath171 .
but then @xmath172 induces a @xmath50 in @xmath7 , if @xmath173 , and @xmath174 induces a @xmath130 in @xmath7 , if @xmath164 , a contradiction .
so ( iv ) holds .
@xmath139 [ cl : bi ] for each @xmath148 $ ] ( @xmath146 mod 5 ) , the following hold : * @xmath175 is a clique .
* @xmath176 induces a @xmath114-free subgraph of @xmath7 . *
@xmath177 = \emptyset = [ b_i , b_{i-2}]$ ] .
suppose that ( i ) does not hold .
then there exist vertices @xmath178 such that @xmath164 .
but , then @xmath179 induces a @xmath50 in @xmath7 , a contradiction .
so ( i ) holds .
suppose that ( ii ) does not hold .
suppose that @xmath180 $ ] contains an induced @xmath114 , say @xmath160 .
then by the definition of @xmath181 s , @xmath161 induces a @xmath130 in @xmath7 , a contradiction .
so ( ii ) holds .
suppose that ( iii ) does not hold .
then there exist vertices @xmath182 and @xmath183 such that @xmath173 . by symmetry
, we may assume that @xmath184 .
but , then @xmath185 induces a @xmath50 in @xmath7 , a contradiction .
so ( ii ) holds .
@xmath139 [ cl : aibi ] for each @xmath148 $ ] ( @xmath146 mod 5 ) , the following hold : * @xmath186 $ ] are complete . * @xmath187 = \emptyset$ ] .
suppose that ( i ) does not hold .
then there exist vertices @xmath162 and @xmath188 such that @xmath164 .
but , then @xmath189 induces a @xmath130 in @xmath7 , a contradiction .
so ( i ) holds .
suppose that ( ii ) does not hold .
then there exist vertices @xmath162 and @xmath190 such that @xmath173 .
but , then @xmath191 induces a @xmath50 in @xmath7 , a contradiction .
so ( ii ) holds .
@xmath139 by claim [ cl : ai](iv ) , we may assume that @xmath192 .
now , by using the above claims , we prove the theorem in two cases ( the other cases are symmetric ) . then by claim [ cl : ai](iii ) , @xmath193 is a clique .
then we define @xmath194 , @xmath195 , and @xmath196 , and we have the following : [ cl:1-v1 ] @xmath102 induces a @xmath114-free subgraph of @xmath7 . by claim [ cl : ai](i ) , @xmath197 $ ] is @xmath114-free , and by claim [ cl : bi](ii ) , @xmath198 $ ] is @xmath114-free .
also , @xmath199 $ ] is complete ( by claim [ cl : aibi](i ) ) .
so , by the definition of @xmath200 , @xmath201 and @xmath202 , by ( r1 ) , @xmath203 $ ] is @xmath114-free , since it is a join of two @xmath114-free graphs , namely @xmath197 $ ] and @xmath204 $ ] .
now , @xmath205 is a clique ( by claim [ cl : bi](i ) ) , @xmath206 = \emptyset$ ] ( by claim [ cl : aibi](ii ) ) , and @xmath207 = \emptyset = [ b_4 , b_2]$ ] ( by claim [ cl : bi](iii ) ) .
thus , by ( r1 ) , @xmath208 $ ] is @xmath114-free , as it is a union of two disjoint @xmath114-free graphs , namely @xmath203 $ ] and @xmath209 $ ] .
hence the claim holds .
@xmath139 [ cl:1-v2v3 ] @xmath103 and @xmath129 are cliques in @xmath7 .
the proof follows by the definition of @xmath193 , @xmath210 and @xmath211 , and by claims [ cl : bi](i ) and [ cl : aibi](i ) .
@xmath139 so , by claims [ cl:1-v1 ] and [ cl:1-v2v3 ] , we see that @xmath128 is a required partition of @xmath79 .
the proof is similar to that of case 1 .
hence the theorem is proved .
the following corollary is an improvement over that in @xcite , where it is shown that for every ( @xmath71)-free graph @xmath7 , @xmath72 .
[ bound-2k2-gem ] let @xmath7 be a ( @xmath71)-free graph .
then @xmath212 . .
consider the complement @xmath40 of @xmath7 .
then @xmath40 is a ( @xmath213)-free graph . if @xmath40 is perfect , then @xmath214 , and the corollary holds .
if @xmath40 is imperfect , then by theorem [ 2k2-gem - comp - struc ] , @xmath40 is connected and there exists a partition ( @xmath215 ) of @xmath40 such that @xmath102 induces a @xmath114-free subgraph of @xmath40 , and @xmath103 and @xmath129 are cliques in @xmath40 .
so , @xmath216)+\theta([v_2 \cup v_3])$ ] . since every @xmath114-free graph is perfect ( by ( r2 ) )
, we have @xmath217 ) + 2 \leq \alpha(h)+2 $ ] , and the corollary follows .
first we prove a structure theorem for the class of ( @xmath123)-free graphs .
[ 2k2 - 4-wheel - struc ] let @xmath7 be a connected ( @xmath123)-free graph .
then @xmath7 is either a pseudo - split graph or there exists a partition @xmath218 of @xmath79 such that 1 .
@xmath208 $ ] is either a pseudo - split graph of @xmath7 with @xmath219 ) \leq \omega(g ) -1 $ ] or the complement of a bipartite graph of @xmath7 , and 2 .
@xmath106 is an independent set , for each @xmath220 .
moreover , if @xmath102 induces a pseudo - split graph of @xmath7 , then @xmath221 .
let @xmath7 be a connected ( @xmath123)-free graph .
if @xmath7 is @xmath50-free , then @xmath7 is a pseudo - split graph , and the theorem holds .
suppose that @xmath7 contains an induced @xmath50 , say @xmath132 with vertex - set @xmath222 , and edge - set @xmath223 .
define sets @xmath224 and @xmath225 . throughout this proof , we take all the subscripts of @xmath135 to be modulo 4 . [
cl : l1 ] if @xmath226 , then @xmath227 . otherwise , @xmath228 induces a @xmath229 in @xmath7 , a contradiction .
@xmath139 so , for any @xmath226 , there exists an index @xmath230 $ ] such that @xmath231 and @xmath232 . for @xmath233
$ ] , let : @xmath234 moreover , let @xmath235 , @xmath236 , and @xmath237 .
then , by claim [ cl : l1 ] , @xmath238 . now : [ cl : l2 ] the following hold : * if @xmath239 , then @xmath240 . *
@xmath241 is an independent set .
\(i ) we may assume that @xmath242 , and suppose to the contrary that @xmath243 .
then @xmath244 induces a @xmath8 in @xmath7 , a contradiction .
so ( i ) holds .
suppose that ( ii ) does not hold .
then there exist two adjacent vertices , say @xmath83 and @xmath168 in @xmath241 .
but , then @xmath245 induces a @xmath8 in @xmath7 , a contradiction .
so ( ii ) holds .
@xmath139 [ cl : l1-y ] for each @xmath233 $ ] ( @xmath146 mod 4 ) , the following hold : * @xmath246 is an independent set . *
if @xmath247 , then either @xmath248 or @xmath249 . *
if @xmath250 , then @xmath251 .
we prove the claim for @xmath252 .
\(i ) suppose to the contrary that there exist two adjacent vertices , say @xmath83 and @xmath168 in @xmath253 .
then @xmath254 induces a @xmath8 in @xmath7 , a contradiction .
so ( i ) holds .
suppose that ( ii ) does not hold .
then there exist vertices @xmath255 , @xmath256 and @xmath257 .
then since @xmath258 or @xmath259 or @xmath260 do not induce a @xmath8 in @xmath7 , @xmath261 induces a triangle in @xmath7 .
but , then @xmath262 induces a @xmath229 in @xmath7 , a contradiction .
so ( ii ) holds .
suppose that ( iii ) does not hold .
then there exist vertices @xmath263 and @xmath264 .
but then @xmath265 induces a @xmath229 in @xmath7 , if @xmath173 , or @xmath266 induces a @xmath8 in @xmath7 , if @xmath164 , a contradiction .
so ( iii ) holds . @xmath139 [ cl : yi ] for each @xmath267 , @xmath268 is a union of a clique and an independent set .
we prove the claim for @xmath252 .
first , we show that @xmath269 $ ] is @xmath270-free .
suppose to the contrary that @xmath269 $ ] contains an induced @xmath270 , say @xmath160 .
then by the definition of @xmath271 , @xmath272 induces a @xmath229 in @xmath7 , a contradiction .
so , @xmath269 $ ] is @xmath270-free , and hence it is a union of cliques .
then since @xmath7 is @xmath8-free , it follows that @xmath271 is a union of a clique and an independent set , and the claim holds .
@xmath139 by claim [ cl : yi ] , for each @xmath267 , we define @xmath273 , where @xmath274 is a clique , and @xmath275 is an independent set .
[ cl : yizi-1 ] for each @xmath276 , @xmath277= \emptyset$ ] .
we prove the claim for @xmath252 .
suppose to the contrary that there exist vertices , say @xmath278 and @xmath279 such that @xmath280 .
but , then @xmath281 or @xmath282 induces a @xmath229 in @xmath7 , a contradiction .
so the claim holds .
@xmath139 [ cl : yizi-2 ] for each @xmath276 , if @xmath283 , then @xmath284 is an independent set .
we prove the claim for @xmath252 .
let @xmath285 .
up to symmetry , we may assume that @xmath286 .
by claim [ cl : yizi-1 ] , @xmath287 = \emptyset$ ] .
now , we show that @xmath288 is an independent set .
suppose to the contrary that there exist adjacent vertices , say @xmath18 and @xmath19 in @xmath288 .
then since @xmath287 = \emptyset$ ] , we have @xmath289 and @xmath290 .
but , then @xmath291 induces a @xmath8 in @xmath7 , a contradiction .
so the claim holds .
@xmath139 now , by using claim [ cl : l1-y](iii ) , we prove the theorem in two cases . by claim [ cl : l1-y](ii ) and by symmetry
, we may assume that either @xmath292 or @xmath293 .
then we define @xmath294 , @xmath295 , @xmath296 , @xmath297 .
further : if @xmath298 , then we define @xmath299 and @xmath300 ; and if @xmath293 , then we define @xmath301 and @xmath302 .
now , by claims [ cl : l2 ] and [ cl : l1-y](i ) , and by the definition of @xmath274 s and @xmath275 s , we see that @xmath208 $ ] is isomorphic to the complement of a bipartite graph , and @xmath106 s are independent sets , for each @xmath220 .
so , @xmath218 is a required partition of @xmath79
. we may assume up to symmetry that @xmath252 and @xmath303 .
then by claim [ cl : yizi-2 ] , @xmath288 is an independent set .
then , we define @xmath304 , @xmath305 , @xmath306 , @xmath307 , @xmath308 and @xmath309
. then since @xmath7 is ( @xmath229)-free , @xmath102 induces a pseudo - split graph in @xmath7 .
also , @xmath219 ) \leq \omega(g ) -1 $ ] .
so , by claims [ cl : l2 ] and [ cl : l1-y](i ) , we see that @xmath106 s are independent sets , for each @xmath310 , and hence @xmath218 is a required partition of @xmath79 .
this completes the proof of the theorem . .
] [ bound-2k2-wheel ] let @xmath7 be a ( @xmath123)-free graph .
then @xmath311 . .
let @xmath7 be a ( @xmath123)-free graph .
we may assume that @xmath7 is connected .
we use theorem [ 2k2 - 4-wheel - struc ] . if @xmath7 is a pseudo - split graph , then , by ( r5 ) , @xmath312 .
so , suppose that @xmath79 admits a partition as in theorem [ 2k2 - 4-wheel - struc ] .
now : \(a ) suppose that @xmath102 induces a pseudo - split graph with @xmath219 ) \leq \omega(g ) -1 $ ] .
then @xmath313 .
so , @xmath314 ) + 3 $ ] .
then by ( r4 ) and ( r5 ) , @xmath315 ) + 1 + 3 \leq \omega(g)+3 $ ] .
\(b ) suppose that @xmath208 $ ] is isomorphic to the complement of a bipartite graph , and @xmath106 s are independent sets , for each @xmath220 .
then since @xmath208 $ ] is perfect , it follows by ( r4 ) that @xmath315 ) + 5\leq \omega(g)+5 $ ] .
hence the corollary is proved .
we use the structure theorem for @xmath316-free graphs proved in @xcite .
we need the following definition .
let @xmath7 be a graph on @xmath16 vertices @xmath317 , and let @xmath318 be any @xmath16 vertex disjoint graphs . then an _ expansion _
@xmath319 of @xmath7 @xcite is the graph obtained from @xmath7 by ( i ) replacing the vertex @xmath135 of @xmath7 by @xmath320 , @xmath321 , and \(ii ) joining the vertices @xmath322 , @xmath323 iff @xmath135 and @xmath324 are adjacent in @xmath7 . an expansion is also called a _ composition _ ; see @xcite .
if @xmath320 s are complete , it is called a _ complete expansion _ of @xmath7 . by a result of lovsz @xcite , if @xmath325 are perfect , then @xmath319 is perfect .
[ structure thm ] if @xmath7 is a connected @xmath316-free graph , then @xmath7 is either chordal or there exists a partition @xmath128 of @xmath79 such that ( 1 ) @xmath208 \cong k_m^c$ ] , for some @xmath326 , ( 2 ) @xmath327 \cong k_t$ ] , for some @xmath328 , ( 3 ) @xmath329 $ ] is isomorphic to a graph obtained from one of the basic graphs @xmath330 ( @xmath331 ) shown in figure [ basic - graphs - p2p3-c4 ] by expanding each vertex indicated in circle by a complete graph ( of order @xmath332 ) , ( 4 ) @xmath333 = \emptyset$ ] , and ( 5 ) @xmath334 $ ] is complete ( see figure [ basic - graphs - p2p3-c4 ] for the set @xmath90 ) . for @xmath335 $ ] ,
let @xmath336 denote the class of graphs obtained from @xmath330 ( see figure [ basic - graphs - p2p3-c4 ] ) by the operations stated in theorem [ structure thm ] .
[ bound-2k2-paraglider ] let @xmath7 be a ( @xmath124)-free graph .
then @xmath312 . .
consider the complement @xmath40 of @xmath7 .
then @xmath40 is a @xmath316-free graph . if @xmath40 is chordal
, then @xmath40 is perfect and so @xmath214 , and the corollary holds .
suppose that @xmath40 is not chordal .
let @xmath337 ( @xmath338 ) denote the components of @xmath40 . then
since @xmath40 is not chordal and since @xmath40 is @xmath316-free , by theorem [ structure thm ] , we may assume that there exists a component , say @xmath92 of @xmath7 such that @xmath339 admits a partition @xmath128 as in theorem [ structure thm ] where @xmath329 $ ] contains either a @xmath340 or a @xmath341 , and @xmath329 \in { \cal{g}}_t$ ] , for @xmath342 $ ] . then since @xmath40 is ( @xmath343)-free , @xmath344 , for each @xmath345 .
so @xmath346 .
now , @xmath216 ) + \theta([v_2\cup v_3 ] ) + ( k-1 ) = |v_1| + \theta([v_2\cup v_3 ] ) + ( k-1 ) = \theta([v_2\cup v_3 ] ) + \alpha(h)-\alpha(g_t)$ ] .
it is easily verified that @xmath347 ) \leq \alpha(g_t)+1 $ ] .
hence , @xmath348 , and the corollary is proved .
the graphs @xmath349 show that the bound in corollary [ bound-2k2-paraglider ] is tight . in order to prove our next results , we need the following notation .
let @xmath7 be a connected graph that contains an induced diamond , say @xmath352 , with vertex set @xmath353 and edge set @xmath354 .
define sets @xmath224 and @xmath225 . moreover , let : @xmath355 , \\
y_1 & = & \{x \in l_1 \mid n(x)\cap
l_0 = \{v_1 , v_{2}\}\},\\ y_2 & = & \{x \in l_1 \mid n(x)\cap l_0 = \{v_2 , v_{3}\}\},\\ z_1 & = & \{x\in l_1 \mid n(x)\cap l_0 = \{v_{1 } , v_3\}\},\\ z_2 & = & \{x\in l_1 \mid n(x)\cap l_0 = \{v_{1 } , v_2 , v_3\}\}.\end{aligned}\ ] ] then we have the following lemma , and we leave its proof as it can be routinely verified .
[ 2k2-struc - diamond ] let @xmath7 be a connected @xmath8-free graph that contains an induced diamond @xmath352 .
let @xmath356 , subsets of @xmath357 , and @xmath241 be defined as above . then the following hold : 1 .
2 . we have either @xmath359 or @xmath360 .
@xmath361 , @xmath288 , @xmath362 and @xmath241 are independent sets .
4 . @xmath363 = \emptyset$ ] .
[ 2k2-hvn - struc ] let @xmath7 be a connected ( @xmath126)-free graph .
then @xmath7 is either a ( @xmath8 , diamond)-free graph or there exists a partition @xmath364 of @xmath79 such that 1 . @xmath102 induces a ( @xmath8 , paw)-free graph of @xmath7 with @xmath219 ) \leq \omega(g ) -1 $ ] , and 2 .
@xmath106 is an independent set , for each @xmath310 .
let @xmath7 be a connected ( @xmath126)-free graph .
if @xmath7 is diamond - free , then the theorem holds .
suppose that @xmath7 contains an induced diamond , say @xmath352 .
we use lemma [ 2k2-struc - diamond ] . by ( 2 ) and by symmetry
, we may assume that @xmath360 .
now , since @xmath7 is @xmath365-free , we have the following : * for any @xmath366 , @xmath367 induces a paw - free graph with @xmath368 ) \leq \omega(g ) -1 $ ] .
* @xmath369 is an independent set ( by using ( 3 ) ) .
define @xmath370 , @xmath371 , @xmath372 , and @xmath373 .
then by ( 3 ) and ( 4 ) , and by the above properties , we see that @xmath364 is a required partition of @xmath79 , and the theorem is proved . [ bound-2k2-hvn ]
let @xmath7 be a ( @xmath126)-free graph
. then @xmath374 .
let @xmath7 be a connected ( @xmath126)-free graph .
we use theorem [ 2k2-hvn - struc ] .
if @xmath7 is a ( @xmath8 , diamond)-free graph , then by ( r8 ) , @xmath312 , and the corollary holds .
suppose that @xmath79 admits a partition as in theorem [ 2k2-hvn - struc ] .
so , @xmath314 ) + 3 $ ] . since @xmath375 ) \leq \omega([v_1 ] ) + 1 $ ] ( by ( r7 ) ) , we have @xmath315 ) + 1 + 3 \leq \omega(g ) + 3 $ ] , as desired .
[ 2k2-k5-e - struc ] let @xmath7 be a connected ( @xmath125)-free graph .
then @xmath7 is either a ( @xmath8 , diamond)-free graph or there exists a partition @xmath376 of @xmath79 such that 1 . @xmath102 induces a ( @xmath8 , diamond)-free graph of @xmath7 with @xmath219 ) \leq \omega(g ) -1 $ ] , and 2 .
@xmath106 is an independent set , for each @xmath377 .
let @xmath7 be a connected ( @xmath125)-free graph .
if @xmath7 is diamond - free , then the theorem holds .
suppose that @xmath7 contains an induced diamond , say @xmath352 .
we use lemma [ 2k2-struc - diamond ] . by ( 2 ) and by symmetry
, we may assume that @xmath360 .
now , since @xmath7 is ( @xmath378)-free , we have the following : * for any @xmath366 , @xmath367 induces a diamond - free graph with @xmath368 ) \leq \omega(g ) -1 $ ] .
* @xmath379 is an independent set .
define @xmath370 , @xmath371 , @xmath380 , @xmath373 , and @xmath381 .
then by ( 3 ) and ( 4 ) , and by the above properties , we see that @xmath376 is a required partition of @xmath79 , and the theorem is proved .
[ bound-2k2-k5-e ] let @xmath7 be a ( @xmath125)-free graph
. then @xmath382 .
let @xmath7 be a connected ( @xmath125)-free graph .
we use theorem [ 2k2-k5-e - struc ] . if @xmath7 is a ( @xmath8 , diamond)-free graph , then by ( r8 ) , @xmath312 , and the corollary holds .
suppose that @xmath79 admits a partition as in theorem [ 2k2-k5-e - struc ] .
so , @xmath314 ) + 4 $ ] . since @xmath375 ) \leq \omega([v_1 ] ) + 1 $ ] ( by ( r8 ) ) , we have @xmath315 ) + 1 + 4 \leq \omega(g ) + 4 $ ] , as desired .
[ general-3 ]
let @xmath40 be any graph .
suppose that for every @xmath384-free graph @xmath385 , @xmath386 .
then for every ( @xmath387-free graph @xmath7 , we have @xmath388 . .
let @xmath7 be a ( @xmath387-free graph .
if @xmath7 is an edgeless graph , then the theorem is obvious .
so we may assume that there exist adjacent vertices , say @xmath137 and @xmath389 in @xmath79 . for each @xmath267
, let @xmath390 .
also , let @xmath391 and @xmath392 .
then , we have the following : 1 . since @xmath7 does not induce a @xmath393 , we have : for any @xmath366 , @xmath367 induces a @xmath40-free graph .
so , for each @xmath276 , @xmath394 $ ] is a @xmath40-free graph with @xmath395 ) \leq \omega(g ) -1 $ ] .
2 . since @xmath7
does not induce a @xmath8 , we see that @xmath132 is an independent set . now , @xmath396 ) + \chi([a_2 ] )
+ \chi([c\cup \{v_1\ } ] ) = \chi([a_1 \cup b \cup \{v_2\ } ] ) + \chi([a_2 ] ) + \chi([c\cup \{v_1\}])$ ] . since for every ( @xmath11)-free graph @xmath385 , @xmath386 , and since @xmath397 is an independent set ( by ( ii ) ) , we have , by ( r4 ) , @xmath398 ( by ( i ) ) , as desired . [ 2k2-k5-bound ] [ bound-2k2-k5 ]
let @xmath7 be a ( @xmath77)-free graph
. then @xmath399 . .
since @xmath7 is a ( @xmath8 , @xmath400)-free graph , and since for every ( @xmath64)-free graph @xmath385 , @xmath401 ( see @xcite ) , the corollary follows by theorem [ general-3 ] .
in this section , we show that some superclasses of @xmath8-free graphs are @xmath1-bounded .
if @xmath7 is a graph and if @xmath402 is an edge in @xmath7 , then we simply write @xmath403 to denote the set of all vertices in @xmath7 that are not adjacent to both @xmath404 and @xmath405 in @xmath7 .
the proof of the following theorem is very similar to the proof of wagon @xcite for the class of @xmath8-free graphs , and we give it here for completeness .
[ general ] let @xmath406 be a class of graphs and let @xmath7 be any graph .
suppose that @xmath406 is @xmath1-bounded with @xmath1-binding function @xmath2 .
suppose that for every edge @xmath407 in @xmath7 , @xmath408 \in \cal{h}$ ] .
then @xmath409 . .
let @xmath410 and let @xmath411 be a complete subgraph of @xmath7 with @xmath412 , and @xmath413 .
then every vertex in @xmath414 is not adjacent to at least one vertex in @xmath411 .
otherwise , @xmath415 induces a clique of size larger than @xmath416 which is a contradiction . for
each @xmath417 $ ] , @xmath418 , let @xmath419 , where @xmath420 is the edge @xmath421 , and let @xmath422~ \mbox{is~complete}\}$ ] .
moreover , let @xmath423 and @xmath424
. then @xmath425 . 1 .
since for every edge @xmath407 in @xmath7 , @xmath408 \in \cal{h}$ ] , we have @xmath426 \in \cal{h}$ ] .
@xmath427 is an independent set . if not , then there exist adjacent vertices , say @xmath83 and @xmath168 in @xmath181 .
but , then @xmath428 induces a clique of size @xmath429 , a contradiction .
so , @xmath430 } \chi([a(e_{ij } ) ] ) + \sum_{i = 1}^{\omega}\chi([b_i\cup\{v_i\}])$ ] .
then by ( i ) and ( ii ) , and by ( r4 ) , @xmath430 } f(\omega([a(e_{ij } ) ] ) ) + \sum_{i = 1}^{\omega}\chi([b_i\cup\{v_i\}])$ ] . then since @xmath431 )
\leq \omega$ ] and since @xmath427 is an independent set , for each @xmath432 $ ] , @xmath418 , we have @xmath433 , and the theorem is proved
. then we immediately have the following . .
since @xmath7 is @xmath8-free , for each @xmath417 $ ] , @xmath418 , @xmath435 is an independent set in @xmath7 .
so , @xmath431 ) \leq 1 $ ] , and hence the corollary follows from the proof of theorem [ general ] .
fldes , s. , hammer , p. l. : split graphs , proceedings of the eighth south - eastern conference on combinatorics , graph theory and computing ( 1977 ) , congressus numerantium * xix * , winnipeg : utilitas math . , 311 - 315 . | a hereditary class @xmath0 of graphs is @xmath1-_bounded _ if there is a @xmath1-_binding function _ ,
say @xmath2 such that @xmath3 , for every @xmath4 , where @xmath5 ( @xmath6 ) denote the chromatic ( clique ) number of @xmath7 .
it is known that for every @xmath8-free graph @xmath7 , @xmath9 , and the class of ( @xmath10)-free graphs does not admit a linear @xmath1-binding function . in this paper , we are interested in classes of @xmath8-free graphs that admit a linear @xmath1-binding function . we show that the class of ( @xmath11)-free graphs , where @xmath12 admits a linear @xmath1-binding function .
also , we show that some superclasses of @xmath8-free graphs are @xmath1-bounded . .
chromatic number ; clique number ; graph classes ; @xmath8-free graphs . |
soon after the westerbork radio synthesis telescope ( wsrt ) went online , @xcite made the first report of a correlation between 1415 mhz radio and 10 @xmath0 m mid - infrared luminosities for a sample of seyfert galaxy nuclei .
this discovery was soon extended to the nuclei of normal spirals , though the dispersion was found to be smaller in the correlation for the original sample of seyferts ( @xcite ) .
it was not until the coming of iras that the optically thin radio continuum emission from galaxies was found to be better correlated with the far infrared ( fir ) dust emission of galaxies _ without _ an active galactic nucleus ( agn ) ( @xcite ) .
the connection between radio and infrared emission from galaxies is that they are both powered by massive stars , as pointed out originally for starbursts by @xcite .
young massive stars , which heat up dust to provide the fir emission , are the same stars which end as supernovae ( sne ) and bring about the synchrotron emission observed at radio wavelengths .
if this general picture is correct , the fact that the mean free path of uv photons ( @xmath3100 pc ) which heat the dust is much less than the diffusion length for a cr@xmath2 ( @xmath31 - 2 kpc ) suggests that the radio image should resemble a smeared version of the infrared image .
this idea was first introduced by @xcite , who attempted to model the propagation of cr@xmath2s by smearing iras data of galaxies using parameterized kernels containing the physics of the cr@xmath2 propagation and diffusion , to better match the morphology of the corresponding radio data . in an attempt to better understand the fir - radio correlation
, we are using mips infrared data from the _ spitzer _ infrared nearby galaxies survey ( sings ) legacy science project ( @xcite ) . in this initial study
, we examine the variations of the fir - radio correlation on sub - kpc scales within four of the nearest face - on galaxies in the sings sample : ngc 2403 , ngc 3031 ( m81 ) , ngc 5194 ( m51a ) , and ngc 6946 . using high resolution _ spitzer _ imaging , we are also able to test the smearing model of @xcite with greater accuracy , at higher spatial resolution , and in more galaxies , with the aim of gaining better insight into cr@xmath2 diffusion and confinement within galaxy disks . for a complete discussion ,
we direct the reader to @xcite .
in figure [ qmaps ] we plot @xmath4 } { f_{\nu}(22~\rm{cm})[\rm{jy}]}\right)$ ] for each galaxy .
we find that elevated @xmath1 ratios are associated with bright structures in the infrared and radio images of each galaxy .
the most obvious cases are seen for the bright spiral arms of ngc 3031 , ngc 5194 and ngc 6946 .
all three galaxies show enhanced @xmath1 ratios along their arms , with local peaks centered on h ii regions , and depressed ratios located in the quiescent inter - arm and outer - disk regions of each galaxy . for ngc 2403 , which does not have a grand - design spiral morphology
, we still see @xmath1 peaks associated with h ii regions . to quantify radial variations in @xmath1 , along with any dependencies of @xmath1 on the 70 @xmath0 m surface brightness within each galaxy , we perform aperture photometry using apertures having diameters equal to the fwhm of the mips 70 @xmath0 m psf ( i.e. 17 ) .
the measured dispersion of @xmath1 within each galaxy is less than @xmath30.25 dex .
we find a general trend of increasing @xmath1 ratios with increasing 70 @xmath0 m surface brightness and decreasing radius . the residual dispersion around the trend of @xmath1 with increasing radius is found to be larger than the residual dispersion around the trend of @xmath1 with increasing 70 @xmath0 m surface brightness by @xmath30.1 dex .
this suggests that the star formation sites within the disk are more important in determining the overall appearance of the @xmath1 maps compared to the exponential profiles of the disks themselves . in figure [ globloc ]
we show our local sub - kpc @xmath1 ratios together with global @xmath1 values for 1752 galaxies from the study of @xcite . using their cataloged 1.4 ghz nrao vla sky survey ( nvss ) and 60 and 100 @xmath0 m iras data , we converted iras based @xmath5 values to _ spitzer _ @xmath1 values using the sed models of @xcite .
we find that the dispersion in the local sub - kpc @xmath1 ratios within our sample galaxies is comparable to that of global ratios . in the @xcite sample , @xmath1 appears to be roughly constant with increasing galaxy luminosity while , within each disk , @xmath1 ratios clearly increase with the 70 @xmath0 m luminosity .
this difference in the observed trend between @xmath1 and luminosity within and among galaxies can be explained by the diffusion of cr@xmath2 within the galaxy disks .
to help understand the underlying physics of the correlation , we look to see whether the image - smearing model works to improve the correlation between the infrared and radio morphologies of galaxies .
determining the functional form of the best fit smearing kernel provides insight into the propagation and diffusion characteristics of cosmic rays within galaxy disks . in this study
we examine the behavior for gaussian and exponential kernels oriented in the planes of the sky and each galaxy disk .
we define the residuals between the smeared 70 @xmath0 m map ( @xmath6:@xmath7shape , @xmath8orientation , and @xmath9=scale - length ) and the observed 22 cm map ( r ) by , @xmath10^{2}}{\sum r_{j}^2 } , \label{phi}\ ] ] where @xmath11 and @xmath12 indexes each pixel . in figure [ res70 ]
we plot @xmath13 , as a function of the scale - length ( @xmath9 ) , for each galaxy and find that the image - smearing technique improves the overall correlation between the radio map and the 70 @xmath0 m image by an average of @xmath30.2 dex .
unlike @xcite , we find exponential kernels are preferred independent of projection over gaussian kernels , suggesting that additional processes such as escape and decay appear necessary to describe their evolution through the galaxy disks .
we find distinct behavior between our active and less active star - forming galaxies .
figure [ res70 ] shows that scale - lengths @xmath141 kpc work best to improve the correlation for the more active star - forming galaxies ( ngc 5194 and ngc 6946 : sfr@xmath154 m@xmath16/yr ) , while scale - lengths @xmath171 kpc work best for the less active star - forming galaxies ( ngc 2403 and ngc 3031 : sfr @xmath141m@xmath16/yr ) .
we also find that by smearing the 70 @xmath0 m images , we are able to reduce the slope of the observed non - linearity in @xmath1 with 70 @xmath0 m surface brightness by @xmath1725% for each galaxy .
this suggests that the non - linearity may be due to the diffusion of cr@xmath2s and is not expected to be found in the _ global _ correlation when integrating the flux over entire galaxies .
in examining the behavior of @xmath1 within the disks of our sample galaxies , along with our image - smearing analysis , we find : 1 .
an empirical trend of @xmath1 decreasing with declining 70 @xmath0 m surface brightness and increasing radius to be a general property within the galaxies .
however , the dispersion measured in @xmath1 at constant 70 @xmath0 m surface brightness is found to be smaller than at constant radius by @xmath30.1 dex , suggesting that star formation sites are more important in determining the @xmath1 disk appearance than the underlying exponential disks .
2 . the dispersion in the _ global _
fir - radio correlation is comparable to the dispersion for @xmath1 on sub - kpc scales within the galaxy disks .
also , the trend of increasing @xmath1 ratio with increasing 70 @xmath0 m luminosity within each galaxy is not observed in the _ global _ correlation .
the phenomenological modeling of cosmic ray electron ( cr@xmath2 ) diffusion using an image - smearing technique is successful as it both decreases the measured dispersion in @xmath1 and reduces the slope of the observed non - linearity in @xmath1 with 70 @xmath0 m surface brightness .
the latter suggests that the non - linearity may be due to the diffusion of cr@xmath2s from star - forming regions .
exponential kernels work marginally better to tighten the correlation than gaussian kernels , independent of projection .
this result suggests that cr@xmath2 evolution is not well described by a pure random - walk diffusion and requires additional processes such as escape and decay .
our two less active star - forming galaxies require kernels having larger scale - lengths to improve the correlation while our two more active galaxies require smaller scale - lengths .
this difference may be due to time - scale effects in which there has been a deficit of recent cr@xmath2 injection into the ism of the two less active star - forming galaxies , thus leaving the underlying diffuse disk as the dominant structure in the morphology .
< widest bib entry > bicay m. d. and helou , g. 1990 , apj , 362 , 59 dale , d. a. and helou , g. 2002 , apj , 576 , 159 de jong , t. , klein , u. , wielebinski , r. , and wunderlich , e. 1985 , a&a , 147 , l6 harwit , h , and pacini , f 1975 , apjl , 200 , l127 helou , g. , soifer , b. t. , and rowan - robinson , m. 1985 , apjl , 298 , l7 kennicutt , r. c. jr . , et al .
2003 , pasp , 115 , 928 marsh , k. a. and helou , g. 1998 , apj , 493 , 121 murphy , e. j. , et al .
2005 , apj , submitted van der kruit , p. 1971
, a&a , 15 , 110 van der kruit , p. 1973 , a&a , 29 , 263 yun , m. s. , reddy , n. a. , and condon , j. j. 2001 , apj , 554 , 803 | we present an initial look at the far infrared - radio correlation within the star - forming disks of four nearby , nearly face - on galaxies ( ngc 2403 , ngc 3031 , ngc 5194 , and ngc 6946 ) . using _ spitzer _
mips imaging and wsrt radio continuum data , we are able to probe variations in the logarithmic 70 @xmath0m/22 cm ( @xmath1 ) flux density ratios across each disk at sub - kpc scales .
we find general trends of decreasing @xmath1 with declining surface brightness and with increasing radius .
we also find that the dispersion in @xmath1 within galaxies is comparable to what is measured _ globally _ among galaxies at around 0.2 dex .
we have also performed preliminary phenomenological modeling of cosmic ray electron ( cr@xmath2 ) diffusion using an image - smearing technique , and find that smoothing the infrared maps improves their correlation with the radio maps .
the best fit smoothing kernels for the two less active star - forming galaxies ( ngc 2403 and ngc 3031 ) have much larger scale - lengths than that of the more active star - forming galaxies ( ngc 5194 and ngc 6946 ) .
this difference may be due to the relative deficit of recent cr@xmath2 injection into the interstellar medium ( ism ) for the galaxies having largely quiescent disks .
a more complete discussion of this proceedings article can be found in @xcite .
infrared : galaxies radio continuum : galaxies cosmic rays : galaxies |
incommensurate magnetic structures in insulators are as a rule caused by a competition between two or more magnetic interactions . in the classic description of a spiral spin structure in mno@xmath0 yoshimori attributed the spiral spin arrangement to a competition between two distinct antiferromagnetic ( af ) heisenberg exchange interactions in the crystal.@xcite we shall refer to this kind of spiral state as `` exchange spiral '' .
an alternative mechanism is realized in a few systems with non - centric crystal structures .
isotropic exchange terms in the hamiltonian compete with dzyaloshinskii - moriya ( d - m ) interactions.@xcite the former are proportional to @xmath5 , i.e. , the scalar product of interacting spins , and thus favor a collinear state .
the d - m term is given by a _
vector _ product and is usually written as @xmath6)$ ] , where @xmath7 is the so - called dzyaloshinskii vector that characterizes the oriented bond between spins @xmath8 and @xmath9 .
d - m interactions favor a relative angle of @xmath10 between spins . unlike isotropic exchange , d - m coupling is of relativistic origin , i.e. , is a result of spin - orbit interactions , and ties the spin system to the underlying crystal lattice via orbital degrees of freedom .
the best known examples of real compounds where this second scenario ( `` relativistic spiral '' ) is realized are the mnsi@xcite and fege.@xcite recently we have observed an incommensurate spiral spin structure in ba@xmath11cuge@xmath0o@xmath1 and explained it within this framework.@xcite while the ground states in both models ( `` exchange '' and `` relativistic '' ) may be similar , under certain conditions they behave very differently in applied magnetic fields .
if only isotropic exchange interactions are present , the direction of magnetic field is irrelevant to the spin structure : the spin space may be freely rotated without a change in interaction energy and the spins are not tied to any particular direction in the crystal . for `` relativistic ''
spiral magnets the spin structure in an external field @xmath4 in general may be strongly dependent on the relative orientation of @xmath4 and the dzyaloshinskii vectors , although in several special cases of high - symmetry crystal structures this is not so . in fege for example ,
the interactions of each spin with its neighbors are characterized by dzyaloshinskii vectors pointing in all three equivalent orthogonal directions . in this situation applying a very small magnetic field always leads to a re - orientation of the spin rotation plane normal to the field direction , and , consequently , the propagation vector becomes aligned along the field .
the length of the magnetic propagation vector is to a good approximation field - independent . in materials of lower crystal symmetry
, as for example in hexagonal cscucl@xmath12@xcite and rbmnbr@xmath12 @xcite , or tetragonal ba@xmath11cuge@xmath0o@xmath1 , the spin rotation plane can not be freely re - oriented without affecting the dzyaloshinskii energy . as a result , the magnetic structure undergoes drastic changes for certain field geometries , and the length of the propagation vector becomes strongly field - dependent .
an extreme example of this behaviour is the commensurate - incommensurate transition that we have recently observed and theoretically analyzed for ba@xmath11cuge@xmath0o@xmath1.@xcite as a magnetic field is applied along the unique @xmath13 axis of the tetragonal structure , the initially uniform sinusoidal spin - spiral is distorted . in a non - zero field
it may be viewed as a soliton lattice , a regular arrangement of anti - phase domain wall boundaries separating regions of commensurate antiferromagnetic spin - flop phase . as the field
is increased , the period of the structure , given by the distance between solitons , increases and diverges at some critical field ,
@xmath14 t , above which a commensurate spin - flop state is observed . in this work
we address the field dependence of the magnetic structure of ba@xmath11cuge@xmath0o@xmath1 in a magnetic field @xmath4 applied in the @xmath2 tetragonal plane .
we show that the square - lattice spin arrangement in this geometry allows an almost unhindered re - orientation of the spin plane with only a very small change in the period of the structure , much like in cubic fege .
the principal difference between fege and ba@xmath11cuge@xmath0o@xmath1 is that in the former material the propagation vector rotates towards the direction of applied field .
we find that for ba@xmath11cuge@xmath0o@xmath1 the propagation vector tends to form a finite angle with @xmath4 , that depends on the orientation of @xmath4 itself .
the rotation of the propagation vector with increasing field is continuous .
the observed behaviour is found to be in very good agreement with theoretical predictions based on a 2-dimensional generalization of the simple ginzburg - landau energy functional that we have previously employed to describe the dzyaloshinskii transition in ba@xmath11cuge@xmath0o@xmath1 .
a single crystal of ba@xmath11cuge@xmath0o@xmath1 , the same sample that was used in our previous work , was studied in a neutron diffraction experiment at the high flux isotope reactor at oak ridge national laboratory , on the hb-3 triple - axis spectrometer .
the sample , roughly @xmath15 mm@xmath16 in volume , was mounted in a small aluminum container where it was secured with compressed al - foil .
the container was in turn mounted on a precision micro - goniometer , that was used to align the @xmath2 crystallographic plane in the scattering plane of the spectrometer , prior to setting up the sample in the cryomagnet .
the magnetic field was produced by using a split - coil _ horizontal - field _ superconducting solenoid .
the construction of the magnet provides two pairs of windows for the incident and outgoing neutron beams .
the larger pair of windows , each window being @xmath17 wide , is oriented perpendicular to the field direction .
an additional narrower pair of windows ( @xmath18 ) is positioned around the direction of magnetic field . within the cryomagnet
the sample could be manually rotated around the vertical axis _ in situ_.
we estimate the uncertainty in the initial alignment of the sample relative to the field direction to be of the order of @xmath19 . after this initial setting the relative rotation of the sample in the magnet
was done with much higher accuracy , @xmath20 , as it could be directly followed by monitoring the in - plane bragg reflections .
measurements were done for several values of the angle @xmath21 between the magnetic field and the @xmath22 $ ] direction , namely , for @xmath2345.0 , 50.0 , 52.9 , 57.5 , 60.2 , 61.7 , 61.9 , 71.5 and 75.2@xmath24 , the field always being applied in the @xmath2 plane of the crystal .
although the magnet is capable of much higher field strengths , stray fields are a serious problem , and all the measurements were done in @xmath25 t. the temperature of the sample was controlled by a standard he - flow cryostat , allowing us to perform measurements in the temperature range @xmath26@xmath27 k. as we will discuss in detail below , the magnetic structure changes in applied fields , but this effect shows a very strong field - hysteresis . to avoid this complication all the measurements
were done using field - cooling .
the sample was first warmed up to @xmath28 k , well above the temperature of magnetic ordering @xmath29 k,@xcite the desired strength and direction of the magnetic field were set , and only then was the sample brought down to base temperature , @xmath30 k , where most of the measurements were performed . at all times
the measurements were done in 3-axis mode to reduce the signal to background ratio .
@xmath31 collimations were used with pyrolitic graphite ( pg ) ( 002 ) reflections for monochromator and analyzer , with @xmath32 mev neutron energy and a pg filter positioned in front of the sample .
the crystal and magnetic structures of ba@xmath11cuge@xmath0o@xmath1 are discussed in detail in our previous publications on the subject,@xcite and only the most essential features are reviewed here .
the magnetcic cu@xmath33 ions form a square lattice in the @xmath2 tetragonal plane of the crystal .
the principal axes of this square lattice , hereafter referred to as the @xmath34 and @xmath35 axis , are along the @xmath22 $ ] and @xmath36 $ ] directions , respectively . to complete the coordinate system we shall choose the @xmath37 axis to run along @xmath38 $ ] . in the ordered phase ( below @xmath29 k ) the spins lie in the @xmath39 plane and the magnetic propagation vector
is @xmath40 , where @xmath41 .
the magnetic structure is a distortion of a nel spin arrangement : a translation along @xmath42 induces a spin rotation by an angle @xmath43 ( relative to an exact antiparallel alignment ) in the @xmath39 plane . along the @xmath44 $ ] direction
nearest - neighbor spins are perfectly antiparallel .
nearest - neighbor spins from adjacent cu - planes are aligned parallel to each other .
obviously , domains with propagation vectors @xmath45 are equivalent and always equally represented in a zero - field - cooled sample .
this can be seen in fig .
[ cont](a ) which shows a contour plot of elastic intensities measured at @xmath30 k on a mesh of points around the @xmath46 position in zero field .
we clearly see 4 magnetic reflections symmetrically grouped around the af zone center .
the two pairs of peaks , @xmath47 and @xmath48 , correspond to the two magnetic domains .
the propagation vector @xmath49 , measured relative to the af zone center , is strictly along the @xmath22 $ ] ( or @xmath44 $ ] ) direction .
the weak feature around @xmath46 is an artifact that is temperature - independent and was previously identified as a multiple scattering peak.@xcite cooling the sample in a very weak magnetic field ( @xmath50 mt ) always produces a single - domain structure .
when a magnetic field is applied along the @xmath51 $ ] direction the diffraction pattern begins to change .
the propagation vector @xmath49 starts to rotate in reciprocal space towards the direction of applied field [ fig .
[ cont](b ) ] , eventually becoming perfectly aligned with it [ fig .
[ cont](c ) ] .
note that all this happens in rather small fields , less than @xmath52 t. this is to be compared with the field along the @xmath13 axis , @xmath53 t , that is required to induce the ci transition studied previously . since @xmath54 , we can conclude that for a horizontally applied field the rotation of @xmath55 occurs in fields that are an order of magnitude smaller than the dzyaloshinskii energy . in the following discussion it is convenient to introduce notations for some angles in our experiment ( fig .
[ notation ] ) .
we shall denote by @xmath21 the angle between the magnetic field @xmath4 and the @xmath22 $ ] direction in the crystal .
@xmath56 will stand for the angle between @xmath57 $ ] and @xmath49 . using @xmath22 $ ] as
a reference is very convenient since i ) @xmath58 for @xmath59 and ii ) the principal axes of the square - lattice arrangement of the magnetic cu sites in ba@xmath11cuge@xmath0o@xmath1 are along @xmath22 $ ] and @xmath44 $ ] , respectively . in these terms , for @xmath60 , as @xmath61 is increased , @xmath56 continuously increases from @xmath62 to @xmath63 . for several directions of the magnetic field
we have performed careful measurements of @xmath64 .
the results are summarized in fig .
[ angles ] .
measurements with @xmath65 were done on the magnetic satellites around the @xmath46 af zone - center , and those around @xmath66 were used for higher values of @xmath21 .
this was necessary to work around the geometrical constraints set by the cryostat windows .
as mentioned , for @xmath60 the propagation vector starts out at @xmath58 for @xmath59 and rotates all the way towards the field direction ( @xmath67 ) , as in the case of fege . for arbitrary field direction
though , while @xmath56 levels off above @xmath68 t , it does so at a smaller value , i.e. , before @xmath49 reaches the direction of applied field .
if we plot @xmath56 measured at a relatively high field ,
@xmath69 t , we find that the saturation value for @xmath56 is always @xmath70 ( fig .
[ linear ] ) .
thus at high fields the @xmath46 vector always bisects the angle formed by the magnetic field @xmath4 and the propagation vector @xmath49 .
note that in all cases we have @xmath71 .
if the magnetic field forms a smaller angle with the @xmath22 $ ] direction , the magnetic domain with propagation vector @xmath72 , @xmath73 is destroyed , and the diffraction intensity completely shifts over into the @xmath74 domain , returning us to the situation when @xmath21 is effectively greater than @xmath63 .
this fact has been verified experimentally .
an important experimental result is that for all directions and of the applied magnetic field the length of the magnetic propagation vector @xmath75 is only weakly field - dependent .
this can be seen in fig .
[ trajectory ] that shows the trajectory traced by the propagation vector in reciprocal space as the magnetic field is increased .
for @xmath60 this trajectory is an almost perfect circular arc that starts out at @xmath58 and continues all the way to @xmath76 [ fig .
[ trajectory](a ) ] . for arbitrary @xmath21
we see that @xmath77 is indeed slightly field - dependent and the trajectory has a characteristic s - shape .
this , however , is a weak effect , barely detectable with the precision of our experiment , limited by the @xmath78-resolution of the spectrometer .
having discussed the field - dependence of the propagation vector , we now turn to that of the spin orientation . typically to solve the magnetic structure one has to measure the intensities of several bragg peak . in the present experiment , however , we did not have this luxury .
the severe geometrical constraints imposed by the narrow cryostat windows eliminate simultaneous access to several magnetic peaks .
many of these could be observed by rotating the sample within the magnet , however , in this case the direction of the magnetic field relative to the sample is changed . only in the case of @xmath60 , thanks to the presence of two orthogonal pairs of windows , could we simultaneously observe more than one set of magnetic satellites , namely those around @xmath46 and @xmath79 .
the intensities of these reflections are plotted against @xmath61 in fig .
[ intensity ] for @xmath60 .
these data are not sufficient to independently determine the magnetic structure for each value of applied field .
we can however assume that for all cases , just as for @xmath59 , the spin arrangement is a flat magnetic spiral , with the normal to the spin rotation plane confined in the @xmath2 plane of the crystal .
the assumption is quite reasonable if we consider that in our experiments @xmath80 t and @xmath81 mev , so the zeeman energy is at least an order of magnitude smaller than the exchange energy ( @xmath82 mev ) .
a slight conical distortion of the planar spiral structure is of course inevitable in finite magnetic fields , but the tilt towards the direction of @xmath4 will remain very small . if we denote by @xmath83 the angle between @xmath22 $ ] and @xmath84 , the normal to the spin rotation plane ( @xmath85 for @xmath59 ) , the intensities of magnetic satellites around @xmath46 and @xmath79 are determined by the spin polarization factors in the neutron diffraction cross - section : @xmath86 from fig . [ intensity ] we see that at high fields @xmath87 increases by roughly a third of its original value , while @xmath88 decreases by roughly the same amount . this observation is consistent with @xmath89 at high fields , as could be expected : a sufficiently strong magnetic field will always align the spin rotation plane perpendicular to itself , independent of the field direction . for arbitrary @xmath21
we therefore expect @xmath90 in sufficiently high fields .
experimentally , for all directions of magnetic field studied , this rule was found to be consistent with the observed intensity increase in the satellites around @xmath46 or @xmath66 that occurs upon increasing the magnetic field from @xmath59 to @xmath69 t. as will be discussed in detail in the next section , theory predicts that for _ arbitrary _ values of @xmath61 one has @xmath91 since @xmath64 is directly measured in our experiments , we can plot the expected field dependencies of @xmath87 and @xmath88 using eqs.([inten ] ) and ( [ rule ] ) .
these are shown in solid lines in fig .
[ intensity ] and are reasonably consistent with experimental data
. we can now summarize all the experimental results in a short list of `` rules '' : i ) as the magnetic field is increased , @xmath49 rotates in reciprocal space for small @xmath61 , but slows down and eventually stops for @xmath92 t ; ii ) @xmath75 is only weakly field - dependent ; iii ) at high fields the @xmath51 $ ] direction bisects the angle formed by the vectors @xmath4 and @xmath49 ; iv ) at high fields @xmath84 is pointing along the direction of applied field ; and v ) it appears that for arbitrary @xmath61 the @xmath51 $ ] direction always bisects the angle formed by the vectors @xmath49 and @xmath84 .
we shall now demonstrate that the observed behavior can be well understood using a generalization of the approach that we have previously employed to quantitatively analyze the dzyaloshinskii transition in ba@xmath11cuge@xmath0o@xmath1 .
while in our previous studies the direction of the propagation vector remained constant , which enabled us to use an effectively 1-dimensional model , in the present experiment @xmath49 rotates in the @xmath2 plane , and we have to extend our expression for the free energy of the spin structure to work in 2 dimensions .
in general , any almost - antiferromagnetic spin structure can be described in terms of the unit vector @xmath93 of local staggered magnetization .
this is the continuous - limit , when nearest- neighbor spins are almost antiparallel and , in consequence , the magnetic propagation vector is close to that of a nel structure . for a `` relativistic '' spiral , such as the one in ba@xmath11cuge@xmath0o@xmath1
, the continuous approximation is expected to work well when @xmath94 , @xmath95 , a condition well satisfied in our experiments .
the simplest form of the free energy that for @xmath59 gives a planar spin spiral with propagation vector along @xmath34 and the spins rotating in the @xmath96 plane , is @xmath97 , where @xmath98 is the temperature dependent spin stiffness ( see refs . ) , @xmath99 ( @xmath100 being the lattice constant ) is the nearest neighbor cu - cu distance , and @xmath101 is the angle by which spins are rotated at one step of the spiral . for the `` relativistic '' spiral in ba@xmath11cuge@xmath0o@xmath1 , @xmath102 . of course
, on a square lattice the spiral can equally well propagate in the direction of the symmetrically equivalent @xmath35-axis ( with spins rotating in the @xmath103 plane ) , and a similar term where @xmath35 is interchanged with @xmath34 has to be added to the free energy functional . the purely ferromagnetic interactions between the cu - layers in ba@xmath11cuge@xmath0o@xmath1 ( ref . ) can be accounted for by including the term @xmath104 , where @xmath105 is the spin stiffness anisotropy factor , @xmath106 for ba@xmath11cuge@xmath0o@xmath1 .
finally , the interaction with external magnetic field is given by the zeeman term @xmath107 $ ] ( for details see refs . ) .
in this last expression @xmath108 and @xmath109 are the longitudinal and transverse local staggered susceptibilities of the almost af spin structure , respectively . for the classical spin model at @xmath110 one
gets @xmath111 and @xmath112 .
combining all of the above terms gives us the simplest expression for the free energy per cu - plane , that is consistent with symmetric properties of the system and gives the `` right answer '' for @xmath59 : @xmath113 \label{free}\end{aligned}\ ] ] let us first consider the system in the absence of a magnetic field .
we want to find a solution for a _ general _ direction of the propagation vector in the @xmath114 plane . for the case
when @xmath49 forms the angle @xmath56 with the @xmath34-axis it is convenient to change the coordinate system to @xmath115 with @xmath116 along the propagation vector , @xmath117 perpendicular to @xmath116 in @xmath114 plane , and @xmath118 .
the derivatives and unit vectors are changed according to : @xmath119 with these expressions eq .
( [ free ] ) after some algebra can be rewritten as : @xmath120 \nonumber\end{aligned}\ ] ] for @xmath59 , since we have selected @xmath116 to be the direction of the propagation vector , @xmath121 , while @xmath122 .
it is straightforward to verify that the above expression is minimized when i ) @xmath93 is periodic with the period @xmath123 and ii ) along its vector of propagation @xmath124 uniformly rotates in a plane that is perpendicular to @xmath125 , i.e. , the `` bisection rule ''
@xmath126 formulated in the previous section is satisfied .
in zero external field all spiral structures that conform with this bisection rule are degenerate , i.e. , they have the _ same _ free energy .
it is now easy to understand what happens in the case @xmath127 .
from all the possible spiral structures , energetically degenerate at @xmath59 , the system will pick the one that takes the most advantage of the zeeman energy @xmath128 , namely that which has its spin plane normal to the field direction .
independent of the value of @xmath61 ( always assuming @xmath129 ) , instead of two equivalent domains seen at @xmath59 one gets a single domain with @xmath130 and @xmath90 .
the particular case of @xmath131 is of special interest . in this case
the propagation vector is also directed along @xmath132 and one has a `` screw - type '' spiral with all spins perpendicular to the propagation axis .
such a `` screw - type '' structure is realized in mnsi and fege for arbitrary direction of @xmath49 . while the above result accounts for the experimentally observed behavior for @xmath92 t , we still have to explain why the rotations of the spin plane and the propagation vector are continuous and in small fields some intermediate structure is realized .
obviously this is due to some anisotropy effects that pick the propagation vector along @xmath22 $ ] ( or @xmath44 $ ] ) and the normal to the spin plane along @xmath36 $ ] ( or @xmath22 $ ] ) for @xmath59 in the first place .
the most likely source of anisotropy is spin - orbital interaction .
for the present discussion however , the actual origin of magnetic anisotropy is of little importance : we can take it into account by introducing a phenomenological term into eq .
( [ free ] ) .
this is done under the assumption that the anisotropy energy @xmath133 , as well as the zeeman energy @xmath134 , are much smaller than the energy scales of dzyaloshinskii or exchange interactions .
neither magnetic field , nor anisotropy can distort any of the planar spiral structures in this limit .
their only effect is to pick the one that gives the greatest gain in zeeman and anisotropy energies .
@xmath133 can now be written as a function of @xmath135 or @xmath56 , which is in essence the same thing , since for all the structures we are dealing with @xmath136 . since at @xmath59 we know from experiment that @xmath85,@xcite the anisotropy term must be a minimum at this point .
it must also comply with the 4-fold symmetry of the crystal , i.e. , it must be invariant under @xmath137 .
in the most general case it is written as @xmath138 .
it is also convenient to rewrite the zeeman energy in terms of @xmath56 and @xmath21 : @xmath139 . minimizing @xmath140 with respect to @xmath135
we obtain : @xmath141 or , substituting @xmath126 , @xmath142 this expression gives us the direction of the propagation vector and the orientation of the spin rotation plane for arbitrary @xmath61 .
we can now verify that the analytical results obtained in the previous section are consistent with our experiments on ba@xmath11cuge@xmath0o@xmath1 . to begin with
, all the approximations that were made in the above calculations are justified .
indeed , the anisotropy energy in ba@xmath11cuge@xmath0o@xmath1 corresponds to magnetic fields of several tenths of a tesla , fields in which the propagation vector actively rotates .
the energy of dzyaloshinskii interactions , as previously mentioned , corresponds to fields of the order of @xmath143 t , while the exchange energy is roughly 10 times as large .
equation ( [ answer ] ) can be directly used to fit the experimentally measured @xmath64 .
the solid lines in fig .
[ angles ] are the result of a _ global _ fit of this expression to all our data collected at different @xmath21 .
the only adjustable parameter was the anisotropy coefficient @xmath144 , while we assumed @xmath145 .
the refined value for @xmath146 is @xmath147 t@xmath148 .
the accuracy of the single - parameter fit is quite remarkable .
it is interesting to note that our theory predicts qualitatively different behaviors for the cases of @xmath60 and @xmath149 , which is particularly easy to understand in the case @xmath150 . for @xmath60 eq .
( [ answer ] ) turns into @xmath151 .
we see that @xmath64 changes continuously for @xmath152 , has a kink at this value and remains constant ( equal to @xmath63 ) above this threshold .
on the other hand , for @xmath153 the @xmath56 approaches @xmath70 asymptotically in high fields , @xmath64 is a smooth function and there is no threshold field .
precisely this kind of behavior is observed in experiment ( fig . [ angles ] ) .
the limited data that we have for the orientation of the spin plane are also totally consistent with theory .
the only effect that our theoretical model fails to account for is the observed slight variation of @xmath75 .
this phenomenon may be a result of corrections to the continuous - limit approximation that we have ignored .
alternatively , it may be a purely quantum effect , related to the small value ( @xmath154 ) of spins involved and the quasi-2-dimensional nature of the system .
again we emphasize the key difference between fege and mnsi on one side , and ba@xmath11cuge@xmath0o@xmath1 on the other . in the former two systems the dzyaloshinskii vector is pointing along the bond between interacting spins , while in ba@xmath11cuge@xmath0o@xmath1 @xmath7 is orthogonal to this bond . as a result , in fege the normal to the spin plane @xmath84 , and the propagation vector @xmath49 are always collinear .
in contrast , in ba@xmath11cuge@xmath0o@xmath1 this colinearity is replaced by the `` bisection rule '' : @xmath49 and @xmath84 form equal angles with the @xmath100 axis of the crystal , but are not colinear . in conclusion , we believe that the present work gives a fairly complete picture of the rather exotic _ static _ magnetic properties of the square - lattice dzyaloshinskii - moriya antiferromagnet ba@xmath11cuge@xmath0o@xmath1 .
much remains to be learned in the study of magnetic critical behavior , as well as the dynamical properties of the soliton lattice , realized in finite fields applied along the @xmath13-axis .
this study was supported in part by nedo ( new energy and industrial technology development organization ) international joint research grant and the u.s .
-japan cooperative program on neutron scattering .
we thank scott moore and brent taylor for expert technical assistance .
oak ridge national laboratory is managed for the u.s .
d.o.e . by lockheed martin energy research corporation under contract de - ac05 - 96or22464 .
work at brookhaven national laboratory was carried out under contract no .
de - ac02 - 76ch00016 , division of material science , u.s . d.o.e . | the magnetic structure of ba@xmath0cuge@xmath0o@xmath1 is investigated by neutron diffraction in magnetic fields applied along several directions in the @xmath2 plane of the crystal . in relatively weak fields , @xmath3 t ,
the propagation vector of the spin - spiral rotates to form a finite angle with the field direction .
this angle depends on the orientation of @xmath4 itself .
the rotation of the propagation vector is accompanied by a re - orientation of the plane of spin rotation in the spiral .
the observed behaviour is well described by a continuous - limit form of a free energy functional that includes exchange and dzyaloshinskii - moriya interactions , as well as the zeeman energy and an empirical anisotropy term . |
conventional parton densities are defined in terms of an integral over all transverse momentum and virtuality for a parton that initiates a hard scattering .
while such a definition of an integrated parton density is appropriate for very inclusive quantities , such as the ordinary structure functions @xmath0 and @xmath1 in dis , the definition becomes increasingly unsuitable as one studies less inclusive cross sections . associated with the use of integrated parton densities
are approximations on parton kinematics that can readily lead to unphysical cross sections when enough details of the final state are investigated .
we propose that it is important to the future use of pqcd that a systematic program be undertaken to reformulate factorization results in terms of fully unintegrated densities , which are differential in both transverse momentum and virtuality .
these densities are called `` doubly unintegrated parton densities '' by watt , martin and ryskin @xcite , and `` parton correlation functions '' by collins and zu @xcite ; these authors have presented the reasoning for the inadequacy , in different contexts , of the more conventional approach .
the new methods have their motivation in contexts such as monte - carlo event generators where final - state kinematics are studied in detail .
even so , a systematic reformulation for other processes to use unintegrated densities would present a unified methodology .
these methods form an extension of @xmath2-factorization , which has so far been applied in small-@xmath3 processes and , as the css formalism , in the transverse - momentum distribution of the drell - yan and related processes .
densities for the @xmath4 of @xmath5 pairs in photoproduction , and for the @xmath6 .
( c ) and ( d ) : comparison of use of @xmath2 densities and full simulation . ] @xmath7{l } \includegraphics[scale=0.5]{nlo - box } \end{tabular } } \;+\ ; { \begin{tabular}[c]{l } \includegraphics[scale=0.5]{nlo - box - crossed } \end{tabular } } $ ] the problem that is addressed is nicely illustrated by considering photoproduction of @xmath5 pairs . in figs .
[ fig : ccbar.1 ] , we compare three methods of calculation carried out within the cascade event generator @xcite : * use of a conventional gluon density that is a function of parton @xmath3 alone . * use of a @xmath2 density that is a function of parton @xmath3 and @xmath2 .
these are the objects usually called `` unintegrated parton densities '' . * use of a doubly unintegrated density that is a function of parton @xmath3 , @xmath2 and virtuality , that is , of the complete parton 4-momentum .
the partonic subprocess in all cases is the lowest order photon - gluon - fusion process @xmath8 ( fig .
[ fig : gamma.g.fusion ] ) .
two differential cross sections are plotted : one as a function of the transverse momentum of the @xmath5 pair , and the other as a function of the @xmath6 of the pair . by @xmath6
is meant the fractional momentum of the photon carried by the @xmath5 pair , calculated in the light - front sense as @xmath9 here @xmath10 is the electron beam energy and the coordinates are oriented so that the electron and proton beams are in the @xmath11 and @xmath12 directions respectively . in the normal parton model approximation for the hard scattering ,
the gluon is assigned zero transverse momentum and virtuality , so that the cross section is restricted to @xmath13 and @xmath14 , as shown by the solid lines in fig .
[ fig : ccbar.1](a , b ) .
when a @xmath2 dependent gluon density is used , quite large gluonic @xmath2 can be generated , so that the @xmath15 distribution is spread out in a much more physical way , as given by the dashed line in fig .
[ fig : ccbar.1](a ) . but
as shown in plot ( b ) , @xmath6 stays close to unity .
neglecting the full recoil mass @xmath16 ( produced in the shaded subgraph in fig [ fig : gamma.g.fusion ] ) is equivalent of taking @xmath17 with @xmath18 being the virtuality of the gluon in fig .
[ fig : gamma.g.fusion ] , @xmath2 its transverse momentum and @xmath3 its light cone energy fraction .
this gives a particular value to the gluon s @xmath19 .
when we also take into account the correct virtuality of gluon , there is no noticeable change in the @xmath15 distribution see fig .
[ fig : ccbar.1](c ) ( dashed line ) since that is already made broad by the transverse momentum of the gluon .
but the gluon s @xmath19 is able to spread out the @xmath6 distribution , as in fig .
[ fig : ccbar.1](d ) with the dashed line .
this is equivalent with a proper treatment of the kinematics and results in @xmath20 , where @xmath16 is the invariant mass of the beam remnant , the part of the final state in the shaded blob in fig .
[ fig : gamma.g.fusion ] .
this change can be particularly significant if @xmath3 is not very small .
note that if partons are assigned approximated 4-momenta during generation of an event in a mc event generator , the momenta need to be reassigned later , to produce an event that conserves total 4-momentum .
the prescription for the reassignment is somewhat arbitrary , and it is far from obvious what constitutes a correct prescription , especially when the partons are far from a collinear limit
. a treatment with fully unintegrated pdfs should solve these problems .
if , as we claim , an incorrect treatment of parton kinematics changes certain measurable cross sections by large amounts , then we should verify directly that there are large discrepancies in the distributions in partonic variables themselves .
we see this in fig .
[ fig : ccbar.2 ] . graph ( a ) plots the gluonic transverse momentum divided by the charm - pair mass . as is to be expected , the typical values are less than one , but there is a long tail to high values .
but the use of full parton kinematics does not have much of an effect , the unintegrated parton distributions already providing realistic distributions in transverse momentum .
on the other hand , a simple collinear approximation for showering sets the remnant mass , @xmath16 , to zero . as can be seen from the formulae for the gluon virtuality , this only provides a good approximation to the gluon kinematics if @xmath16 is much less than @xmath2 . in reality , as we see from graph ( b ) , there is a long tail to large values of @xmath21 , and the tail is much bigger when correct kinematics are used . a more correct comparison uses @xmath22 , with an extra factor of @xmath3 .
even then , there is a large effect , shown in graph ( c ) .
the vertical scale is logarithmic , so the absolute numbers of events are relatively small , but the tail is broad . finally , graph ( d ) shows that the distribution in @xmath23 itself is very broad , extending to many tens of gev .
this again supports the argument that unless a correct treatment of parton kinematics is made , very incorrect results are easily obtained .
it is important to note that , for the cross sections themselves , the kinematic variables used in fig .
[ fig : ccbar.1 ] are normal ones that are in common use .
many other examples are easily constructed .
clearly , the use of the simple parton - model kinematic approximation gives unphysically narrow distributions .
the correct physical situation is that the gluon surely has a distribution in transverse momentum and virtuality , and for the considered cross sections neglect of parton transverse momentum and virtuality leads to wrong results .
it is clearly better to have a correct starting point even at lo , for differential cross sections such as we have plotted .
the standard treatment of parton kinematics involves replacing the incoming parton momentum @xmath24 by its plus component only : @xmath25 .
there are actually two parts to this .
the first is to neglect the @xmath26 and transverse components of @xmath24 with respect to the large transverse momenta in the calculation of the numerical value of the hard - scattering amplitude ; this is a legitimate approximation , readily corrected by higher order terms in the hard scattering .
the second part is to change the kinematics of the final - state particles , @xmath27 and @xmath28 , so that their sum is @xmath29 plus the approximated gluon momentum .
it is this second part that is problematic , for it amounts to the replacement of the momentum conservation delta function @xmath30 by @xmath31 .
these delta - functions are infinitely different , point - by - point .
only when integrated with a sufficiently smooth test function can they be regarded as being approximately the same , as in a fully inclusive cross section . in an event generator ,
the effect is to break momentum conservation , which is restored by an ad hoc correction of the parton kinematics .
note that the change of parton kinematics is only in the hard scattering , i.e. , in the upper parts of the graphs .
parton kinematics are left unaltered within the parton density part , and the integrals over @xmath2 and virtuality are part of the standard definition of integrated pdfs .
the situation is ameliorated by inclusion of nlo terms , and perhaps also by some kind of resummation .
but these do not correct the initial errors in the approximation , and lead to a very restricted sense in which the derivation of the cross section can be regarded as valid .
furthermore , when much of the effect of nlo terms is to correct the kinematic approximations made in lo , this is an inefficient use of the enormous time and effort going into nlo calculations .
a case in point is the bfkl equation , where @xmath32 of the ( large ) nlo corrections are accounted for @xcite by the correction of kinematic constraints in the lo calculation .
the physical reasoning for the absolute necessity of fully unintegrated densities is , we believe , unquestionable .
therefore it is highly desirable to reformulate perturbative qcd methods in terms of doubly unintegrated parton densities from the beginning .
a full implementation will be able to use the full power of calculations at nlo and beyond . among other things , a full implementation , as in @xcite ,
will provide extra factorization formulae for obtaining the values of the unintegrated densities at large parton transverse momentum and virtuality .
this will incorporate all possible perturbatively calculable information , so that the irreducible nonperturbative information , that must be obtained from data , will be at low transverse momentum and virtuality .
in addition , the implementation will quantify the relations to conventional parton densities . with the most obvious definitions , the integrated pdfs are simple integrals of the unintegrated densities .
however , in full qcd a number of modifications are required @xcite , so that the relations between integrated and unintegrated pdfs are distorted . the fact that we propose new and improved methods does not invalidate old results in their domain of applicability .
the work of watt , martin and ryskin , and of collins and zu provides a start on this project ; but much remains to be done to provide a complete implementation in qcd ; for example , there is as yet no precise , valid , and complete gauge - invariant operator definition of the doubly unintegrated densities in a gauge theory .
the outcome of such a program should have the following results : 1 .
lowest order calculations will give a kinematically much more realistic description of cross sections .
this may well lead to nlo and higher corrections being much smaller numerically than they typically are at present , since the lo description will be better .
2 . it will also obviate the need for separate methods ( resummation or the css technique ) , which are currently applied to certain individual cross sections like the transverse - momentum distribution for the drell - yan process .
all these and others will be subsumed and be given a unified treatment .
3 . a unified treatment will be possible for both inclusive cross sections using fixed order matrix element calculations and for monte - carlo event generators .
4 . for a long - term theoretical perspective
, the doubly unintegrated distributions will interface to methods of conventional quantum many - body physics much more easily than regular parton densities , whose definitions are tuned to their use in ultra - relativistic situations .
this program is , of course , technically highly nontrivial if it is to be used in place of conventional methods with no loss of predictive power .
a start is made in the cited work . among the main symptoms of the difficulties
are that the most obvious definition of a fully unintegrated density is a matrix element of two parton fields at different space - time points , which is not gauge - invariant .
it is often said that the solution is to use a light - like axial gauge @xmath33 .
however , in unintegrated densities , this leads to divergences see @xcite for a review and the definitions need important modification , in such a way that a valid factorization theorem can be derived .
we also have to ask to what extent factorization can remain true in a generalized sense .
hadron - hadron collisions pose a particular problem here , because factorization needs a quite nontrivial cancellation arising from a sum over final - state interactions .
this is not compatible with simple factorization for the exclusive components of the cross section , and makes a distinction between these processes and exclusive components of dis , for example .
this work is supported in part by the u.s . doe .
watt , g. and martin , a. d. and ryskin , m. g. , eur .
j. * c31 * , 73 ( 2003 ) watt , g. and martin , a. d. and ryskin , m. g. , phys . rev . * d70 * , 014012 ( 2004 ) collins , john c. and zu , xiaomin , jhep * 03 * , 059 ( 2005 ) h. jung and g. salam * 19 * , 351 ( 2001 ) . h. jung , comp .
. comm . * 143 * , 100 ( 2002 ) . `
http://www.quark.lu.se/~hannes/cascade/ ` kwiecinski , j. and martin , alan d. and outhwaite , j. j. , eur .
j. * c9 * , 611 ( 1999 ) collins , john c. , acta phys . polon . * b34 * , 3103 ( 2003 ) | associated with the use of conventional integrated parton densities are kinematic approximations on parton momenta which result in unphysical differential distributions for final - state particles .
we argue that it is important to reformulate perturbative qcd results in terms of fully unintegrated parton densities , differential in all components of the parton momentum . |
the notion of gravitational mass for a composite body is known to be non - trivial in general relativity and related to several paradoxes . the role of the classical virial theorem in establishing of the equivalence between averaged over time
gravitational mass and energy is discussed in detail in refs.[1,2 ] for different types of classical composite bodies . in particular , for electrostatically bound two bodies , it is shown that the gravitational field is coupled to a combination @xmath0 , where @xmath1 is the kinetic energy and @xmath2 the potential energy . since the classical virial theorem states that the following time average is equal to zero , @xmath3
, we can conclude that averaged over time gravitational mass is proportional to the total amount of energy , @xmath4 [ 1,2 ] : @xmath5
the main goal of our paper is to study the quantum problem concerning the passive gravitational mass of a composite body . as the simplest example , we consider a hydrogen atom in the earth s gravitational field , where we take into account only kinetic and coulomb potential energies of an electron in a curved spacetime .
we claim three main results in the paper ( see also refs . [ 3,4 ] ) .
our first result is that the equivalence between passive gravitational mass and energy in the absence of gravitational field survives , at the macroscopic level , in the quantum case .
more strictly speaking , we show that the expectation value of the mass is equal to @xmath6 for stationary quantum states due to the quantum virial theorem [ 5 ] .
our second result is a breakdown of the equivalence between passive gravitational mass and energy , at the microscopic level , for stationary quantum states due to the fact that passive gravitational mass operator does not commute with energy operator . as a result
, there exist a non - zero probability that a measurement of passive gravitational mass can yield a value , which is different from @xmath6 , given by the einstein s equation .
our third result is a suggestion of a realistic experiment to detect this inequivalence by measurement of electromagnetic radiation , emitted by a macroscopic ensemble of hydrogen atoms , supported by and moving in the earth s gravitational field , using spacecraft or satellite .
here , we derive the lagrangian and hamiltonian of a hydrogen atom in the earth s gravitational field , taking into account couplings of kinetic and potential coulomb energies of an electron with a weak gravitational field .
note that we keep only terms of the order of @xmath7 and disregard all tidal effects .
therefore , we can write the interval , using a weak field approximation [ 6 ] : @xmath8 where @xmath9 is the gravitational constant , @xmath10 is the velocity of light , @xmath11 is the earth mass , @xmath12 is the distance between center of the earth and center of mass of a hydrogen atom ( i.e. , a proton ) .
then , in the local proper spacetime coordinates , which in the first approximation in @xmath13 are @xmath14 the classical lagrangian and action of an electron have the following standard forms : @xmath15 where @xmath16 is the bare electron mass , @xmath17 and @xmath18 are the electron charge and velocity , respectively ; @xmath19 is a distance between electron and proton .
it is possible to show that the lagrangian ( 4 ) can be rewritten in coordinates @xmath20 as @xmath21 which corresponds to the following hamiltonian : @xmath22
the hamiltonian ( 6 ) can be quantized by substituting a momentum operator , @xmath23 , instead of canonical momentum , @xmath24 : @xmath25 where we introduce passive gravitational mass operator of an electron to be proportional to its weight operator in a weak gravitational field ( 2 ) , @xmath26 note that the hamiltonian ( 7),(8 ) contains the virial contribution ( i.e. , the last term ) to the mass operator and , therefore , does not commute with electron energy operator , taken in the absence of the field .
below , we discuss some consequences of eqs.(7),(8 ) .
suppose that we have a macroscopic ensemble of hydrogen atoms with each of them being in a ground state with energy @xmath27 .
then the expectation value of the electron mass operator ( 8) per atom is @xmath28 where the third term in eq.(9 ) is zero in accordance with the quantum virial theorem [ 5 ] .
therefore , we conclude that the equivalence between passive gravitational mass and energy survives at a macroscopic level for stationary quantum states . let us discuss how eqs.(7),(8 ) break the equivalence between passive gravitational mass and energy at a microscopic level . here
, we illustrate the above mentioned inequivalence , using the following thought experiment .
suppose that at @xmath29 we create a ground state wave function of a hydrogen atom , corresponding to the absence of gravitational field , @xmath30 in gravitational field ( 2 ) , wave function ( 10 ) is not anymore a ground state .
it is possible to show that a general solution of the schrdinger equation , corresponding to the hamiltonian ( 7),(8 ) , can be written as @xmath31 \exp[-i m_e c^2 ( 1+\phi / c^2 ) t/\hbar ] \exp[-i e_n(1+\phi / c^2 ) t/\hbar ] .\ ] ] we wish to draw attention to the fact that wave function ( 11 ) is a series of eigenfunctions of passive gravitational mass operator ( 8) .
@xmath32 is a normalized wave function of an electron in a hydrogen atom in the absence of gravitational field , corresponding to energy @xmath33 . in accordance with quantum mechanics , probability that , at @xmath34 , an electron occupies excited state with energy @xmath35 is @xmath36
d^3 { \bf r } = - ( \phi / c^2 ) \int \psi^*_1(r ) r \psi'_n(r ) d^3 { \bf r } , \nonumber\\ \int \psi^*_1(r ) r \psi'_n(r ) d^3 { \bf r } = v_{n,1}/ ( \hbar \omega_{n,1 } ) , \ \hbar \omega_{n,1 } = e_n - e_1 , \
n \neq 1 , \nonumber\\ v_{n,1}= \int \psi^*_1(r ) \hat v({\bf r } ) \psi_n(r ) d^3 { \bf r } , \ \
\hat v({\bf r } ) = 2 \frac{\hat { \bf p}^2}{2 m_e } - \frac{e^2}{r}.\end{aligned}\ ] ] let us discuss eq.(12 ) .
we stress that it directly demonstrates that there is a finite probability , @xmath37 ^ 2 \ , \
n \neq 1,\ ] ] that , at @xmath34 , an electron occupies n - th ( @xmath38 ) energy level , which breaks the expected einstein s equation , @xmath39 .
in fact , this means that quantum measurement of passive gravitational mass in a quantum state with a definite energy ( 10 ) gives the following quantized values : @xmath40
here , we describe a realistic experiment : we consider a hydrogen atom being in its ground state at @xmath29 , located at distance @xmath41 from the center of the earth and having the wave function : @xmath42 \exp[-im_ec^2(1+\phi'/c^2 ) t /\hbar ] \exp[-ie_1(1+\phi'/c^2)t/\hbar ] .\ ] ] the atom is supported by in the earth s gravitational field and moving from the earth at constant velocity , @xmath43 ( where @xmath44 is the fine structure constant ) , by spacecraft or satellite .
it is possible to show that the electron wave function and the time dependent perturbation for the hamiltonian ( 7)-(8 ) , in the coordinate system related to center of mass of the atom , can be expressed as @xmath45 \exp[-im_ec^2(1+\phi'/c^2 ) t /\hbar ] \nonumber\\ \exp[-ie_n(1+\phi'/c^2)t/\hbar ] , \end{aligned}\ ] ] @xmath46 we wish to draw attention that in a spacecraft ( satellite ) , which moves with constant velocity , gravitational force , which acts on each individual hydrogen atom , is compensated by some non - gravitational forces .
this causes very small changes of a hydrogen atom energy levels and is not important for our calculations .
therefore , the atoms do not feel directly gravitational acceleration , @xmath47 , but feel , instead , gravitational potential , @xmath48 . it is important
that , if excited levels of a hydrogen atom were strictly stationary , then probability to find the passive gravitational mass to be quantized with @xmath38 ( 14 ) would be @xmath49 ^ 2 \
, n \neq 1 , \
|\phi(r'+ut)| \ll |\phi(r')| , u \ll \omega_{n,1 } r , \ ] ] which coincides with eq .
( 13 ) . in reality
, the excited levels spontaneously decay with time and , therefore , it is possible to observe the quantization law ( 14 ) indirectly by measuring electromagnetic radiation from a macroscopic ensemble of the atoms . in this case , eq.(18 ) gives a probability that a hydrogen atom emits a photon with frequency @xmath50 during the time of the experiment . to estimate the probabilities ( 18 ) we use the following numerical values of the earth s mass , @xmath51 , and its radius , @xmath52 .
it is important that , although the probabilities ( 18 ) are small , the number of photons , @xmath53 , emitted by macroscopic ensemble of the atoms , can be large .
for instance , for 1000 moles of hydrogen atoms , @xmath53 is estimated as @xmath54 ^ 2 , \ \
n ( 2 \rightarrow 1 ) = 0.9 \times 10 ^ 8 , \end{aligned}\ ] ] which can be experimentally detected .
[ here , @xmath55 stands for a number of photons , emitted with frequency @xmath56 . ]
we are thankful to n.n .
bagmet ( lebed ) , v.a .
belinski , steven carlip , li - zhi fang , douglas singleton , and v.e .
zakharov for useful discussions .
this work was supported by the nsf under grant dmr-1104512 . | the simplest quantum composite body , a hydrogen atom , is considered in the presence of a weak external gravitational field .
we define an operator for the passive gravitational mass of the atom in the post - newtonian approximation of the general relativity and show that it does not commute with its energy operator .
nevertheless , the equivalence between the expectation values of the mass and energy is shown to survive at a macroscopic level for stationary quantum states .
breakdown of the equivalence between passive gravitational mass and energy at a microscopic level for stationary quantum states can be experimentally detected by studying unusual electromagnetic radiation , emitted by the atoms , supported by and moving in the earth s gravitational field with constant velocity , using spacecraft or satellite |
for @xmath0 a parry number , boyd @xcite introduced the notion of the beta - conjugates of @xmath1 in the context of the rnyi - parry numeration system @xcite @xcite @xcite @xcite .
as he has shown it in numerous examples , the investigation of beta - conjugates is an important question .
these beta - conjugates , up till now defined for parry numbers , are canonically associated to @xmath1 and to the dynamics of the beta - transformation . our aim is to show that their definition can be given in a larger context , namely for any algebraic number @xmath0 , and that the theory of puiseux provides a geometric origin to the beta - conjugates of @xmath1 ; for doing it , once @xmath1 is given by its minimal polynomial , we first put into evidence that a germ of curve at @xmath5 " does exist and develop new tools deduced from the canonical decomposition of this germ in order to express the beta - conjugates of @xmath1 in terms of the puiseux expansions @xcite @xcite of the germ . though the existence of this germ of curve was discovered by the author some years ago
, the present note is the first account on it and its potential applications .
it establishes a deep relation between the theory of singularities of curves in algebraic geometry and the dynamical system of numeration @xmath6 , t_{\beta})$ ] where @xmath0 is an algebraic number and @xmath7 is the beta - transformation .
the existence of this germ of curve brings new tools to the rnyi - parry numeration system , namely the puiseux series associated to the germ , and defines new directions of research for old questions .
for instance , if @xmath8 is a sequence of salem numbers which converges to a real number @xmath1 , then it is known @xcite that @xmath1 is a pisot or a salem number , but how is distributed the collection of the beta - conjugates and the galois conjugates of @xmath9 , with @xmath10 large enough , with respect to that of the limit @xmath1 ?
this question is merely a generalization of the classical question of how is distributed the collection of the galois conjugates of @xmath9 with respect to that of @xmath1 ? why should we add the beta - conjugates ? because a new phenomenon appears which generally does not exist with only the galois conjugates
: under some assumptions the collections of galois- and beta- conjugates may have equidistribution limit properties on the unit circle ( 3.6 in @xcite ) if the two collections of conjugates are simultaneously considered .
both collections of conjugates are expected to play a role in limit and dynamical properties of convergent sequences of real algebraic numbers @xmath11 in general .
a basic question is then to understand the role and the relative density of the beta - conjugates in this possible equidistribution process , in particular if the limit @xmath1 is an integer @xmath12 or is equal to @xmath13 ( context of the conjecture of lehmer ) . conversely the curve canonically associated with this numeration dynamical system is of interest for itself ( critical points , monodromy , ... ) .
it will be studied elsewhere . in this first contribution
we obtain useful expressions for the beta - conjugates as puiseux expansions of @xmath1 and of the minimal polynomial of @xmath1 , towards this goal . as usual
now we use the new terminology , which is in honor of w. parry . the old terminology used by w.
parry himself in @xcite transforms as follows : we now call _ parry number _ a @xmath1-number @xcite , and _ parry polynomial _ of a parry number @xmath1 the characteristic polynomial @xcite of the @xmath1-number @xmath1 . as previously a _
simple _ parry number @xmath1 is a parry number @xmath1 for which the rnyi @xmath1-expansion @xmath14 of unity is finite ( i.e. ends in infinitely many zeros ) .
the exact definitions are given in section [ s3 ] .
if @xmath1 is a parry number , the roots of the parry polynomial of @xmath1 , denoted by @xmath15 , are called the conjugates of @xmath1 .
a conjugate of @xmath1 is either a galois conjugate of @xmath1 or a beta - conjugate , if the collection of beta - conjugates of @xmath1 is not empty .
let @xmath0 be a real number and @xmath16 be the rnyi @xmath1-expansion of 1 .
since this rnyi @xmath1-expansion of 1 controls the language in base @xmath1 @xcite , the properties of the analytic function constructed from it , called parry upper function at @xmath1 , defined by @xmath17 , is of particular importance .
ito and takahashi @xcite have shown that the parry upper function at a parry number @xmath1 , of the complex variable @xmath18 , is related to the ( artin - mazur ) dynamical zeta function @xmath19 \mid t_{\beta}^n(x ) = x\}}{n } z^n\right)\ ] ] of the beta - transformation @xmath7 ( artin and mazur @xcite , boyd @xcite , flatto , lagarias and poonen @xcite , verger - gaugry @xcite @xcite ) .
namely , if @xmath1 is a nonsimple parry number , with @xmath20 ( where @xmath21 means infinitely repeated ) , @xmath22 where @xmath23 is the reciprocal of the parry polynomial @xmath24 of @xmath1 , of degree @xmath25 ( @xmath26 is the preperiod length and @xmath27 is the period length in @xmath14 , if @xmath1 is a nonsimple parry number , with the convention @xmath28 for a finite rnyi @xmath1-expansion of unity ( for @xmath1 a simple parry number ) , with the convention @xmath29 if @xmath14 is a purely periodic expansion @xcite ) ; if @xmath1 is a simple parry number , with @xmath30 , then @xmath31 the zeros of @xmath2 are the poles of @xmath4 .
the set of zeros of @xmath2 is the set @xmath32 of the reciprocals of the conjugates @xmath33 of @xmath1 .
the geometry of the conjugates @xmath33 of @xmath1 was carefully studied by solomyak @xcite @xcite : these conjugates all lie in solomyak s fractal @xmath34 , a compact connected subset of the closed disc @xmath35 in the complex plane ( figure [ fractalsolo ] ) , having a cusp at @xmath36 , a spike on the negative real axis , symmetrical with respect to the real line @xcite @xcite .
if @xmath0 is an algebraic number but not a parry number , some relations are expected between @xmath2 and @xmath4 , though not yet determined .
indeed , on one hand , @xmath2 is an analytic function on the open unit disc which admits @xmath37 as natural boundary by szeg - carlson - poly s theorem @xcite @xcite ; @xmath2 admits @xmath5 as zero of multiplicity one , which is its only zero in the interval @xmath38 . on the other hand @xmath4
is an analytic function defined on the open unit disc @xmath39 , which admits a nonzero meromorphic continuation on @xmath40 , by @xcite @xcite @xcite , or by baladi - keller s theorem 2 in @xcite . whether the zeros of @xmath2 correspond to poles of @xmath4 is unknown .
the behaviour of the dynamical zeta function @xmath4 on the unit circle remains unknown , i.e. we do not know whether @xmath37 is a natural boundary for @xmath4 or not .
but the multiplicity of the pole @xmath5 of @xmath4 is known to be one @xcite @xcite @xcite . for @xmath0 an algebraic number , as a consequence of theorem 1 in @xcite , the coefficients in obey the following asymptotics of growth ( pollicott , 5.2 in @xcite ) : for any @xmath41 there exist an integer @xmath42 and constants @xmath43 , with @xmath44 , and @xmath45 , such that @xmath46 \mid t_{\beta}^n(x ) = x\ } ~=~ \sum_{i=1}^{m } c_{i,\beta } \lambda_{i,\beta}^{n } + o((1+\delta)^n).\ ] ] in the case where @xmath0 is a parry number
, @xmath4 is a rational fraction and , from and , transforms into the following exact formula ( after pollicott , 1 in @xcite ) : @xmath47 \mid t_{\beta}^n(x ) = x\ } ~=~ \sum_{i=1}^{k } \ , ( \rho_i)^n - \sum_{i=1}^{d_p } \,(\beta^{(i)})^n,\ ] ] where @xmath48 is the collection of @xmath49-th roots of unity , @xmath50 if @xmath1 is nonsimple with @xmath20 and @xmath51 if @xmath1 is simple with @xmath14 of length @xmath26 ( i.e. @xmath30 ) .
moreover , @xmath1 is a perron number since it is a parry number ( lind , @xcite ) : hence the asymptotic growth of is dictated by the geometry and the moduli of the beta - conjugates of @xmath1 , all being algebraic integers lying in solomyak s fractal @xmath34 , of modulus less than or equal to @xmath52 , and by the geometry and the moduli of the galois conjugates of @xmath1 , all being less than @xmath1 , by definition .
our objective consists in showing that a germ of curve exists in a neighbourhood of the point @xmath53 in @xmath54 ( this point being the origin of this germ ) each time @xmath0 is a real algebraic number , that is , roughly speaking , a germ of curve located at the reciprocal @xmath5 of the base of numeration @xmath1 .
the construction of this germ of curve comes from a ( unique ) writting of the one - variable analytic function @xmath2 as a ( unique ) two - variable analytic function parametrized by @xmath55 and @xmath56 , where @xmath57 is the reciprocal of the minimal polynomial @xmath58 of @xmath1 : @xmath59 where @xmath60[z]$ ] , deg@xmath61deg @xmath1 , is convergent , with coefficients in @xmath62 , possibly in some cases in the algebraic number field @xmath63 , or in a finite algebraic extension of @xmath64 .
the existence of this germ of curve arises from the fact that @xmath0 is a real number which is an algebraic number , since it is constructed from the imposed parametrization @xmath65 , which makes use of the minimal polynomial of @xmath1 .
this parametrization of @xmath66 leads to the identity .
applying the theory of puiseux @xcite @xcite to provides a canonical decomposition of this germ into irreducible curves , conjugacy classes , as stated in theorem [ puiseuxfbeta ] .
this decomposition brings to light several new features of the parry upper function @xmath2 : \(i ) a new definition of the beta - conjugates of @xmath1 in terms of the puiseux expansions of the germ ( definition [ betaconjugatepuiseuxfbetanewdef ] ) , \(ii ) the explicit relations between the field of coefficients of the puiseux series of the germ @xmath67 , and the beta - conjugates , \(iii ) a product formula , as given by ; in particular , if @xmath1 is a parry number , from and , this product gives an analog of the euler product of the riemann zeta function for the dynamical zeta function @xmath4 , where the product is taken over the different rational conjugacy classes of the germ ( as given by ) .
in addition to the usual galois conjugation relating the roots of the minimal polynomial of @xmath1 , a new conjugation relation , called puiseux - conjugation " , among the beta - conjugates , is defined .
the reader accustomed to numeration systems and to the theory of puiseux for germs of curves can skip section [ s3 ] and section [ s4 ] to proceed directly to beta - conjugates in section [ s5 ] .
the present note finds its origin in @xcite , for the parametrization by @xmath68 , and in the two articles @xcite @xcite , for the idea of developping a two - variable analytic function canonically associated with the beta - transformation and the minimal polynomial of the base of numeration @xmath1 .
let us recall them . in theorem iv in @xcite , for constructing convergent families of salem numbers @xmath69
for which the limit is a ( nonquadratic ) pisot number @xmath70 , salem introduces polynomials of the following type @xmath71 where @xmath72 and @xmath73 is the minimal polynomial of the limit @xmath70
. we may consider @xmath74 in , in one or the other form , as parametrized by the couple @xmath75 ( ordered pair ) .
this parametrization , and its consequences , were developped and extended by boyd @xcite to a more general form , by adding ingeniously and in a
profitable " way a second variable @xmath76 , as follows @xmath77 with @xmath78 integers .
the advantage of introducing a second variable @xmath76 , as continuous parameter " , lies in the fact that an algebraic curve @xmath79 is associated to @xmath80 , with a finite number of branches and multiple points @xcite .
boyd @xcite shows that the existence of this curve gives a deep insight into the geometry of the roots of @xmath81 , for some values of @xmath76 , in particular those roots on the unit circle .
using these polynomials bertin and boyd @xcite explore the interlacing of the galois conjugates of salem numbers with the roots of associated polynomials ( theorem a and theorem b in @xcite ) .
a salem number is an algebraic integer @xmath11 for which all the galois conjugates lie in the closed unit disc , with at least one conjugate on the unit circle ; the degree of a salem number is even , greater than @xmath82 , and its minimal polynomial is reciprocal ( a salem number is galois - conjugated to its inverse ) @xcite .
a perron number is either 1 or an algebraic integer @xmath0 such that all its galois conjugates @xmath15 satisfy : @xmath83 for @xmath84 , if the degree of @xmath1 is denoted by deg@xmath85 ( with @xmath86 ) .
a pisot number @xmath1 is a perron number @xmath87 which has the property : @xmath88 for @xmath89 ( with @xmath86 ) .
let @xmath0 be a real number and define the beta - transformation @xmath90 \to [ 0,1 ] , x \to \{\beta x\}$ ] ( @xmath91 , resp .
@xmath92 , denotes the closest integer to the real number @xmath93 , @xmath94 , resp .
@xmath95 , and @xmath96 its fractional part ) .
denote @xmath97id , @xmath98 , and @xmath99 , @xmath100 ( the dependency of each @xmath101 to @xmath1 will not be indicated in the sequel ) .
the digits @xmath101 belong to the finite alphabet @xmath102 .
the rnyi @xmath1-expansion of unity is denoted by @xmath103 obtained by the greedy algorithm applied to @xmath13 by the successive negative powers of @xmath1 .
the set of successive iterates of @xmath13 under @xmath7 , hence the sequence @xmath104 , has the important property that it controls the admissibility of finite and infinite words written in base @xmath1 over the alphabet @xmath105 , that is the language in base @xmath1 , by the so - called conditions of parry @xcite @xcite @xcite .
a parry number @xmath1 is a real number @xmath11 for which the sequence of digits @xmath104 in the rnyi @xmath1-expansion of unity @xmath106 either ends in infinitely many zeros , in which case @xmath14 is said to be finite and @xmath1 is said a simple parry number , or is eventually periodic . in the second case ,
if the preperiod length is zero , @xmath14 is said to be purely periodic .
the set of parry numbers is denoted by @xmath107 .
let @xmath108 be the set of algebraic numbers .
denote by @xmath109 , resp .
@xmath110 , resp .
@xmath111 , the set of salem numbers , resp .
pisot numbers , resp .
perron numbers .
after bertrand - mathis @xcite , schmidt @xcite , lind @xcite , the following inclusions hold @xmath112 the question of the dichotomy @xmath113 is an important open question , which amounts to finding a method for discrimating when a perron number @xmath11 is a parry number or not . in particular , for salem numbers , though conjectured to be nonempty with a positive density @xcite , the set @xmath114 is not charaterized yet . for now , it is a fact that all the small salem numbers , for instance those given by lehmer in @xcite , and many others known , are parry numbers @xcite @xcite .
the set of simple parry numbers contains @xmath115 and is dense in @xmath116 @xcite .
let @xmath1 be a parry number , with @xmath20 . if @xmath117 , the integer @xmath26 is the preperiod length of @xmath14 ; if @xmath118 , the period length of @xmath14 is @xmath27 .
the iterates of @xmath13 under @xmath7 are polynomials : @xmath119 ( by induction ) .
this observation allows boyd in @xcite to define uniquely the parry polynomial of @xmath1 . indeed ,
writting @xmath120 , we have @xmath121 and @xmath1 satisfies the polynomial equation @xmath122 , where @xmath123 the parry polynomial @xmath24 of the parry number @xmath1 , monic , of degree @xmath124 , multiple of the minimal polynomial @xmath58 of @xmath1 , can also be defined from the rational fraction @xmath125 : its reciprocal @xmath126 , of the complex variable @xmath18 , is the denominator of the meromorphic function @xmath4 , given in both cases by and ( simple " case ) .
boyd @xcite defines the beta - conjugates of @xmath1 as being the roots of @xmath24 , canonically attached to @xmath1 , which are not the galois conjugates of @xmath1 .
beta - conjugates are algebraic integers .
for any real number @xmath0 , from the sequence @xmath127 we form the parry upper function @xmath17 at @xmath1 , of the complex variable @xmath18 .
the terminology parry upper " comes from the fact that @xmath104 gives the upper bound for admissible words in base @xmath1 , where being lexicographically smaller than this upper bound , with all its shifts , means satisfying the conditions of parry for admissibility @xcite @xcite @xcite .
when @xmath1 is a parry number , the inverses @xmath128 of the zeros @xmath129 of the analytic function @xmath2 are exactly the roots of the parry polynomial @xmath24 of @xmath1 ( from , ; @xcite ) .
in particular we have @xmath130 by .
the multiplicity of the root @xmath5 in @xmath2 is one by the fact that @xmath131 .
hence in the factorization of @xmath24 the multiplicity of the minimal polynomial @xmath58 of @xmath1 is one .
but the determination of the multiplicity of a beta - conjugate of @xmath1 and of the factorization of the parry polynomial of @xmath1 is an open problem @xcite @xcite .
we give a partial solution to this problem by showing how this factorization can be deduced from the germ of curve at @xmath5 " and the theory of puiseux .
though the degree @xmath132 of the parry polynomial @xmath24 of a parry number @xmath1 be somehow an obscure function of @xmath1 , the parry polynomial @xmath24 , say @xmath133 , has the big advantage , as compared to the minimal polynomial @xmath58 of @xmath1 , to exhibit a naive height h@xmath134 in @xmath135 @xcite .
this control of the height by the base of numeration @xmath1 has an important consequence : given a convergent family of parry numbers @xmath136 , an equidistribution limit theorem for the conjugates @xmath137 holds with a limit measure which is the haar measure on the unit circle @xcite , under some assumptions .
fractal @xmath34 is densely occupied by all the conjugates of all the parry numbers @xcite , with a major concentration of conjugates occuring in a neighbourhood of the unit circle .
beta - conjugates are then equivalently defined either as roots of @xmath24 , as inverses of zeros of @xmath2 , as inverses of poles of the dynamical zeta function @xmath4 .
the three equivalent definitions arise from the relations and ( simple " case ) , deduced from @xcite @xcite .
the galois- and beta- conjugates @xmath15 of a parry number @xmath1 all lie in solomyak s fractal @xcite , represented in figure [ fractalsolo ] .
the left extremity of the spike on the real negative axis is @xmath138 and the general bound @xmath139 holds for all @xmath10 and all parry numbers @xmath1 ; this upper bound was also found by flatto , lagarias and poonen @xcite .
.,width=264 ] let @xmath1 be a parry number .
the three following assertions are obviously equivalent : ( i ) @xmath1 has no beta - conjugate , ( ii ) the parry polynomial of @xmath1 is irreducible , ( iii ) the parry polynomial of @xmath1 is equal to the minimal polynomial of @xmath1 .
for some families of parry numbers @xcite @xcite it is possible to deduce the irreducibility of their parry polynomials . by szeg - carlson - poly theorem @xcite ,
the parry upper function @xmath2 is a rational fraction if and only if @xmath1 is a parry number @xcite .
if @xmath0 is an algebraic number , but not a parry number , @xmath2 is an analytic function on the open unit disc with the unit circle as natural boundary . for @xmath0 any algebraic number , except a parry number , we define a beta - conjugate of @xmath1 as the inverse of a zero of the function @xmath2 , if it exists . a priori
, it may happen that @xmath2 admits the only zero @xmath5 in its domain of definition @xmath40 , with @xmath37 as natural boundary .
the problem of the existence of zeros of @xmath2 in @xmath40 is linked to the gappiness ( the terminology _ gappiness _ was introduced in @xcite as a notion which is much weaker than that of _ lacunarity _ ; indeed _ lacunarity _ is classically associated to hadamard gaps ) of the sequence @xmath140 and its diophantine approximation properties @xcite @xcite ; this gappiness can not be too large at infinity and the ostrowski
quotients of the gaps " are dominated by @xmath141 , where m@xmath85 is the mahler measure of @xmath1 . by a theorem of fuchs @xcite ,
if @xmath2 is such that @xmath140 admits hadamard gaps , then the number of zeros of @xmath2 is infinite in @xmath40 .
this occurence , of having hadamard gaps , is conjectured to be true for infinitely many transcendental numbers @xmath0 but to be impossible as soon as @xmath0 is an algebraic number . if @xmath0 is an algebraic number , the number of zeros of @xmath2 in @xmath40 , i.e. the number of beta - conjugates of @xmath1 of modulus @xmath11 , is conjectured to be finite .
this finiteness property of the number of beta - conjugates would be in agreement with the existence of an integer @xmath142 in , in the context of the dynamical zeta function .
in the sequel , we will follow casas - alvero @xcite , duval @xcite , walker @xcite , walsh @xcite and restrict ourselves to what is needed for the application of the theory of puiseux to beta - conjugates of algebraic numbers @xmath11 , to fix notations .
the terminology `` fractionary '' is taken from @xcite .
let @xmath49 be a ( commutative ) field of characteristic zero and let @xmath143 $ ] .
we consider the formal equation @xmath144 and are interested in solving it for @xmath145 , that is we want to find some sort of series in @xmath146 , say @xmath147 , with coefficients in @xmath49 , such that @xmath148 @xmath149 being the series in @xmath146 obtained by substituting @xmath147 for @xmath145 in @xmath150 .
the series @xmath147 is called a @xmath145-root of @xmath150 . when @xmath151 , this general problem was considered by newton . in the following
we will consider @xmath151 and will consider rationality questions over smaller fields @xmath49 in section [ s6 ] . for solving ,
we need to deal with series in fractionary powers of @xmath146 .
first , let us define the field of fractionary power series over @xmath62 .
denote @xmath152 the field of the formal laurent series @xmath153 an element of @xmath154 has the form @xmath155 the field of fractionary power series is denoted by @xmath156 and by definition is the direct limit of the system @xmath157 where , for @xmath158 dividing @xmath159 ( with @xmath160 ) , @xmath161 a puiseux series is by definition a fractionary power series @xmath162 for which the order in @xmath146 @xmath163 is ( strictly ) positive .
a natural representant of its class in the direct limit is such that @xmath158 and gcd@xmath164 have no common factor ; then @xmath158 is called the ramification index ( or polydromy order ) of @xmath165 , denoted by @xmath166 .
if @xmath167 is a puiseux series , with @xmath168 its ramification index , the series @xmath169 , will be called the conjugates of @xmath165 , where @xmath170 the set of all ( distinct ) conjugates of @xmath165 is called the conjugacy class of @xmath165 .
the number of different conjugates of @xmath165 is @xmath166 .
let us recall the newton polygon of a two - variable formal series .
let @xmath171\ ] ] and obtain the discrete set of points with nonnegative integral coefficients @xmath172 called the newton diagram of @xmath150 .
let @xmath173 be the first quadrant in the plane @xmath174 and consider @xmath175 then the convex hull of @xmath176 admits a border which is composed of two half - lines ( a vertical one , an horizontal one , parallel to the coordinate axes ) and a polygonal line , called the newton polygon of @xmath150 , joining them , denoted by @xmath177 . the height @xmath178 of @xmath150 is by definition the maximal ordinate of the vertices of the newton polygon @xmath177 .
if @xmath179 is a puiseux series , write @xmath180 , the @xmath181 being the conjugates of @xmath182 .
the series @xmath183 is irreducible in @xmath184 $ ] .
the theory of puiseux allows a formal decomposition as follows .
[ decompomainthm ] for any @xmath185 $ ] , * there are puiseux series @xmath186 , in @xmath187 so that @xmath150 decomposes in the form @xmath188 where @xmath189 , and @xmath190 is an invertible series in @xmath184 $ ] , * the height of the newton polygon of @xmath191 is the sum of the ramification indices @xmath192 and the @xmath145-roots of @xmath150 are the conjugates of the @xmath193 .
the newton - puiseux algorithm applied to the newton polygon @xmath177 of @xmath150 allows to compute all the @xmath145-roots of @xmath194 and the ramification indices @xcite @xcite @xcite .
let @xmath49 be a ( commutative ) field of characteristic zero and @xmath195 an element of @xmath196 $ ] such that @xmath197 .
a parametrization of @xmath191 is an ordered pair @xmath198 of elements of @xmath199 $ ] which satisfies * @xmath200 and @xmath201 are not simultaneously identically zero , * @xmath202 , * @xmath203 $ ]
. denote @xmath204 the ring of convergent power series , and turn to convergence questions .
let @xmath205 be a fractionary power series , with @xmath206 .
we say that @xmath165 is a convergent fractionary power series if and only if the ordinary power series @xmath207 has nonzero convergence radius .
this condition does not depend upon the integer @xmath158 and the set of convergent fractionary power series @xmath208 is a subring of @xmath209 . if @xmath165 is convergent , with @xmath210 , one may compose the polydromic ( multivalued ) function @xmath211 and the analytic function defined by @xmath212 in a neighbourhood of @xmath213 : we obtain a polydromic function @xmath214 , defined in a neighbourhood of @xmath215 , which we call the ( polydromic ) function associated with @xmath165 .
if @xmath165 is convergent , all its conjugates are also convergent and any of them defines the same polydromic function @xmath214 as @xmath165 .
if @xmath165 is convergent , the associated function @xmath214 takes @xmath166 different values on each @xmath216 in a suitable neighbourhood of @xmath217 . in the context of convergent
series the theory of puiseux makes theorem [ decompomainthm ] more accurate as follows .
[ decompconvergentthm ] if @xmath218 is a convergent series , then all its @xmath182-roots are convergent , and there are an invertible series @xmath219 and a nonnegative integer @xmath220 , both uniquely determined by @xmath150 , and convergent puiseux series @xmath221 , uniquely determined by @xmath150 up to conjugation so that @xmath222 if @xmath150 is a polynomial in @xmath145 , i.e. if @xmath223[y]$ ] , and if the coefficients @xmath224 of the puiseux expansions involved in its decompositon are algebraic numbers , denote @xmath225 the number field generated by the coefficients .
assume @xmath226 < + \infty$ ] and let @xmath227 $ ] .
let @xmath228 , the @xmath220 embeddings of @xmath229 into @xmath108 .
denote @xmath230 the @xmath229-rational conjugacy class of @xmath231 . by proposition 2.1 in walsh @xcite , assuming that all the puiseux expansions of @xmath146 in @xmath150 are distinct , @xmath232 is irreducible in @xmath233 $ ] , of degree @xmath234 in @xmath145 , and @xmath235 is irreducible in @xmath236 $ ] of degree @xmath237 in @xmath145 where @xmath238 if , in addition , @xmath150 is assumed convergent , gathering the puiseux expansions by @xmath229-rational conjugacy classes , whose number is ( say ) @xmath239 , the collection of such classes being @xmath240 , allows to write @xmath150 in the form of the product of a unit @xmath241 $ ] by a nonnegative power @xmath242 of the first variable @xmath93 and a product of @xmath239 irreducible polynomials in @xmath243[y]$ ] as follows : @xmath244
let @xmath0 be an algebraic number , not necessarily a parry number . in the sequel we will not consider the case where @xmath0 is a rational integer : indeed , in this case
, @xmath1 has no galois conjugate @xmath245 , and @xmath246 is a polynomial having only the root @xmath5 ; therefore @xmath1 has no beta - conjugate . the key observation , that the three functions @xmath56 , @xmath55 , @xmath2 cancel at @xmath5 , each of them with multiplicity one , leads to consider the point @xmath53 of @xmath247 as natural origin of the germ of curve .
therefore we consider the new variable @xmath248 and make the change of variable @xmath249 into @xmath2 and @xmath55 , as follows : @xmath250 [ fbetatilde ] let @xmath0 be a real number .
then @xmath251 with @xmath252 expanding @xmath253 as a function of @xmath254 readily gives .
let @xmath0 be any real number .
the series @xmath255 have nonegative terms and , by stirling s formula applied to the binomial coefficients , are convergent .
[ lambda_jcontinuous ] let @xmath0 be a real number . for all @xmath100 ,
the map @xmath256 is right - continuous .
the set of discontinuity points is contained in the set of simple parry numbers .
assume @xmath0 a real number which is not an integer .
let us fix @xmath100 .
there exists @xmath257 such that the open interval @xmath258 contains no integer .
then any @xmath259 is such that its rnyi @xmath260-expansion @xmath261 of @xmath13 has digits @xmath262 within the same alphabet which is @xmath263 .
let @xmath264 .
then there exists @xmath265 such that @xmath266 then , for all @xmath259 , since @xmath267 , the following uniform inequality holds : @xmath268 now there are are two cases : either @xmath1 is a simple parry number , or not .
\(i ) assume @xmath0 is not a simple parry number .
then the sequence @xmath269 is infinite ( does not end in infinitely many zeros )
. there exists @xmath270 small enough such that @xmath271 for all @xmath272 with @xmath273 , for which , since @xmath274 , are all continuous , @xmath275 in this nonsimple parry case , recall @xcite that the function @xmath276 is monotone increasing and locally constant when the variable @xmath260 tends to @xmath1 ( i.e. @xmath261 and @xmath14 start by the same string of digits @xmath277 when @xmath260 is close to @xmath1 ) .
\(ii ) assume that @xmath0 is a simple parry number .
let @xmath278 be its rnyi @xmath1-expansion of unity ( @xmath279 ) .
if @xmath280 , there exists @xmath281 , @xmath282 , such that @xmath283 for all @xmath284 , and also holds .
if @xmath285 , we express @xmath1 in base @xmath1 and @xmath260 in base @xmath260 in the sense of rnyi : then we deduce that there exists @xmath281 , @xmath282 , such that @xmath286 implies @xmath287 and @xmath288 in this case , @xmath289
if @xmath290 , we deduce , for all @xmath291 , @xmath292 @xmath293 and , in the case @xmath285 , we decompose the sum @xmath294 as @xmath295 in the upper bound , using and , to obtain @xmath296 as well . if @xmath297 , then @xmath298 ; there exists @xmath281 , @xmath282 , such that @xmath286 implies @xmath299 hence , using and , for @xmath300 , @xmath301 and the right - continuity @xmath302 for @xmath297 .
let us now assume that @xmath0 is an integer .
then @xmath303 , @xmath304 and @xmath305 for @xmath306 .
the same arguments as in ( ii ) , with @xmath307 , lead to the result .
[ pbetatilde ] if @xmath0 is an algebraic number of minimal polynomial @xmath308 , @xmath309 , then @xmath310 with @xmath311 .
the relation @xmath312 leads to @xmath313 the constant term is zero since @xmath314 .
[ existgermthm ] let @xmath0 be an algebraic number and @xmath58 its minimal polynomial .
then there exists a unique polynomial @xmath315[z]$ ] in @xmath316 , deg@xmath317deg @xmath1 , such that @xmath318 is a parametrization of @xmath319 $ ] , i.e. such that @xmath320 uniqueness .
assume that @xmath321 and @xmath322 are such that @xmath323 and @xmath324 are both parametrized by @xmath325 .
then @xmath326 with @xmath327[z]$ ] , deg@xmath328 .
assume @xmath329 and @xmath330 irreducible in @xmath316 ( no loss of generality ) .
then this equation defines a plane curve @xmath331 along with a ramified covering @xmath332 of the complex plane . above all but finitely many points @xmath190 of the @xmath333-plane , the fiber @xmath334 has cardinality @xmath335 .
the implicit function theorem states that there exist @xmath336 analytic functions @xmath337 , @xmath338 , such that @xmath339 and @xmath340 for @xmath341 .
each of them parametrizes one sheet of the covering in a neighbourhood of @xmath190 .
the contradiction comes from the fact that the polynomial @xmath55 is irreducible , of degree @xmath342 , that the parametrization @xmath343 is imposed .
therefore the number of sheets @xmath336 should be equal to @xmath342 . contradiction .
existence : by construction .
let @xmath344 . from
, @xmath345 it follows that @xmath346[z]$ ] , with deg@xmath347 .
the idea consists in replacing all powers @xmath348 , in @xmath349 by polynomials in @xmath316 , of degree @xmath350 , with coefficients in @xmath351 $ ] .
let us prove recursively that @xmath352[z]$ ] , with deg@xmath353 , for all @xmath354 : assume @xmath355 with @xmath356 $ ] and show @xmath357[z]$ ] , with deg@xmath358 .
indeed , @xmath359 @xmath360.\ ] ] hence @xmath361 and the result .
we deduce @xmath362 @xmath363[z].\ ] ] equation is exactly with the usual variable @xmath18 .
we call @xmath67 the germ associated with the analytic function @xmath2 , or with the base of numeration @xmath1 . following theorem [ decompconvergentthm ] and the relations and , the decomposition of the germ @xmath67 shows that the coefficients of its puiseux series do possess a
right - continuity " property , with @xmath1 , via the functions @xmath364 ( proposition [ lambda_jcontinuous ] ) , and an asymptotic " property , linked to the invariants of the companion matrix form of .
this will be developped further elsewhere .
the interest of such a remark may consist in studying globally the properties of the family of germs @xmath365 when @xmath0 varies in the set of algebraic numbers .
[ puiseuxfbeta ] let @xmath0 be an algebraic number , @xmath58 its minimal polynomial and @xmath67 the germ associated with the parry upper function @xmath2 .
then @xmath366 where @xmath367 is an invertible series , and the convergent puiseux series @xmath368 are uniquely determined by @xmath67 , up to conjugation , with @xmath369 @xmath370 and @xmath371 theorem [ decompconvergentthm ] is applied to the germ @xmath66 . since @xmath2 is convergent in a neighbourhood of @xmath5 , @xmath67 and all the puiseux expansions involved in its decomposition are convergent in this neighbourhood .
the power of @xmath333 in is necessarily equal to @xmath13 since @xmath372 , i.e. @xmath217 is a simple zero of @xmath373 . since deg@xmath374 deg @xmath1 , by the definition of the height of the newton polygon of the germ @xmath67
, we readily deduce from theorem [ decompomainthm ] ( ii ) . for @xmath0 any algebraic number , a beta - conjugate @xmath129 of @xmath1 is by definition a complex number such that ( i ) @xmath128 is a zero of @xmath2 which lies in its domain of definition , ( ii ) @xmath129 is not a galois conjugate of @xmath1
. for parry numbers @xmath1 , and show that this definition is exactly the usual one which uses the parry polynomial of @xmath1 @xcite .
equation gives the exhaustive list of zeros of @xmath2 , and therefore suggests the following alternate definition of the beta - conjugates of @xmath1 ( where the natural boundary @xmath37 of @xmath2 is taken into account , if @xmath1 is not a parry number ) .
[ betaconjugatepuiseuxfbetanewdef ] let @xmath0 be an algebraic number .
\(i ) a complex number @xmath129 which satisfied @xmath375 where @xmath376 , @xmath377 , is any @xmath316-root , is called a _ cancellation point _ of the germ @xmath66 .
we say that the cancellation point @xmath129 lies on the @xmath316-root @xmath378 .
the set of cancellation points is denoted by @xmath379 .
equation has to be understood as the composition of the ( convergent ) two analytic functions @xmath380 and @xmath381 with the multivalued ( polydromic ) analytic function @xmath382 .
since @xmath129 is not a galois conjugate of @xmath1 the function @xmath381 does not cancel on a small neighbourhood of @xmath128 ; this give a sense to .
since the puiseux expansions in are convergent , truncating them to a few terms transforms into a finite collection of equations whose solutions provide the geometry of the beta - conjugates of @xmath1 with a certain approximation , controlled by the error terms .
this approach will be continued elsewhere .
\(ii ) if @xmath1 is a parry number , a beta - conjugate of @xmath1 is a cancellation point of the germ .
the set @xmath379 is the set of beta - conjugates of @xmath1 , and @xmath383 solomyak s fractal .
\(iii ) if @xmath1 is not a parry number , a beta - conjugate of @xmath1 is a cancellation point @xmath384 of the germ such that @xmath385 .
\(iv ) a cancellation point @xmath384 , lying on the @xmath316-root @xmath378 , is said puiseux - conjugated to another cancellation point @xmath386 if @xmath387 , lying on a @xmath316-root @xmath388 , is such that @xmath378 and @xmath388 belong to the same conjugacy class of the germ @xmath66 .
if @xmath0 is an algebraic number which is not a parry number the natural boundary @xmath37 of @xmath2 is the natural boundary of at least one of the factors in , but not necessarily of all of them a priori . in other terms it may occur that puiseux - conjugation may be addressed to cancellation points of the germ @xmath67 which lie beyond the natural boundary of @xmath2 , some branches possibly spiraling outside the domain of definition of @xmath2 .
let @xmath1 be a parry number , with @xmath26 as preperiod lenght and @xmath27 as period length in @xmath14 .
then the parry polynomial of @xmath1 is , for non - simple parry numbers , @xmath390 @xmath391 and @xmath392 in the case of pure periodicity . for simple parry numbers , the parry polynomial is @xmath393 with @xmath394 @xcite @xcite @xcite . the height ( @xmath395 maximum of the moduli of the coefficients ) of the parry polynomial lies in @xmath396 ; if @xmath1 is a simple parry number , then it is equal to @xmath397 @xcite . in the decomposition of @xmath24 as the product of irreducible polynomials with coefficients in @xmath389 ,
as @xmath398 we may identify the irreducible factors @xmath399 as arising from the conjugacy classes of the germ @xmath67 .
this requires some assumptions .
[ factoparrypoly ] let @xmath0 be a parry number , @xmath58 its minimal polynomial , @xmath24 its parry polynomial decomposed as @xmath400 into irreducible factors .
let @xmath67 the germ associated with @xmath1 and @xmath229 be the field of coefficients of the puiseux series of @xmath67 .
assume that all the puiseux expansions of @xmath146 in @xmath67 are distinct .
assume @xmath226 < + \infty$ ] and , for each @xmath229-rational conjugacy class @xmath401 , the product @xmath402[y].\ ] ] if @xmath239 is the number of @xmath229-rational conjugacy classes @xmath240 , then \(i ) @xmath403 , and \(ii ) up to the order , @xmath404 this is a consequence of proposition 2.1 in walsh @xcite . under the present assumptions
@xmath405 if @xmath406 and the decomposition of @xmath67 , as given by , allows to write @xmath2 as a product of distinct irreducible factors in @xmath407[y]$ ] . from and the identification of the factors readily gives @xmath408 , the irreducible factors @xmath409 and the unit @xmath410 , with @xmath411 if @xmath1 is simple , with @xmath14 of length @xmath26 , and @xmath412 if @xmath1 is not simple , with @xmath14 of period length @xmath27 . from , the number @xmath413 of irreducible factors which arises from @xmath229-rational conjugacy classes of puiseux expansions is smaller than deg @xmath1 .
using and and assuming the hypotheses of theorem [ factoparrypoly ] we obtain the following reformulation of the dynamical zeta function @xmath4 as a finite product over the @xmath239 @xmath229-rational conjugacy classes , @xmath415 deg @xmath1 , @xmath416 the unit @xmath417 is equal to @xmath418 with @xmath411 if @xmath1 is simple , with @xmath14 of length @xmath26 , and @xmath412 if @xmath1 is not simple , with @xmath14 of period length @xmath27 .
the poles of @xmath4 are either the reciprocals @xmath128 of the cancellation points @xmath129 of the germ @xmath67 of @xmath1 , or the reciprocals of the galois conjugates of @xmath1 . the assumptions in theorem [ factoparrypoly ] could probably be weakened , for obtaining the same decomposition .
the author is indebted to m. pollicott and to m. lejeune - jalabert for valuable comments and discussions . | the beta - conjugates of a base of numeration @xmath0 , @xmath1 being a parry number , were introduced by boyd , in the context of the rnyi - parry dynamics of numeration system and the beta - transformation .
these beta - conjugates are canonically associated with @xmath1 .
let @xmath0 be a real algebraic number .
a more general definition of the beta - conjugates of @xmath1 is introduced in terms of the parry upper function @xmath2 of the beta - transformation .
we introduce the concept of a germ of curve at @xmath3 associated with @xmath2 and the reciprocal of the minimal polynomial of @xmath1 .
this germ is decomposed into irreducible elements according to the theory of puiseux , gathered into conjugacy classes .
the beta - conjugates of @xmath1 , in terms of the puiseux expansions , are given a new equivalent definition in this new context .
if @xmath1 is a parry number the ( artin - mazur ) dynamical zeta function @xmath4 of the beta - transformation , simply related to @xmath2 , is expressed as a product formula , under some assumptions , a sort of analog to the euler product of the riemann zeta function , and the factorization of the parry polynomial of @xmath1 is deduced from the germ . 11m99 , 30b10 , 12y05 .
rnyi , parry number , numeration system , parry polynomial , algebraic integer , dynamics , beta - conjugate , germ of curve , puiseux series , dynamical zeta function , factorization . |
nowadays , co@xmath0 based heusler compounds are successfully used as ferromagnetic electrodes providing a spin polarized electrical current in tunneling magnetoresistive ( tmr ) junctions @xcite . to produce high performance spintronic devices with co@xmath0 based heusler thin films as ferromagnetic electrodes ,
it is necessary to clarify the dependence of the magnetoresistive characteristics on the stoichiometry of the heusler film electrode and its impact on the selection of materials for tmr devices .
however , very little is known about the electronic structure of non - stoichiometric heusler compounds .
furthermore , partially substitution of one of the elements in the ternary heusler @xmath5 ( where , @xmath6 is transition metals and @xmath7 main group element ) can be used to design new materials with predictable properties @xcite , such as tuning the fermi energy in to the middle of the minority band gap @xcite .
the partially substitution of the elements in @xmath5 leads to substitutional quaternary alloys of the type @xmath8 or @xmath9 .
the quaternary alloy co@xmath0cr@xmath10fe@xmath11al has attracted emerged interest as potential material for magnetoelectronics , due to the large mr @xcite .
the band structure calculations confirmed the halfmetallic state of this compound in the ordered @xmath12 structure @xcite .
recently , several groups developed fully epitaxial magnetic tunnel junctions ( mtjs ) based on co@xmath0cr@xmath10fe@xmath11al as a lower electrode and a mgo tunnel barrier @xcite . however , a disorder of the structure results in a strongly reduced magnetic moment as well as a loss of the half - metallic character @xcite . co@xmath0mnz ( z = si , ge , sn ) are representatives of heusler compounds with a specific electronic band structure providing a pronounced dependence of its transport properties on the orientation of the electron spin @xcite . structural and magnetic properties of co@xmath0mnsi were reported for films and single crystals @xcite and the compound was suggested to be suitable for magnetic tunnel junctions ( mtjs ) @xcite
. a high tmr ratio of 179% at room temperature ( rt ) and 683% at 4.2 k was revealed by ishikawa and coworkers @xcite .
the substitution of fe by mn in co@xmath0fe@xmath1mn@xmath13si shows a dependency of half - metallicity and gilbert damping constant on the composition @xmath14 @xcite .
recently , anisotropic magnetoresistance ( amr ) effect was systematically investigated in epitaxially grown co@xmath0fe@xmath1mn@xmath13si heusler films with respect to fe composition @xmath14 and the annealing temperature @xcite .
it was shown that the amr effect can be an indicator of half - metallicity .
an appropriate alternative to co@xmath0mnsi is the isovalent co@xmath0mnge with 29 valence electrons in the primitive cell .
recently , co@xmath0mnge was investigated in greater detail by theoretical and experimental methods in order to prove the half metallic ferromagnetic character , as well as their crystallographic structure , mechanical and transport properties @xcite .
this compound is of interest for spintronic applications because it combines high curie temperature ( 905 k ) @xcite , high magnetic moment ( 5 @xmath15 ) @xcite , and coherent growth on top of semiconductors @xcite .
the non - stoichiometric co@xmath0mnge was successfully used to produce magnetic tunnel junctions @xcite .
hakamata and coworkers fabricated epitaxial mtjs with a co - rich co@xmath0mnge film and reached relatively low tmr ratios of 83% at rt and 185% at 4.2 k @xcite .
thereafter , taira and coworkers investigated the influence of the annealing temperature on the performance of the mtjs .
they achieved a significant increase of the tmr up to 160% at rt ( 376% at 4.2 k ) by annealing at 500@xmath16c @xcite .
this increase was explained in terms of the increase in the interfacial spin polarization at the fermi level associated with the change in the spin - dependent interfacial density of states . on the other hand ,
the effect of defects in co@xmath0mn@xmath17ge@xmath18 and co@xmath0mn@xmath19si films on the tmr ratio was extensively studied in reference @xcite where a strong dependence on the composition and mn content was observed . a high tmr ratio of 650% at 4.2 k and 220% at rt for co@xmath0mn@xmath17ge@xmath18 with @xmath20= 1.40
was found .
the highest tmr ratios for heusler based mtjs of 1135% at 4.2 k and 236% at rt was achieved for mn - rich co@xmath0mn@xmath19si electrodes with @xmath21 = 1.29 .
it was suggested that detrimental co@xmath22 antisites can be suppressed by preparing the films with a mn - rich composition .
the present work reports on the theoretical background and experimental studies of the electronic structure of non - stoichiometric co@xmath0mn@xmath17ge@xmath18 . _ ab - initio
_ electronic structure calculations are used to verify the range of existence of the half - metallic ferromagnetic behavior .
high resolution photoelectron spectroscopy of the core levels and valence states of buried co@xmath0mn@xmath17ge@xmath18 films was used to explore the electronic structure in comparison to the theoretical predictions .
hard x - ray photoelectron spectroscopy ( haxpes ) was previously shown to be a bulk sensitive probe of the electronic structure @xcite .
the use of high - brilliance high - flux x - rays from the third - generation synchrotron radiation sources results the emission of electrons having high kinetic energies , in turn leading to a high probing depth because of the increased electron mean free path @xcite .
recently , several studies using high - resolution haxpes have been realized .
the electronic structure of solids like valence transitions in bulk systems @xcite as well as multilayer systems @xcite and the valence band of buried thin films @xcite have been investigated .
newly , the use of polarized light in combination with haxpes enables the analysis of the symmetry of bulk electronic states @xcite as well as the magnetic properties of buried layers @xcite .
haxpes was used in the present work to study buried co@xmath0mn@xmath17ge@xmath18 films underneath mgo / alo@xmath1 capping layers .
the study of core level photoelectron spectra of the buried heusler thin films gives information on the electronic structure and the chemical composition of these films .
for the present study , special multi - component thin film arrangements were produced that corresponded to a half of a magnetic tunnel junction as used in tmr devices ( see figure [ fig : sample ] ) .
in particular the free electrode was modeled .
the fabricated sample layer structure was as follows : mgo(001 ) substrate ( 0.5 mm ) / mgo buffer layer ( 10 nm ) / co@xmath0mn@xmath17ge@xmath18 ( 30 nm or 50 nm ) / mgo barrier ( @xmath23 ) / alo@xmath1 ( 1 nm ) cap .
nm and 20 nm were chosen for the thickness of the mgo layer .
the topmost alo@xmath1 was used for protection of the hygroscopic mgo layer during contact with air .
the sample structure is sketched in figure [ fig : sample ] . each layer in the sample layer structure was successively deposited in an ultrahigh vacuum chamber ( base pressure : about @xmath25 pa ) through the combined use of magnetron sputtering for co@xmath0mnge and al , and electron beam evaporation for mgo .
the fabrication procedure was the same as in previous works @xcite .
the thin films were deposited at 300 k and subsequently annealed in situ at @xmath26c for 15 min .
co - sputtering from an additional mn target was applied to vary the mn content of the co@xmath0mn@xmath17ge@xmath18 thin films .
x - ray diffraction revealed that the co@xmath0mn@xmath17ge@xmath18 films have a dominating @xmath12 crystalline structure .
the resulting co@xmath0mnge layers used in the photoemission study were estimated to be non - stoichiometric with the compositions of @xmath27 , @xmath28 , and @xmath29 ( see reference @xcite ) .
the mgo barrier was deposited at 300 k. the alo@xmath1 cap was prepared by exposing the sputter - deposited al layer to an o@xmath0 atmosphere of @xmath30 pa for 2 hours .
the precise occupation probability for each particular site of an assumed @xmath12 structure is unknown although the relative amounts of constituents are well - defined .
for this reason we will use the composition formula @xcite derived from theoretical estimations @xcite of the defect formation energies , using the permanent ratio of co : mn : ge as @xmath31 .
firstly , it is assumed that the co - mn - ge alloy series basically does not contain vacancies .
the ge deficiency is compensated by mn ( mn@xmath32 antisites ) .
the lack of mn in its native octahedral position is backfilled by co@xmath22 antisites .
further , the lack of co in the tetrahedral sites is compensated by mn@xmath33 antisites .
this leads to the following compositions for @xmath34 : co@xmath0(mn@xmath13co@xmath1)(ge@xmath35mn@xmath2 ) for mn deficit ( @xmath36 ) and ( co@xmath37mn@xmath38)@xmath0mn(ge@xmath35mn@xmath2 ) for mn excess ( @xmath39 ) . for @xmath34 ,
the @xmath14 and @xmath40 antisite concentrations as function of the mn content parameter @xmath20 are determined by the relations : * @xmath41 ; * @xmath42 , this corresponds to the experiments reported in reference @xcite . [
cols="<,^,^,^,^,^,^,^",options="header " , ] [ tab : sym ] in the regular @xmath12 structure of the x@xmath0yz heusler compounds , the atoms occupy the wyckoff positions 8a ( x@xmath0 ) ( @xmath43 ) , 4a ( y ) ( @xmath44 ) , and 4b ( z ) ( @xmath44 ) of the primitive cell with @xmath45 symmetry ( space group 225 ) .
table [ tab : sym ] summarizes the site occupations of the primitive cell for the different @xmath20 values .
@xmath46 corresponds to a @xmath47 structure according to co@xmath48(mn@xmath49ge@xmath50 ) .
it should be noted that for compositions with @xmath51 ( @xmath52 ) a @xmath53-type structure with @xmath54 symmetry ( space group 216 ) can be expected where the 8c position of the @xmath12 structure is split into 4c and 4d such that all sites ( 4a-4d ) have @xmath43 symmetry . in that case
mn may occupy 4a and 4c whereas co and ge are distributed on 4b and 4d .
further , it is known that the binary alloys co@xmath55mn@xmath55 @xcite or co@xmath56ge@xmath57 @xcite crystallize in @xmath58 lattices with random site occupation and similar lattice parameters .
therefore , other types of disorder or site occupations can not be excluded .
due to the similar scattering amplitudes of the constituents , however , x - ray diffraction is not unambiguous in the structure determination .
to understand the electronic structure of the non stoichiometric compounds , first - principles calculations were performed using the kkr ( korringa - kohn - rostoker ) green s function method as implemented in the munich spr - kkr ( spin - polarized relativistic ) package @xcite .
the chemical disorder is treated by the coherent potential approximation ( cpa ) @xcite .
exchange and correlation are treated within the local spin density approximation ( lsda ) using the parametrization of vosko , wilk and nussair @xcite .
the full symmetry potential method was used in order to account for the non spherical character of the atomic potentials in solids that is rather significant in heusler compounds .
the @xmath59-integration mesh was set to a size of @xmath60 during the selfconsistent cycles resulting in 328 points in the irreducible wedge of the brillouin zone .
the expansion in spherical harmonics up to @xmath61 ( @xmath62-electrons ) was found to be numerically sufficient .
the calculations were performed for a disordered @xmath12 structure to account for the composition dependence .
the site occupations for the different compositions are explained in detail in section [ sec : exp ] . )
series , ( b ) mn - excessive ( @xmath63 ) series .
minority states ( @xmath64 ) are shown on a negative scale.,width=264 ] ) ( b ) mn - excess ( @xmath63).,width=264 ] the results of the electronic structure calculations are discussed in the following .
it is obvious from the comparison of the density of state ( figures [ fig : dos_spin - resolved](a ) and ( b ) ) that the most significant factor influencing the electronic properties is the formation of co@xmath22-antisites in the mn - poor regime ( @xmath65 ) . in full analogy to co - mn - si alloy systems
@xcite , co@xmath22-antisites introduce impurity - like states ( shown separately in figure [ fig : dos_co - impurity ] ) inside of the half - metallic band gap that seriously destroy the spin - polarization ( see figure [ fig : spinpol ] : @xmath65 ) . in general ,
the spin polarisation at the fermi energy is defined by : @xmath66 where @xmath67 and @xmath68 are the fermi velocities for majority and minority electrons , respectively . in photoemission one
has @xmath69 such that the spin polarisation is purely given by the majority ( @xmath70 ) and minority ( @xmath71 ) densities .
the situation is , however , different for the tmr wherefore the values can not be compared directly but only their trend .
it is worthwhile to note that in halfmetallic ferromagnets one has @xmath72 independent of the method . at the fermi energy calculated as a function of mn composition @xmath20.compared
are the values from ab - initio calculations to experimental values calculated from the @xmath73 measured at 4.2 k ( see ref . :
@xcite ) using @xmath74 .
the vertical line assigns @xmath75.,width=302 ] the total magnetic moment in the mn - deficient alloy strongly deviates from the slater - pauling rule ( figure [ fig : moments](a ) : @xmath65 ) .
this reasonably agrees with the measured tmr ratio and magnetic moments @xcite .
( experimental data from @xcite ) ; ( b ) calculated partial magnetic moments .
( see table [ tab : sym ] for the notation of the wyckoff positions 4a , 4b , 8c and their site occupations ; vertical lines are at @xmath75.),width=340 ] in contrast , the excess of mn ( which result in a formation of mn@xmath33 antisites ) does not influence the half - metallic gap noticeable ( figure [ fig : dos_spin - resolved](b ) ) .
the system remains nearly half - metallic .
this restores the slater - pauling behavior of the total magnetic moment ( figure [ fig : moments](a ) : @xmath52 ) .
the slightly reduced value of the spin - polarization for @xmath76 is mainly caused by mn@xmath32 antisites permanently present in the whole series .
this is also reflected in a slightly reduced value of the total magnetic moment compared to the slater - pauling rule .
as it follows from the calculated atomic - resolved magnetic moments shown in figure [ fig : moments](b ) , the total magnetization is mainly controlled by the amount of co@xmath22 and mn@xmath33 antisites . for the mn - deficient system the strong moment at the mn atom ( about @xmath77 ) in the octahedral position
is accompanied by the lower moment of the co@xmath22 antisite ( about @xmath78 ) .
this is the reason for the violation of the slater - pauling rule . on the other hand
, all possible mn permutations always result into half - metallic contributions . for the mn - excessive alloys , formation of the mn@xmath33 antisites at tetrahedral positions exhibit a negative moment of about @xmath79 .
the ferrimagnetic coupling between the regular mn atom and the mn@xmath80 antisite appears because they are nearest neighbors .
obviously , the magnetic interaction between neighboring co atoms is smaller when part of them occupies co@xmath22 antisites due to the smaller extend of the @xmath81-wave functions . from the comparison of the total ( spin - integrated ) density of states ( see figure [ fig : dos_spin - summed ] ) and the photoelectron spectra ( figure [ fig : ef8kev ] ) , it was found that the mn @xmath82 states decrease with decreasing mn content .
for the mn - poor alloys the minimum of the density of state at the fermi energy is destroyed by the appearance of co@xmath22 antisites . with increasing the mn content
the model system behaves as a half - metal following the slater - pauling scenario .
the rapid decrease of the measured magnetization @xcite above @xmath83 ( see figure [ fig : moments](a ) ) may be attributed to difference types of disorder including possible structural distortions in thin films .
( co antisites ) for mn - deficient regime ( @xmath84 ) .
minority states ( @xmath64 ) are shown on a negative scale.,width=226 ]
the haxpes experiments were performed at the undulator beamlines bl15xu and bl47xu of spring-8 . at bl15xu ,
the photon energy was fixed at 5.9469 kev using a si(111 ) double crystal monochromator ( dcm ) and a the 333 reflection of a si channel - cut post monochromator . at bl47xu ,
the photon energy was fixed at 7.9382 kev using a si(111 ) dcm and a si(444 ) channel - cut post monochromator .
the photo - emitted electrons were analyzed for their kinetic energy and detected by a hemispherical analyzer ( vg scienta r4000 ) .
the overall energy resolution ( monochromator plus analyzer ) was set to 250 mev , as verified by spectra of the au valence band at the fermi energy ( @xmath85 ) .
additionally , spectra close to the fermi energy were taken with a total energy resolution of 150 mev .
the angle between the electron spectrometer and photon propagation is fixed at @xmath86 in all experiments . a near normal emission ( @xmath87 ) detection angle was used resulting in an angle of photon incidence of @xmath88 .
the measurements were taken at sample temperatures of 20 k and 300 k. hard x - ray core level spectra were measured on co@xmath0mn@xmath17ge@xmath18 films with different compositions to investigate the influence of the off - stoichiometry on the spin orbit splitting as well as the exchange interaction of the unpaired valence electrons with the core holes .
figures [ fig : cl](a ) and ( b ) compare the hard x - ray co @xmath89 and mn @xmath89 core level spectra of the non - stoichiometric co@xmath0mn@xmath17ge@xmath18 films covered with 2 nm mgo .
the spectra were taken at room temperature with an excitation energy of about 8 kev .
concentrating on the co @xmath89 region , figure [ fig : cl](a ) illustrates a spin - orbit splitting between the co @xmath90 and @xmath91 states of about 14.86 ev .
there is no remarkable change of the splitting or line width detected for the different non - stoichiometric films .
all spectra exhibit a satellite at about 4 ev below the center of the @xmath90 main line .
such satellites were also observed in the previous work @xcite .
the satellite emerges from the interaction of the @xmath89 core hole with the partially filled @xmath81 valence bands . to clarify the detailed origin of the satellite structure
, theoretical calculations are required that treat the details of the exchange interaction between the co @xmath92 electron occupation ( in other words , the band structure ) and @xmath89 core hole states . at the @xmath91 state
it is only seen weakly as a tail at the low energy side due to shorter lifetime of the core hole compared with the @xmath90 state .
figure [ fig : cl](b ) shows the mn @xmath89 photoelectron spectra for the buried heusler thin films co@xmath0mn@xmath93ge@xmath94 and co@xmath0mn@xmath95ge@xmath94 .
the peak intensity of the spin - orbit split @xmath89 states of the mn - rich film co@xmath0mn@xmath93ge@xmath94 were slightly higher than those with less mn .
this is in agreement with the estimated composition of the films . in analogy to the co @xmath89 spectra , the mn @xmath89 core level shows the spin - orbit splitting of about 12.1 ev .
different to the co @xmath89 spectra , the multiplet splitting of the mn @xmath90 state was clearly revealed ( see inset in figure [ fig : cl](b ) ) .
it was shown theoretically and experimentally for the case of mn that the coulomb interaction of the @xmath89 core hole and the @xmath92 valence electrons leads to the splitting of the mn @xmath90 level into several main sublevels @xcite caused by existence of more than one possible final ionic state during ejection of an electron from the @xmath96 shell .
the effect of the exchange interaction between @xmath89 core hole and the partially filled @xmath81-bands is less pronounced at the @xmath91 state but still clearly detectable . the small influence of the composition on the mn @xmath89 derived multiplet is in agreement with the electronic structure calculations where the localized mn @xmath81 bands and the magnetic moment at mn are stable against variation of the composition .
this finding underlines the localized character of the mn valence @xmath81 electrons .
( a ) and mn @xmath89 ( b ) core level of co@xmath0mn@xmath17ge@xmath18 films .
the spectra are taken at room temperature .
the excitation energy was set to @xmath97 kev .
the inset in ( b ) shows the details of the mn @xmath90 peak on an enlarged scale .
, width=302 ] figure [ fig : vbmgo ] compares the hard x - ray photoelectron spectra of the shallow core levels with low binding energy of the 50 nm thick co@xmath0mn@xmath98ge@xmath99 layer buried underneath mgo / alo@xmath1 with different thicknesses of the mgo interlayer .
the intensity ratios of the core level @xmath100 of the films covered with 2 nm or 20 nm thick mgo are 0.95 or 16.45 , respectively .
this strong increase is caused by the increase of the emission from the thick mgo layer .
the inset of figure [ fig : vbmgo ] shows the valence band of the covered thin films . the high intensity at -1.3 ev below @xmath85 is clearly resolved and agrees well with the bulk spectra ( see also : figures [ fig : vb8kev ] and [ fig : ef8kev ] ) . the low lying @xmath101 band below -8 ev is only seen for the film with the 2 nm thick mgo layer
this part , in addition to the lower parts of the @xmath96 and @xmath81 bands , is covered by the emission from the mg @xmath102 and o @xmath89 states when the mgo film is 20 nm thick .
this has been already reported for other heusler compounds @xcite and makes clear that the valence band spectra close to the fermi energy are not influenced by the insulating layers above .
mn@xmath98ge@xmath99 thin films .
the spectra with different thickness of the mgo layer ( 2 nm and 20 nm ) are compared .
the excitation energy was set to 7.9392 kev .
the small peak at about -25 ev is due to the o @xmath103 excitation .
the inset compares the valence band spectra for different mgo thickness.,width=340 ] figure [ fig : vb8kev ] compares the valence band spectra of the non - stoichiometric co@xmath0mn@xmath17ge@xmath18 films to stoichiometric bulk co@xmath0mnge .
the main features of the valence band of the thin films agree well with that of the bulk sample .
the intense , broad band at about -10.5 ev is observed for the buried films as well as for the bulk sample .
this band is related to ge @xmath104 ( @xmath101 ) states and is separated from the remaining valence bands by the @xmath105 hybridization gap at around -8.5 ev .
this low lying band gap is characteristic for well ordered heusler compounds with @xmath12 structure .
this hybridization gap is well pronounced for the bulk sample , confirming the ordered structure .
additional features within the @xmath105 hybridization gap are seen in the spectra of the buried films .
partially , this intensity arises from the o @xmath89 states of the oxide overlayers and thus a direct comparison to the _ clean _ bulk material is not straightforward .
however , the difference in the spectral shapes between the films with altered composition one has the same overlayer and differences in the spectra can be considered as an evidence for the disorder rising up when changing the composition . the spectral shape above about -6 ev are not influenced by the mgo and alo@xmath1 layers as already mentioned above .
overall , the spectra of the co@xmath0mn@xmath17ge@xmath18 films show broadened structures compared to the spectrum of bulk co@xmath0mnge .
this broadening of the bands is explained by the disturbed translational periodicity due to the random site occupation or periodically unoccupied sites in the non - stoichiometric co@xmath0mn@xmath17ge@xmath18 films .
the majority mn @xmath106 states at about -3.9 ev , that are responsible for the localized magnetic moment at the mn atoms , are much less pronounced in the non - stoichiometric films compared to stoichiometric co@xmath0mnge .
all the states far below @xmath85 influence mainly the magnetic but not directly the transport properties .
mn@xmath17ge@xmath18 films and co@xmath0mnge bulk sample .
the excitation energy was set to 7.9392 kev.,width=226 ] most interesting with respect to the transport properties , in particular the use of the films in tmr devices , is the energy range close to the fermi energy .
the maximum in the density of the @xmath81 states at -1.1 ev is clearly resolved in the spectra from all films and the intensity varies when mn content is changed ( see figure [ fig : ef8kev ] ) .
this maximum in the intensity arises from co @xmath106 minority and mn @xmath82 majority states .
obviously , those states appear much stronger and sharper in the stoichiometric bulk sample pointing to their higher localization in the defect free sample .
the intensity directly at @xmath85 is dominated by @xmath106 majority states that are mainly located in the co@xmath0 planes of the compound .
the increase of the mn content enhances the intensity of the highest majority @xmath81-band that is mainly responsible for the electronic transport properties
. such composition dependent behavior of the density of states close to @xmath85 provides the possibility to tune the magnetoresistive characteristics of tunnel junctions by changing the composition of the co@xmath0mn@xmath17ge@xmath18 electrode , as was previously proposed in reference @xcite . .
the spectra are taken for co@xmath0mnge bulk as well as thin co@xmath0mn@xmath93ge@xmath94 , co@xmath0mn@xmath98ge@xmath99 , and co@xmath0mn@xmath95ge@xmath94 films .
the excitation energy was set to 7.9392 kev.,width=188 ] figure [ fig : vb_comp ] compares the valence band spectra on the buried co@xmath0mn@xmath98ge@xmath99 film taken with different photon energies of about 5.946 kev and 7.939 kev .
the differences in the spectra are caused by the different weights of the partial cross - sections for @xmath101 , @xmath96 , and @xmath81 electrons at different excitation energies @xcite . with increasing energies the cross - sections for @xmath81 electron excitation decrease
faster than those for @xmath101 or @xmath96 electrons .
therefore , the contribution from @xmath81 electrons is more pronounced at lower energy photo - excitation .
the observed shift of spectral weight towards the fermi energy in the non - stoichiometric films compared to bulk co@xmath0mnge is in agreement with the changes in the density of states predicted by the calculations .
mn@xmath98ge@xmath99 film excited by different photon energies ( 5.9469 kev and 7.9382 kev ) . , width=302 ] furthermore , the valence band spectra of the + co@xmath0mn@xmath98ge@xmath99 thin films taken at room temperature and at 20 k are compared in figure [ fig : vb_temp ] .
no general changes in the electronic structure are observed in the spectra while changing temperature . at @xmath107
k a small , additional peak appears that is caused by adsorbed carbon oxides on the surface of the samples while cooling .
this observation provides evidence for the requirement on ultra high vacuum for haxpes measurement even though this technique is bulk sensitive . indeed , a massive contamination of the surface in the order of several layers will influence the haxpes spectra .
figure [ fig : vb_temp](b ) shows a broadening of the spectra close to the fermi energy . the broad spectral feature in the vicinity of the fermi energy at 300 k compared to 20 k is caused by the temperature dependence of the fermi - dirac distribution .
this observation is in agreement with reports showing that co@xmath0mnsi films do not exhibit temperature effects in photoelectron spectroscopy @xcite .
mn@xmath98ge@xmath99 film .
( a ) compares the spectra measured at different temperatures ( @xmath108 kev ) .
( b ) shows the energy range close to the fermi energy on an enlarged scale .
, width=302 ]
in summary , the electronic structure of off - stoichiometric co@xmath0mn@xmath17ge@xmath18 alloys was investigated by first - principles calculations and bulk sensitive photoelectron spectroscopy on thin films .
the electronic structure calculations demonstrate that the magnetic moments of co and mn atoms ( including the antisites ) are nearly independent of the stoichiometry .
this finding is in agreement with the core level photoelectron spectra where no differences in the multiplet splitting of the mn @xmath89 states was observed for different compositions .
the formation of co@xmath22 antisites was found to be the most destructive depolarizing mechanism .
corresponding results for the mn - deficient alloy are in reasonable agreement with earlier measurements of tmr and magnetization .
this finding also agrees with the valence band spectra which exhibit a noticeable smearing of the mn @xmath106 and @xmath82 states for the mn - deficient alloys . on the other hand , the relative decrease of the measured magnetization , especially the significant violation of the slater - pauling rule in the mn - excessive regime ,
indicate the presence of additional disorder mechanisms different from the assumed types of antisites disorder .
the valence band spectra of the buried co@xmath0mn@xmath17ge@xmath18 thin films exhibit still some major differences compared to stoichiometric bulk co@xmath0mnge .
however , the electronic structure of the non - stoichiometric films was found to reveal the major characteristic features of the bulk analogue
. the observed shift of spectral weight towards the fermi energy in the non - stoichiometric films compared to bulk co@xmath0mnge is in agreement with the changes in the density of states predicted by the calculations .
temperature effects on the states at the fermi energy that go beyond the broadening of the fermi - dirac distributions are not observed .
this is , indeed , expected from the high curie temperature of the compound .
co - sputtering of co@xmath0mnge together with additional mn is shown to be a well suited technique to tune the stoichiometry of the co@xmath0mn@xmath17ge@xmath18 film composition that provides the possibility to change the electronic structure within the valence band and as a consequence the magnetoresistive characteristics of co@xmath0mn@xmath17ge@xmath18 based tunnel junctions .
the authors thank e. ikenaga ( bl47xu ) and s. ueda ( bl15xu ) for help with the experiments .
this work was funded by the _
deutsche forschungs gemeinschaft dfg _
( tp 1.2-a and 1.3-a of the research unit aspimatt ) and the _ japan science and technology agency jst _ ( dfg - jst project : fe633/6 - 1 ) .
the work at hokkaido university was partly supported by a grant - in - aid for scientific research ( a ) ( grant no .
20246054 ) from the mext , japan , and by the strategic international cooperative program of jst .
the synchrotron radiation haxpes measurements were performed at bl47xu with the approval of the japan synchrotron radiation research institute ( jasri ) ( long - term proposal 2008b0017 , 2009a0017 ) and at bl15xu with the approval of nims ( nanonet support proposal 2008b4903 ) .
the haxpes experiment at bl15xu was partially supported by the nanotechnology network project mext ( japan ) .
55ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) in @noop _
_ , vol . , ( , , ) pp . @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) | fully epitaxial magnetic tunnel junctions ( mtjs ) with off - stoichiometric co@xmath0-based heusler alloy shows a intense dependency of the tunnel magnetoresistance ( tmr ) on the mn composition , demonstrating giant tmr ratios of up to 1995% at 4.2 k for @xcite .
this work reports on the electronic structure of non - stoichiometric co@xmath1mn@xmath2ge@xmath3 thin films with a fixed co / ge ratio of @xmath4 .
the electronic structure was investigated by high energy , hard x - ray photoelectron spectroscopy combined with first - principles calculations .
the high - resolution measurements of the valence band of the non - stoichiometric co@xmath1mn@xmath2ge@xmath3 films close to the fermi energy indicate a shift of the spectral weight compared to bulk co@xmath0mnge .
this is in agreement with the changes in the density of states predicted by the calculations .
furthermore it is shown that the co - sputtering of co@xmath0mnge together with additional mn is an appropriate technique to adjust the stoichiometry of the co@xmath1mn@xmath2ge@xmath3 film composition .
the resulting changes of the electronic structure within the valence band will allow to tune the magnetoresistive characteristics of co@xmath1mn@xmath2ge@xmath3 based tunnel junctions as verified by the calculations and photoemission experiments . |
we present a numerical study of the semi - classical solutions to the following nonlinear schrdinger equations with @xmath0 , @xmath1 when a caustic ( a point or a cusp ) is formed , that is to say , beyond _ breakup time_. since the nonlinearity is homogeneous , the change of unknown function @xmath2 shows that is equivalent to : @xmath3 so that we can always consider initial data of order @xmath4 .
there are several motivations to study the behavior of when a caustic is formed .
first , on a purely academic level , we recall that the description of the caustic crossing is complete in the case of linear equations ; see @xcite . for nonlinear equations ,
very interesting formal computations were proposed in @xcite ( we recall the main idea in section [ sec : analyt ] below ) . for _ dissipative _ nonlinear wave equations , joly , mtivier and rauch @xcite have proved that the amplification of the wave near the caustic can ignite the dissipation phenomenon in such a way that the oscillations ( that carry highest energy ) are absorbed .
the above nonlinear schrdinger equation is the simplest model of a _ conservative , nonlinear equation_. the mass and the energy of the solution are independent of time ( see below ) .
therefore , different nonlinear mechanisms are expected .
we recall in section [ sec : analyt ] some results that have been established rigorously , and give heuristic arguments to extend these results .
this serves as a guideline for the numerical experiments proposed after .
second , may be considered as a simplified model for bose
einstein condensation , which may be modeled ( see e.g. @xcite ) by : @xmath5 with @xmath6 if @xmath7 , and @xmath8 if @xmath9 or @xmath10 .
the power @xmath11 in front of the nonlinearity depends on the rgime considered , and in particular on the respective scales of different parameters ( see e.g. @xcite and references therein ) .
the role of the harmonic potential @xmath12 is to model a magnetic trap . in the semi - classical limit @xmath13 for the linear equation , this potential causes focusing at the origin for solutions whose data are independent of @xmath14 .
this is to be compared with the case of with initial quadratic oscillations as considered below : the initial quadratic oscillations force the solution to concentrate at one point in the limit @xmath15 .
the parallel between and was extended and justified in @xcite for these nonlinear equations . from both points of view , when a caustic point is formed , the caustic crossing may be described in terms of the scattering operator associated to @xmath16 this aspect is recalled in section [ sec : analyt ] . for this reason
, we also pay a particular attention to this operator , independently of the above semi - classical limit .
note that besides the existence of this operator , very few of its properties ( dynamical , for instance ) are known . in this paper , we always assume @xmath17 : one of the reasons is that when @xmath18 , instability occurs , see @xcite .
suppose for instance that @xmath19 solves , and that @xmath20 solves , where @xmath21 replaced by @xmath22 , where @xmath23 is a sequence of real numbers going to zero as @xmath15 .
then there are some choices of @xmath23 for which @xmath24 for some sequence of time @xmath25 ( see @xcite , and @xcite for a similar phenomenon with different initial data ) .
therefore , producing reliable numerical tests in the case @xmath18 ( which is super - critical as far as wkb analysis is concerned @xcite ) seems to be a very delicate issue , that we leave out in the present paper .
the rest of this paper is structured as follows . in section [ sec :
analyt ] , we recall the general approach of wkb analysis for the schrdinger equation , the arguments of @xcite , and the rigorous results available for the semi - classical limit of when a caustic reduced to a point is formed .
we then recall the definition of the scattering operator .
we also give heuristic arguments to tackle the case of a `` supercritical focal point '' , and to guess what the critical indices are when a cusp caustic is formed , instead of a focal point . in section [ sec :
numgen ] , we present the different strategies that have been followed in the literature to study numerically the semi - classical limit for nonlinear schrdinger equations .
numerical experiments on the semi - classical limit for in the presence of a focal point appear in section [ sec : foc ] , and the scattering operator is simulated in section [ sec : scattnum ] .
we present the numerical experiments of the semi - classical limit for in the presence of a cusp caustic in section [ sec : cusp ] , and make conclusive remarks in section [ sec : concl ] .
consider the initial value problem , for @xmath26 : @xmath27 the aim of wkb methods is to describe the asymptotic behavior of @xmath28 as @xmath15 .
for instance , @xmath14 can be related to the planck constant , and the asymptotic behavior of @xmath28 is expected to yield a good description of @xmath28 when @xmath14 is fixed , but small compared to the other parameters .
more precisely , seek @xmath28 of the form @xmath29 plugging this expansion into and canceling the @xmath30 term , we see that the phase @xmath31 must solve the eikonal equation : @xmath32 to cancel the @xmath33 term , the leading order amplitude solves the transport equation : @xmath34 the eikonal equation is solved thanks to hamilton - jacobi theory : @xmath31 is constructed locally in space and time ( see e.g. @xcite for a discussion on this aspect ) .
even if @xmath35 is smooth , @xmath31 develops singularities in finite time in general : the locus where @xmath31 is singular is called _ caustic _ ( see e.g. the second volume of @xcite ) . when @xmath31 becomes singular
, all the terms @xmath36 may become singular as well .
one easily observes that ( [ eq : transport ] ) admits a divergence form " : @xmath37 . to illustrate this general discussion , we consider two examples that will organize the rest of this paper .
+ _ example ( quadratic phase ) .
_ let @xmath38 . then and
can be solved explicitly : @xmath39 this shows that as @xmath40 , @xmath31 and @xmath41 become singular : the wave @xmath19 focuses at the origin .
this example can be viewed as the smooth counterpart of the cauchy problem @xmath42 fourier analysis shows that @xmath43 , the dirac measure at the origin .
+ of course , the solution of can be represented as an oscillatory integral : @xmath44 the caustic set is exactly the locus where the critical points for the phase @xmath45 are degenerate . outside the caustic ,
an approximation of @xmath28 is given by the stationary phase theorem ( that we recalled as simply as possible in @xcite ) .
this leads us to the second example we shall consider numerically : + _ example ( cusp ) .
_ let @xmath7 and @xmath46 .
the set of degenerate critical points for @xmath47 ( caustic ) is given implicitly by : @xmath48 as soon as @xmath49 , a caustic is formed ( see figure 2 in @xcite ) .
+ when considering the asymptotic behavior of @xmath19 beyond the caustic , two main features must be considered : the creation of other phases , and goes to infinity as @xmath13 - rapid oscillations .
the wavelength may be proportional to @xmath14 , or , say , to @xmath50 .
] , and the maslov index ( see @xcite for more general linear equations ) . in the case of a focal point
, the first aspect does not exist : there is no creation of phase , and one phase is enough to describe @xmath28 past the focal point @xmath51 .
one can prove easily the following result : [ lem1 ] let @xmath52 and @xmath53 . if @xmath54 , then the asymptotic behavior ( in @xmath55 ) of the solution @xmath28 to is given by : @xmath56 in this example , the maslov index is @xmath57 . in the case of the cusp
, three phases must be considered to describe the asymptotic behavior of @xmath28 beyond the caustic ( see e.g. @xcite ) .
+ for future discussion on the numerical results , we state the following more precise result , which follows from the stationary phase theorem : [ lem2 ] let @xmath52 and @xmath53 .
if @xmath54 , then the asymptotic behavior of the solution @xmath28 to at time @xmath58 is given by : @xmath59 consider now the perturbation of with a nonlinear term : @xmath60 the sign of the nonlinearity is chosen so that no finite time blow - up occurs
. the following two important quantities are formally independent of time : @xmath61 we refer to @xcite for a justification .
fix the power @xmath62 of the nonlinearity , and consider different values for @xmath63 .
two notions of criticality arise : for the wkb methods on the one hand , and for the caustic crossing on the other hand .
this discussion is presented in @xcite for conservation laws , and we summarize it in the case of . plugging an expansion of the form into , we see that the value @xmath64 is critical for the wkb methods : if @xmath65 , then the nonlinearity does not affect the transport equation ( `` linear propagation '' ) , while if @xmath64 , then the nonlinearity appears in the right hand side of ( `` nonlinear propagation '' ) .
recall that in this paper , we always assume @xmath66 .
therefore , the eikonal equation is not altered : the geometry of the propagation remains the same as in the linear wkb approach , and we have to face the same caustic sets .
the idea presented in @xcite consists in saying that according to the geometry of the caustic , different notions of criticality exist , as far as @xmath63 is concerned , near the caustic . in the linear setting
, the influence of the caustic is relevant only in a neighborhood of this set ( essentially , in a boundary layer whose size depends on @xmath14 and the geometry of @xmath67 ) .
view the nonlinearity in as a potential , and assume that the nonlinear effects are negligible near the caustic : then @xmath68 near @xmath67 .
view the term @xmath69 as a ( nonlinear ) potential .
the average nonlinear effect near @xmath67 is expected to be : @xmath70 where @xmath71 is the region where caustic effects are relevant , and the factor @xmath72 is due to the integration in time ( recall that there is an @xmath14 in front of the time derivative in ) .
the idea of this heuristic argument is that when the nonlinear effects are negligible near @xmath67 ( in the sense that the uniform norm of @xmath73 is small compared to that of @xmath28 near @xmath71 ) , the above approximation should be valid . on the other hand
, it is expected that it ceases to be valid precisely when nonlinear effects can no longer be neglected near the caustic : @xmath73 is of the same order of magnitude as @xmath28 in @xmath74 , or even larger . practically , assume that in the linear case , @xmath28 has an amplitude @xmath75 in a boundary layer of size @xmath76 ; then the above quantity is @xmath77 the value @xmath78 is then critical when the above cumulated effects are not negligible : @xmath79 when @xmath80 , the nonlinear effects are expected to be negligible near the caustic : resuming the terminology of @xcite , we speak of `` linear caustic '' .
the case @xmath81 is called `` nonlinear caustic '' . to conclude this paragraph
, we examine this approach in the case of our two examples . in the case of a focal point
, we have @xmath82 and @xmath83 .
this leads us to the value : @xmath84 in the case of the cusp in dimension one , we have @xmath85 and @xmath86 ( which can be viewed thanks to the airy function and its asymptotic expansion , see e.g. @xcite or @xcite ) , which yields : @xmath87 one aspect of the numerical experiments presented below is to test this notion of criticality in those two examples . in this paragraph , we assume @xmath88 .
a complete justification of the above discussion is available @xcite : + [ cols="<,^,^",options="header " , ] consider @xmath89 $ ] , which includes the caustic crossing .
the above tables means : * if @xmath90 , then @xmath19 can be approximated by @xmath28 for @xmath89 $ ] . * if @xmath91 , then the nonlinearity is negligible near the focal point , but not away from it . * if @xmath92 , then nonlinear effects are relevant near the focal point , and only near the focal point . * if @xmath93 , then the nonlinearity is never negligible .
we give some precisions in some cases of interest for the numerics presented below .
+ in the one - dimensional case @xmath7 , the following pointwise estimate is proved in @xcite when @xmath94 or @xmath95 : @xmath96 setting @xmath97 , we see that @xmath98 the usual energy estimate yields , for @xmath99 : @xmath100 using , we infer , for @xmath89 $ ] : @xmath101\}}\frac{d\tau}{|\tau -1|^\si}+\int_{|\tau-1|\le \e}\frac{d\tau}{\e^\si } \)\lesssim \e^{\alpha -\si}.\ ] ] using the operator @xmath102 and @xmath103 , and gagliardo nirenberg inequalities as in @xcite , we find : [ lem3 ] let @xmath7 , @xmath94 , @xmath53 , and @xmath54 .
then we have , for the solutions of and : @xmath104.\ ] ] in particular , lemma [ lem3 ] implies , at time @xmath58 : @xmath105 ( the above result is true also when @xmath106 , but becomes far less interesting . ) we now explain the critical case @xmath92 .
the nonlinear effects near the focal point are described in terms of the scattering operator associated to the nonlinear schrdinger equation .
we rapidly present this operator @xmath107 in section [ sec : scatt ] .
we have then : @xmath108 where @xmath109 is the conjugate of @xmath107 by the fourier transform ( see @xcite ) , @xmath110 ( since @xmath107 is a nonlinear operator , the normalization of the fourier transform is an important detail . ) to conclude this paragraph , we give a few hints of what happens or may happen when the propagation is linear , and the caustic is super - critical , that is @xmath111 .
first , the conservation of mass and energy seem to rule out the possibility of a concentration of the form @xmath112 for some function @xmath113 independent of @xmath14 .
the above relation holds for @xmath28 and @xmath19 when @xmath114 , with @xmath115 ( for @xmath28 , this is obvious from ) .
when @xmath92 ( linear propagation and nonlinear focal point ) , the above relation still holds , with a different profile @xmath113 ( see @xcite ) .
now we see that the energy is bounded as @xmath13 : @xmath116 plugging a concentrating profile as in in the second term of the energy would yield , thanks to the conservation of mass : @xmath117 which is impossible since the energy is the sum of two positive terms .
this suggests two possible effects : near @xmath118 , the amplification of the solution ( in terms of powers of @xmath14 ) may be weaker than in the linear case ; on the other hand , nonlinear effects near the caustic should affect the _ phase _ of the solution @xmath19 in a rather strong way , causing the appearance of new frequencies .
a partial justification of the last assertion may be found in @xcite . extrapolating this argument
, we expect that in the supercritical case for a cusp ( with @xmath7 : @xmath119 ) , new frequencies appear .
if as in the linear case , three phases are necessary to describe the solution past the caustic , then the nonlinear interaction of these phases might reveal the presence of new frequencies , even on the modulus of @xmath19 ( see sect . [
sec : cusp ] for numerical tests that seem to confirm this heuristics ) .
to explain what the operator @xmath107 mentioned in the previous section is , consider the nonlinear schrdinger equation @xmath120 where @xmath121 is the propagator of the linear equation . to construct the scattering operator ,
we first want to give a meaning to when @xmath122 .
this means that the nonlinear effects are asymptotically negligible as @xmath123 : for instance , we expect at least @xmath124 this gives a rigorous meaning to the relation @xmath125 which aims at saying that as time goes to @xmath126 , the nonlinear dynamics associated to can be compared to the free dynamics given by @xmath127 . to define a scattering operator , we want to be able to say that as @xmath128 as well , the nonlinear effect are asymptotically negligible
that is , there exists @xmath129 such that @xmath130 the scattering operator is then defined as @xmath131 .
since our numerical experiments concern the one - dimensional case , we recall the existence of the scattering operator in this setting , and refer to @xcite for some extensions to the multidimensional framework : define @xmath132 [ prop : scattering ] let @xmath7 , and assume @xmath133 . * for every @xmath134 , there exists a unique @xmath135 such that the maximal solution @xmath136 to satisfies @xmath137 and @xmath138 * for every @xmath135 , there exists a unique @xmath139 such that the maximal solution @xmath136 to with @xmath137 satisfies @xmath140 the scattering operator is @xmath141 .
when @xmath142 , the above conclusions remain , in a neighborhood of the origin .
when @xmath143 , we also have : * for every @xmath144 , there exist a unique solution @xmath145 to and a unique @xmath146 such that : @xmath147 when @xmath148 , the above conclusions are false : if @xmath8 for instance , and if @xmath149 with @xmath150 in @xmath151 as @xmath152 , then @xmath153 .
one can not compare the nonlinear dynamics with the free dynamics ( see @xcite ) .
+ note that even though the scattering is proven to exist , very few of its features are known .
we refer to @xcite for some algebraic properties .
at least , this operator is not trivial : near the origin , it is a non - trivial perturbation of the identity ( see @xcite ) .
hereafter we restrict our discussion to the one - dimensional case , that is to say @xmath7 in all the preceding considerations .
it is interesting to notice that in the numerical literature , the nonlinear equation is treated exactly the same way the linear one would be in the presence of a potential @xmath154 on its right - hand side .
the strategy is called _ time - splitting _ , in its first or second order version ( lie or strang splitting , see e.g. @xcite ) where one alternates every time step @xmath155 between the solving of the laplace operator and the handling of the ( nonlinear ) differential equation . according to section [ sec : scatt ]
, @xmath156 will still stand for the free propagator , whereas we shall use @xmath154 as the ode solver ; lie time - splitting algorithms generate the following type of approximation for , @xmath157^n u^\e(t=0,.),\ ] ] and strang splittings , @xmath158^{n-1 } v(\dt/2 ) u^\e(t=0,.).\ ] ] many references exist ; let us quote only @xcite . on the contrary ,
few rigorous convergence results are available , hence we shall mainly recall the results from @xcite which quantify accurately the _ splitting errors _ assuming each time - step is performed exactly . under this assumption , there holds : ( @xcite , theorem 4.1 ) for any @xmath159 and @xmath160 , there exists a constant @xmath161 depending on the initial data for and @xmath162 , such that for @xmath163 $ ] and @xmath164 , @xmath165 if moreover @xmath166 , then there holds under the same assumptions : @xmath167 this result is concerned with splitting errors only and relies on the knowledge of the exact solution operators @xmath156 and @xmath154 . in order to stick to this framework in the context of smooth solutions , it is rather natural to approximate @xmath156 by means of a fourier scheme taking advantage of optimized fft routines , as proposed in the paper @xcite . moreover
, this will guarantee that the @xmath151 norm ( mass ) of the numerical solution will be conserved up to round - off errors .
unfortunately , the hamiltonian @xmath168 is generally not preserved ; a method conserving both quantities exists ( see the so called mcn algorithm , page 253 of @xcite ) but it would nt be efficient in the semiclassical regime because of the results in @xcite .
this is the main purpose of the paper @xcite to illustrate the ( surprising ) fact that in semi - classical regime , usual finite - difference schemes for can deliver very wrong approximations without any particular sign of instability in case very restrictive meshing constraints turn out to be bypassed
. this can be quite easily checked through the location of caustics , for instance .
the analysis of those standard schemes has been carried out by means of wigner measures , so the conclusions hold essentially for the quadratic observables coming out of the wave function itself .
this class of schemes became popular after the publication of @xcite , mainly because treating the differential part of ( [ eq : nls ] ) by means of a discrete fourier transform looked very much like being the best possible compromise in terms of meshing constraints .
indeed , in the linear case where ( [ eq : schrodlibre ] ) is supplemented with a potential @xmath169 on its right - hand side , it was shown that the time - step @xmath170 could be chosen independent of @xmath14 whereas the space discretization has to satisfy @xmath171 .
this was already much better when compared to finite - differences ; moreover , the method is naturally @xmath151-conservative . in @xcite
, these authors extended their fourier framework " to the weakly nonlinear schrdinger equations of the form ( [ eq : nls ] ) .
however , and despite the fact we do believe these `` fft time - split schemes '' realize the best numerical strategy in terms of gridding , we shall point out some shortcomings of the method in the next section .
we present here a preliminary result about truncation errors in lebesgue spaces for fourier schemes ; its proof follows directly from the strichartz estimates on the torus due to j. bourgain @xcite ( see also @xcite ) , and from the study of fft by m. taylor @xcite . its derivation is not obvious though as it applies directly to widely - used schemes like the one recalled in the forthcoming section .
we restrict our attention to the 1d free schrdinger equation with @xmath172 , and periodic boundary conditions : @xmath173 .
hence we start from @xmath174.\ ] ] we have explicitly : @xmath175 in order to investigate the behavior of the fft - scheme involving a finite even number @xmath176 of modes , we introduce the discrete fourier transform of a continuous function @xmath21 on @xmath177 $ ] as follows : @xmath178 from @xcite , we recall : [ lem_tay]if the continuous function @xmath21 has a convergent fourier series , then : @xmath179 now , what the fourier numerical scheme really computes is : @xmath180 thus the corresponding numerical solution @xmath181 is built : @xmath182 the question is therefore to study the discrepancy between @xmath183 and @xmath181 : @xmath184 the second term @xmath185 in this last equality can be interpreted as a usual truncation error ; it means that some high frequencies in the solution are lost when discretizing the problem on a finite grid . from @xcite , we know that the modulus of this term can be controlled as follows : @xmath186 which is satisfactory provided the initial data @xmath187 is smooth , e.g. @xmath188 .
however , the first term @xmath189 is far more delicate and reveals the propagation of the errors coming from the dft / fft itself .
it can be controlled thanks to the periodic strichartz estimates proved by j. bourgain . from @xcite , we recall : [ lem_bou ] given any complex - valued sequence @xmath190 , the following estimates hold for the corresponding functions in @xmath177 \times [ 0,2\pi]$ ] : @xmath191 the second estimate looks more attractive as it
sees " the finite number of modes .
we therefore deduce that the first term can be controlled in @xmath192 by means of : @xmath193 for instance , if @xmath187 is a finite superposition of fourier modes , then it is clear that this term cancels for @xmath194 large enough as @xmath195 in the summation ; obviously , the second term @xmath185 vanishes too .
the general case is nt completely clear yet .
this section aims at visualizing the asymptotics previously recalled ; namely we shall compare numerical approximations of and in 1d ( @xmath7 ) for various values of the parameters @xmath63 and @xmath196 .
the initial wave function is rather simple : @xmath197.\ ] ] numerical results have been obtained through the time - splitting fft schemes recalled in the previous section ; we used 1024 modes and fixed @xmath198 .
it is convenient to observe results in @xmath58 since @xmath199 .
this case corresponds to @xmath200 and @xmath201 ; we expect to observe a decay of the absolute errors between @xmath28 and @xmath19 for values @xmath0 .
this is indeed the case , but fig .
[ pf1 ] shows even a bit more , namely it compares pointwise the following quantities : ( recall lemma [ lem1 ] ) @xmath202 on the left in fig .
[ pf1 ] , we obviously observe that the absolute errors are slightly bigger when considering the solution of the nonlinear equation , @xmath19 .
however , even for the free solution @xmath28 , one sees that the error does nt vanish despite the fact no time - splitting algorithm is needed . as the way of discretizing the solution reveals itself important , we include here the corresponding scilab routine for the free equation : clear;deff([y]=phase(x),[y=-0.5*(x-@xmath203)@xmath204 ; ] ) deff([y]=position(x),[y = exp(-2*(x-@xmath203)@xmath204) ] ) deff([y]=az(x),[y = position(x).*exp(i*phase(x)./epsilon) ] ) nmax=2@xmath205;n=-(nmax)/2:(nmax/2)-1;epsilon=1.0/150 ; xstart=0;xstop=2*@xmath203;dx=(xstop - xstart)/nmax;xstop = xstop - dx ; a = xstart : dx : xstop , initialdata = az(a ) ; vepsilon = fftshift(fft(initialdata,-1 ) ) ; vepsilon = exp(-i*epsilon*(@xmath206)).*vepsilon ; vepsilon = fft(fftshift(vepsilon),1 ) ; + clearly , its outcome is in agreement with lemma [ lem1 ] since @xmath14 is already quite small .
the maslov index is visible , up to an error around @xmath207 for 1024 fourier modes .
we now put @xmath208 and the outcome is displayed on fig .
[ pf2 ] ; we still compare the same quantities .
absolute errors on wave functions ( left side ) are much bigger for @xmath19 in this case .
in particular , no new frequencies appear in the numerical solutions .
the nonlinear effect boils down to a small change on the modulus of @xmath19 .
we close this first series of tests by considering @xmath200 and @xmath209 as shown in fig .
[ pf3 ] . of course , as no pointwise convergence is expected in this case , absolute errors are even bigger for both wave functions ( left side ) and moduli ( right side ) . of course , the size of the error on the modulus is much bigger too , and one should be extremely careful about the credit to give to the numerical simulations in the supercritical case . indeed , this is a regime where a small error can be amplified at leading order ( see @xcite ) .
we aim now at illustrating the results on scattering theory through numerical computations still achieved through time - splitting fft schemes .
the algorithm we used for the approximation of the scattering operator is based on a nonlinear time - splitting routine flanked by two free evolution steps ( implemented the way recalled in the previous section ) : @xmath210 with @xmath156 , @xmath211 standing for the solution operators of equations and in 1-d with @xmath212 respectively .
we used @xmath213 in the computations hereafter .
as no small parameter @xmath14 is present in the problem , one may think that no major obstacle exists in carrying out this program ; this is nt correct as the free evolutions can ( and do ! ) dramatically increase the size of the computational domain for large @xmath214 . it is interesting to notice that , in case one wants to use fft - based schemes , both the direct computation for small @xmath14 and the scattering operator approximation lead to a large computational domain difficulty " : in the fourier space for the first case , in the usual space for the second . a way to understand the scattering operator is to visualize the average effects of the nonlinearities appearing in equations of the form @xmath215 for various values of @xmath216 and @xmath217 .
intuitively , as @xmath196 increases , the nonlinearity becomes shorter range .
similarly , as @xmath217 increases , the nonlinearity becomes stronger , and it should take a larger amount of time before we can consider it has become negligible .
in all the tests we performed , it was somehow surprising to observe how fast the algorithm converges : one does not have to consider `` very large '' values of @xmath214 so that @xmath218 becomes stable and visually independent of @xmath214 .
the parameter @xmath217 controls in some sense the strength of the nonlinearity inside , as can be seen on fig .
this figure displays the position density of the initial data , the scattered solution for @xmath213 and a mixed state " @xmath219 . as our time - splitting / fft algorithm preserves only the @xmath151 norm , but not the hamiltonian , we first restricted ourselves to moderate values of @xmath220 ( defocusing case ) .
however , as a numerical experiment , we wanted to display the outcome of our scheme for the stronger case @xmath221 on fig .
[ run2 ] : notice the change of shape in the scattered solution . moreover , on this figure , we also tried to show what happens for @xmath222 , that is to say for the focusing case despite there may be finite time blow - up ( but there is scattering for small data ) .
we checked that the energy associated to this data is ( and remains ) positive , a case where the virial identity , @xcite , does not imply blow - up .
the computational domain for these runs was @xmath223 $ ] and @xmath224 fourier modes were used .
now let s observe the effects of lowering the @xmath196 value while keeping other parameters equal , see fig .
it is interesting to see that the change of shape appearing for @xmath221 is stronger than in the preceding case . on the contrary ,
the increase of the numerical solution s support is slightly less important .
this hints that increasing the @xmath196 value tends to expand the support of the scattered solution whereas increasing the @xmath226 value ( defocusing case ) leads to an oscillatory behavior .
however , we stress that since the energy , @xmath227 of the numerical solution changes more with a bigger @xmath217 ( its mass being always kept constant ) , these oscillations might be spurious . we actually do nt know how this fact can be decided ; our profiles have been checked to be stable on a finer grid . in order to get some numerical evidence about the dependence of the scattered solution on @xmath196
, we display on fig .
[ run4 ] the outcome for @xmath228 .
it is quite clear that the scattered solutions for both values of @xmath217 are less peaked .
their support is bigger and the oscillations for @xmath221 are weaker , their frequency remained the same though .
this agrees with the behavior we sketched in the preceding subsection as @xmath196 and @xmath217 vary .
let us now go back to comparing the quadratic observables generated by numerical approximation of equations and endowed with a small parameter @xmath14 in 1-d . in this section we fixed @xmath229 .
figure [ run - cusp ] displays the position density of the initial data for both equations , i.e. @xmath230,\ ] ] together with the position density of the numerical approximations of , in @xmath231 .
the point here is to investigate what happens for the case of such a self - interfering gaussian pulse , since no scattering theory is known for this problem .
what we would like to check is whether the theoretical results on the focus point recalled and visualized in the preceding sections can be thought of as a guideline for this more complex case involving a non - trivial caustic
. we shall observe position densities for the unique value of @xmath232 as a similar behavior has been seen to hold for different nonlinearities with convenient values of @xmath63 .
@xmath233 fourier modes have been used in order to produce these results .
this case could be referred to as subcritical since it is noticeable on the top of fig .
[ run - cusp ] that the free and the nonlinear numerical solutions do agree for this reasonably small value of @xmath14 .
in particular , the frequencies of oscillations are identical .
this is very similar compared to the behavior investigated in @xcite .
the parameter @xmath63 is now in a
critical range " as we observe that both solutions differ much more , but the frequency of the oscillations looks like being still the same in both cases . in order to establish this fact , we display on the left of fig.[fft - cusp ] the fft of the position densities : a peak at the same frequency is clearly noticeable .
the nonlinear effect manifests itself through a change of order zero in the moduli , as we already observed on the right side of fig .
[ pf2 ] ; notice also the similarity with the scattering state shown on fig .
[ run2 ] ( right ) .
this does agree with the @xmath236 value derived in section [ sec : heur ] in this last case , there is no similarity no more between the approximate solutions of , , as seen on both fig .
[ run - cusp ] and [ fft - cusp ] . especially , the right side of fig .
[ fft - cusp ] reveals that new frequencies show up inside the position density of the nonlinear solution .
we have therefore a change of order zero in the moduli and in the frequency .
this is of course reminiscent of fig .
[ pf3 ] in which a frequency doubling seems to show up in the supercritical regime .
we have presented the semi - classical limit for the nonlinear schrdinger equation in the presence of a caustic . when the caustic is reduced to a point ,
the numerical experiments are in good agreement with the analytical results as far as the notion of criticality is concerned .
however in the critical case , described by a nonlinear scattering operator , the leading order nonlinear effects are rather hard to visualize in the semi - classical limit .
this is why we simulated the scattering operator in a separate way .
our numerical tests give encouraging evidence of new phenomena concerning the phase of the wave in the supercritical case when a focal point is formed ( appearance of new frequencies ) . in the presence of a cusp caustic , the numerical experiments are in good agreement with the heuristic arguments that we presented here , for which no rigorous justification is available so far .
, _ an introduction to nonlinear schrdinger equations _ , in : nonlinear waves ( sapporo , 1995 ) , eds .
r. agemi , y. giga , and t. ozawa , gakuto international series , math .
sciences and appl . , gakktosho , tokyo , 1997 , pp .
85133 . , _ a case study on the reliability of multiphase wkb approximation for the one - dimensional schrdinger equation _ , in : numerical methods for hyperbolic and kinetic problems , vol . 7 of irma lect .
soc . , zrich , 2005 , pp . | the aim of this text is to develop on the asymptotics of some 1-d nonlinear schrdinger equations from both the theoretical and the numerical perspectives , when a caustic is formed .
we review rigorous results in the field and give some heuristics in cases where justification is still needed .
the scattering operator theory is recalled .
numerical experiments are carried out on the focus point singularity for which several results have been proven rigorously .
furthermore , the scattering operator is numerically studied .
finally , experiments on the cusp caustic are displayed , and similarities with the focus point are discussed . |
an osfr is made of a film of a chemically resistant gel , of uniform thickness @xmath1 .
one face is kept in contact with the homogeneous contents of a continuous stirred tank reactor ( cstr ) .
the other face is pressed against an impermeable wall ( fig .
[ osfr ] ) .
the coupled dynamical equations for the concentrations in the cstr and in the gel are respectively @xcite : @xmath2 and @xmath3 where @xmath4 , @xmath5 and @xmath6 are the concentrations of species @xmath7 respectively in the input flow , in the cstr , and inside the gel , @xmath8 is the corresponding diffusion coefficient , @xmath9 the residence time of the reactor ,
@xmath10 the ratio of the volume of the gel to the volume of the cstr , and @xmath11 the distance to the cstr / gel interface .
the @xmath12 s are the reaction rates . in the right hand side of eq .
( [ eqcstr ] ) , the three terms represent the change of the concentrations per time unit .
the first term gives the contribution of the reaction .
the second one represents the input and output flows of the species .
it contains all the expandable control parameters of the system , namely @xmath9 and the @xmath4 .
the third term results from the diffusive flux of the species at the interface between the gel and the cstr and represents the feedback of the gel contents on the cstr dynamics .
when the volume of the cstr is large in regard to the volume of the gel ( @xmath13 ) , this last term can generally be dropped so that the chemical state of the cstr is independent of the state of the gel and the concentrations in the cstr define a dirichlet boundary condition for eq .
( [ eqgel ] ) at @xmath14 , whereas a no - flux boundary condition is applied at @xmath15 . to avoid ambiguities ,
let us make more precise the terms we use .
whereas in usual osfrs the thickness @xmath1 is small in regard to the two other dimensions , in this paper , we want to investigate the behavior of the system along the direction orthogonal to the faces in contact with the cstr and the impermeable wall .
thus , we shall consider systems made of ` flat ' gels in which one of the dimensions parallel to these faces ( @xmath16 in fig .
[ osfr ] ) is still smaller and negligible .
such systems can be considered as two - dimensional .
nevertheless , the size @xmath1 , in the direction orthogonal to the faces , remains small in regard to the second direction parallel to these faces .
size @xmath1 and the direction orthogonal to the faces will be further referred to as the ` depth ' or the ` width ' of the gel , the negligible dimension @xmath16 becoming the ` thickness ' . a number of reactions which present autocatalytic steps are bistable in a cstr @xcite .
these reactions normally exhibit ` clock ' behavior in batch conditions , i.e. a more or less long induction period followed by a fast switch to the equilibrium . in these reactions ,
the transformation rate of the substrate(s ) tremendously increases during the switch which occurs when the level of the autocatalytic species is large enough . for residence times much shorter than the typical reaction time ,
i.e. at high flow rates , the composition inside the cstr is close to the composition in the input flow .
it is referred as the ` flow state ' or state f. for residence times much larger than the typical reaction time , i.e. at low flow rates , the reaction is almost completed in the cstr and the composition in the cstr is close to a thermodynamic equilibrium .
this is referred as the ` thermodynamic state ' or state t. for intermediate flow rates , both states are stable for the same composition of the input flows .
spatial bistability can be observed when such a reaction is operated in an osfr , the cstr of which is set in the flow state .
although this phenomenon , which extends the concept of bistability to space , has been previously described in simple models @xcite and in real experiments @xcite , its importance has been recognized only recently and a detailed analysis was worked out in relation with the development of the osfrs @xcite . to understand the basic principles of spatial bistability , let us consider the concentration profile of a substrate of concentration @xmath17 and diffusion coefficient @xmath18 consumed in the reaction and let us imagine that the depth @xmath1 of the gel can be changed in a continuous way .
the time taken by fresh reactants fed by diffusion at the cstr boundary to reach the opposite boundary is of the order of @xmath19 .
we start from @xmath20 and increase @xmath1 adiabatically so that the system is always in its asymptotic stationary state .
let @xmath21 be the induction time for the cstr composition ( in state f ) . while @xmath1 is small enough for @xmath22 to hold , the whole volume of the gel remains in state f. we shall still refer this state of the gel as a f state .
when @xmath23 reaches a value of the order of @xmath21 , for a value @xmath24 , the fast reaction starts at the impermeable wall and a steep front connecting the f state to a t state forms .
such a state will be referred as a ` mixed state ' or ft state .
but , the amount of the autocatalytic species is now large at the impermeable wall of the gel ( @xmath15 ) and can diffuse backwards switching on the reaction , so that the front can recede to a value @xmath25 .
actually , if one assumes that the reaction rate is extremely large inside the front , one can neglect the rate of the consumption of the substrate elsewhere and the concentration of the latter can be approximated by the solution of a diffusion equation between a fixed value at @xmath14 ( the cstr ) and @xmath26 ( the front ) .
then , it can easily be shown that : @xmath27 where @xmath28 is the total amount of substrate converted by time unit in the gel and @xmath29 is the variation of concentration between the f and the t states @xcite .
this value is practically independent of @xmath1 , so that one can now decrease @xmath1 down to a value @xmath30 before the system switches back from the ft to the f state .
thus , for @xmath31 , both concentrations profiles f and ft are stable .
this defines the spatial bistability .
the extent of the bistability domain depends on the concentrations in the cstr always close to those in the input flow and the residence time , so that various nonequilibrium phase diagrams can be built @xcite . in spite of its apparent crudeness ,
this theory gives accurate results .
extensive studies have been performed experimentally and numerically on the chlorine dioxide - iodide ( cdi ) reaction . up to now , differences in diffusion coefficients did not seem to play any role .
actually , whether the system is ruled by equal diffusion coefficients or by long range inhibition , it exhibits a standard spatial bistability .
we shall see that the situation is fairly more complex when long range activation comes into play .
the chlorite - tetrathionate ( c - t ) reaction has essentially been studied in closed systems for the determination of stoichiometric and kinetic aspects @xcite as well as a source for propagating cellular fronts @xcite . in excess of chlorite ,
the reaction kinetics is well approximated by the following overall balance equation @xmath32 and the reaction rate by @xmath33}}{{\rm d } t } } = k\mbox{[clo$_2 ^ -$][s$_4$o$_6^{2-}$][h$^+$]$^2 $ } \label{kinlaw}\ ] ] since , during the reaction , large ph changes are observed , the fast dissociation equilibria of @xmath34 and @xmath35 have also to be accounted for .
this rate law exhibits quadratic autocatalysis in [ @xmath36 .
although the validity of this rate law has been determined for a limited range of ph ( 4.7 - 5.5 ) @xcite , it has been successfully used in much larger domains . nevertheless , the original value @xmath37 m@xmath38s@xmath39 had to be adjusted to fit with experiments . the value @xmath40 m@xmath38s@xmath39 has recently been proposed to fit the experimental data of the bistability domain of the reaction in a cstr @xcite and will be used in the following . it is well known that , whereas most species have a diffusion coefficient close to 10@xmath41 @xmath42s@xmath39 , the autodiffusion coefficient of the proton h@xmath0 is almost ten times larger . nevertheless , accounting for the electroneutrality condition and the presence of the slowest ions , the effective diffusion of h@xmath0 is lower @xcite .
solving reaction - diffusion equations that include all ionic effects is a formidable task which has been only performed on simple models @xcite . for practical reasons ,
it is reasonable to use a constant and uniform effective value for each diffusion coefficient .
we shall assume that all diffusion coefficients are equal to a unique effective value @xmath43 , except for h@xmath0 and oh@xmath44 which have both a higher autodiffusion coefficient and are closely related .
let @xmath45 be a common value for these two species . with these assumptions , it was shown in ref @xcite that the experimental data can be correctly interpreted only if @xmath46 ranges slightly above 3 .
thus it is necessary to take into account that the autocatalytic species h@xmath0 diffuses much faster than the inhibitory species , so that the reaction - diffusion system enters the class of long range activation . in order to make extensive computations
, we shall first define a simple model , the properties of which are close enough to the original system to draw significant conclusions .
our main purpose is that , due to its formal simplicity , this model can be further used as a prototype for long range activation .
the model relies on two strong assumptions : a ) the kinetics is completely described by eqs .
( [ kin ] , [ kinlaw ] ) .
b ) the concentrations at the cstr boundary are fixed to their values in the input flow , i.e. we assume a residence time @xmath47 ( infinite flow rate ) . in these conditions ,
the reaction - diffusion system is : @xmath48[{\mbox{\rm s$_4$o$_6^{2-}$}}][{\mbox{\rm h$^+$}}]^2 + d_1\nabla^2[{\mbox{\rm clo$_2 ^ -$ } } ] \nonumber \\ { \frac{\partial { \mbox{\rm s$_4$o$_6^{2-}$}}}{\partial t } } & = & -2k[{\mbox{\rm clo$_2 ^ -$}}][{\mbox{\rm s$_4$o$_6^{2-}$}}][{\mbox{\rm h$^+$}}]^2 + d_1\nabla^2[{\mbox{\rm s$_4$o$_6^{2-}$ } } ] \label{model1}\\ { \frac{\partial { \mbox{\rm h$^+$}}}{\partial t } } & = & 12k[{\mbox{\rm clo$_2 ^ -$}}][{\mbox{\rm s$_4$o$_6^{2-}$}}][{\mbox{\rm h$^+$}}]^2 + d_2\nabla^2[{\mbox{\rm h$^+$ } } ] \nonumber\end{aligned}\ ] ] in this paper , we do not consider short transients that follow a change of parameters and only consider the asymptotic regime . eliminating the reaction term between the two first equations , one is left with a diffusion equation with a fixed boundary condition at @xmath14 . in the long time limit , the solution is @xmath49 - 7[{\mbox{\rm s$_4$o$_6^{2-}$}}]=2[{\mbox{\rm clo$_2 ^ -$}}]_0 - 7[{\mbox{\rm s$_4$o$_6^{2-}$}}]_0 $ ] , where index 0 holds for the concentrations fixed at the boundary .
then [ ] can be expressed as a function of [ ] and can be eliminated from eqs .
( [ model1 ] ) .
one gets the two - variable reduced system : @xmath50 where we have introduced the new time and space scales @xmath51_0)^3t$ ] , @xmath52_0)^{3/2}r$ ] and the following normalized variables and parameters : @xmath53/[{\mbox{\rm clo$_2 ^ -$}}]_0 $ ] , @xmath54/[{\mbox{\rm clo$_2 ^ -$}}]_0 $ ] , @xmath46 and @xmath55_0/[{\mbox{\rm clo$_2 ^ -$}}]_0 $ ] .
this system is formerly identical to the system used in previous studies in closed systems @xcite . according to the definition of @xmath56
, one has @xmath57 at the cstr boundary .
thus , the control parameters are @xmath58 , @xmath59 , and @xmath60 , where @xmath58 is the fixed value of @xmath61 at the cstr boundary .
parameter @xmath60 represents the normalized excess of chlorite .
provided it is not too large this is the case in most experiments it was found to play a minor role in the model and does not bring qualitative changes in the results .
thus , it was fixed at @xmath62 which would correspond to exact stoichiometric proportions of chorite and tetrathionate .
note that , if @xmath63 , the third equation could be eliminated in the same way and the system would be reduced to a unique variable @xmath56 .
a similar reduction to the same one - variable equation could also be done for a homogeneous cstr .
a one variable system can be bistable but never oscillating or excitable , since two dynamical time scales at least are necessary .
this suggests that the c - t reaction can be neither oscillating or excitable in a cstr .
more sophisticated models support this assertion , which is confirmed by experiments .
this model suffers from serious limitations .
first , at the cstr boundary , we use fixed concentrations equal to the concentrations in the input flow . since , in the flow state , the concentrations should be close to those in the input flows , this seems to be a reasonable approximation . nevertheless , at finite residence times , the values of @xmath58 could correspond to non existent or unstable states of the cstr .
the approximation is strictly valid only in the practically unreachable limit of infinite flow rates ( @xmath47 ) . in the same way
, one can fix @xmath58 to arbitrary high values for which the cstr would necessarily be in a t state .
this can lead to ambiguities in the definition of the spatial states f and t and to experimentally unreachable regions in phase diagrams so that some care is necessary for their interpretations .
finally , values of @xmath58 can also be taken arbitrary low , down to zero , an unrealistic situation since the dissociation equilibrium of water imposes that @xmath64 $ ] remains positive and finite .
thus , this model must be only thought as a ` toy model ' , with the advantage that the kinetics remains closely related to that of a real system .
this will allow us to evidence dynamical phenomena induced by long range activation .
such models have also been recently used by benyaich et al . in studies of spatial bistability @xcite . to make direct comparison with experiments ,
a more elaborated model has to be devised . to avoid oddities due to the simple model oversimplifications
, one has to account , not only for the dynamics of the cstr at @xmath14 included in the eq .
( [ eqcstr ] ) , but also for the way in which the h@xmath0 ions are introduced in the reactor and the equilibria in which they are involved . in the extended model that we used to make comparisons with the experimental results ,
the following fast equilibria are taken into account in addition to eq .
( [ kin ] ) : @xmath65 with rate law @xmath66[{\mbox{\rm oh$^-$}}]\ ] ] and @xmath67 with rate law @xmath68-k_a^-\;[{\mbox{\rm h$^+$}}][{\mbox{\rm so$_4^{2-}$}}]\ ] ] discarding the species that do not contribute to kinetics in eqs .
( [ kin ] , [ kineau ] , [ kinsulf ] ) , one retains the six variables @xmath64 $ ] , @xmath69 $ ] , @xmath70 $ ] , @xmath71 $ ] , @xmath72 $ ] and @xmath73 $ ] .
the following values are used for the kinetic constants : @xmath74 m s@xmath39 , @xmath75 with @xmath76 , @xmath77 m s@xmath39 , @xmath78 with pk@xmath79 .
the diffusion coefficient @xmath43 is fixed to a standard value @xmath43=10@xmath41 @xmath42s@xmath39 and the fast species diffusion coefficient @xmath45 is fixed to @xmath80 @xmath42s@xmath39 to fit at best the experimental data .
accurate one - dimensional numerical solutions of the resulting reaction - diffusion system have been obtained by an implicit finite difference method for stiff systems with adaptative grid in time and space .
unfortunately , due to the higher complexity and the extreme stiffness of the model , it was impossible to make systematic studies of this model in higher dimensions .
two - dimensional computations were limited to a few points in parameter space . in order to avoid problems in the definition of the input concentrations and to make easier the comparisons with the experimental data and a previous work , we use , as the main control parameter , the quantity @xmath81 defined in the experimental section below . in order to resolve the concentration profiles in the depth @xmath1 of the reactor and to minimize both technical difficulties and parasitic edge effects ,
we have developed one - side - fed flat annular reactors , i.e. annular osfrs .
complete descriptions and technical details on such reactors can be found in ref .
these osfrs are made of a flat annular piece of 2% agarose gel , of thickness @xmath820.5 mm .
they have a fixed outer radius @xmath83=12.5 mm and an inner radius @xmath84 such that the width @xmath85 ranges from 0.5 to 3 mm .
the agarose gel can be considered as chemically inert in regard to the reaction . in this soft hydrogel ,
the diffusion coefficients are close to those in pure water .
the ratio @xmath86 is large enough for the curvature effects to be negligible .
the inner rim of the annulus is tighten against the impermeable boundary while the outer rim is in contact with the contents of a cstr of 25 @xmath87 , fed through a single inlet port with appropriately premixed solutions of reagents stored in separated reservoirs .
the residence time of the reactor @xmath9=600
s , the temperature t=25@xmath88c , and the input flow concentrations of sodium chlorite [ naclo@xmath89@xmath90 m and potassium tetrathionate [ k@xmath91s@xmath92o@xmath93@xmath94=0.5@xmath95 m are kept fixed during all the experiments . besides the two above mentioned reagents , variable mixtures of perchloric acid and sodium hydroxide
are added to the input flow of reagents to control the ph of the premixed feed solution .
this is done by pumping different volume ratios of acid and base , keeping the overall flow rate constant . for experimental and graphic convenience
, this relative distribution is characterized by a control parameter @xmath81 on which the acid and base flow concentrations depend linearly : one has [ hclo@xmath92]@xmath94 = @xmath97 m and [ naoh]@xmath94 = @xmath98 m. the ph of the feed mixture decreases from a basic to an acid state as @xmath81 increases from 0 to 1 .
the parameter @xmath81 effectively controls the ph of the solution and is the only expandable control parameter in this series of experiments .
the patterns are made visible by bromophenol blue ( 1% solution ) , a ph sensitive indicator that turns from dark purple in the basic state to pale yellow in the acid state .
patterns are monitored by a ccd camera connected to a time lapse video recorder .
the dynamics of the c - t reaction operated in the cstr exhibits bistability in the range @xmath99 between a f branch ( with ph values typically above 9 ) and a t branch ( with ph values typically below 3 ) .
the upper limit corresponds to the value of @xmath81 where the input flow switches from a basic to an acid solution .
the lower limit was used to adjust the rate constant @xmath100 to the value @xmath40
m@xmath38s@xmath39 @xcite .
in the preliminary experiments of ref .
@xcite , we had observed , in addition to spatial bistability , oscillations of the ft state front and excitability phenomena in the domain where the sole f state is stable .
while the stability of concentration profiles can be determined with 1-d simulations , the latter needs 2-d computations .
thus , the stability limits are determined in 1-d calculations by the same accurate implicit method as the full model and the existence of excitability domain is studied in two dimensional simulations , performed with two different numerical methods , namely an implicit finite difference hopscotch method with automatic step control and a finite element method .
first , choosing a typical depth value @xmath101 ( expressed in reduced units precedently defined ) , we have built a stability diagram in the plane @xmath102 ( fig .
[ diagdv ] ) . at low values of @xmath58 and @xmath59
, there is a large region of spatial bistability that extends down to @xmath103 . when @xmath104 the ` acid ' state must obviously be obtained by a continuous decrease of @xmath58 from another ` acid ' state or by an acid perturbation of the system . in fig .
[ proftoy ] , we have represented the two stable concentration profiles through the depth of the gel for a given set of control parameters .
contrary to the standard case , in the ft state , these profiles are always monotonous and do not exhibit a steep front .
the states are nevertheless well characterized by the low ( ` basic ' ) or high ( ` acid ' ) value of @xmath61 at @xmath15 , the impermeable boundary of the gel .
when @xmath105 , the system exhibits temporal oscillations in place of bistability .
since the system can not oscillate in homogeneous conditions , these oscillations clearly result from the different time scales introduced by the differential diffusion .
although the problem is of spatial nature , the stability limits present the characteristic topology of the so - called ` cross - shaped diagram ' that connects bistability and oscillations of numerous reactions in homogeneous conditions @xcite .
a second diagram in the plane @xmath106 has been extensively computed for @xmath59=5.5 , a value chosen for the richness of the dynamical situations exhibited at this diffusion ratio ( fig .
[ diagvl ] ) . for small depths , the system is always in the flow ( or ` basic ' ) state , as expected . at intermediate values of depth , there is a large domain of oscillations .
it vanishes at large @xmath58 values for which one continuously changes from the ` basic ' to the ` acid ' state when @xmath1 is increased . for large depths ,
the ft ( or ` acid ' ) state is again present down to @xmath103 , which defines a domain of spatial bistability at small @xmath58 values .
the most interesting part of this diagram is the region labeled ` bist.2 ' in fig .
[ diagvl](b ) . in this narrow region ,
the system still exhibits spatial bistability , but the ` acid ' state becomes oscillating .
thus , the bistability domain is divided in two regions , one of standard bistability between two stationary concentration profiles ( ` bist.1 ' ) , and a second one between a stationary profile and an oscillating one ( ` bist.2 ' ) . moreover ,
when one decreases @xmath1 and crosses the stability limit of the oscillatory ` acid ' state , the unique stable state , namely the f ( ` basic ' ) state , exhibits particular properties .
if , in this stable state , an acid perturbation ( such as @xmath107 ) is applied close to the impermeable wall , the perturbation does not smoothly decay to the stable state but stimulates a further increase of @xmath108 ions concentration in its vicinity . a pair of pulses forms and propagates along the wall in opposite directions .
after relaxation of transients , they travel with constant shape , amplitude and velocity , which precisely defines the phenomenom of excitability .
the limits of this excitability domain are reported in fig .
[ diagvl](a ) .
this domain extends all the way to the oscillation zone . in the vicinity of this latter zone ,
the recovery tail of the pulse can become oscillatory in space .
although , for more efficiency , these limits have been accurately computed in a rectangular reactor , we have also also performed some runs in an annulus , analog to that of the experimental device .
an example of an excitation pulse propagating in such an annulus ( partial view ) is represented in fig .
[ pulse ] .
, @xmath109 .
concentration map of @xmath61 : from minimum ( black ) to maximum ( white).,width=321 ] the limits of this excitability domain are reported in fig.[diagvl](a ) .
this domain extends to the oscillation zone . in the vicinity of this latter zone
, the recovery tail of the pulse can become oscillatory in space .
in this section , we report the numerical results that have been obtained with the extended model and simultaneously make a direct comparison with the experimental data .
we consider the existence and stability of the f state and the ft state in the plane ( @xmath110 ) .
spatial bistability is observed over a significant domain of the plane .
for instance , in fig .
[ numprof ] , we show the two stable concentration profiles ( f and ft states ) computed for a same set of control parameter corresponding to a point located inside the bistability domain .
= 1 mm , @xmath81=0.72 .
dotted lines : f state ; full lines : ft state.,width=321 ] contrary to the case of the simple model , the ft state exhibits a steep front .
the concentration profile of the autocatalytic species @xmath108 is plotted in logarithmic scale to emphasize on the position and the stiffness of the front .
the concentration of @xmath111 and oh@xmath44 are represented in linear coordinates to underline the linear drop of the substrates concentrations as predicted by eq .
( [ valdelta ] ) . in experiments ,
the ph color indicator reveals the two different states @xcite . in the ft state ,
the gel is found to remain in a basic state only in a narrow band located along the outer rim .
the nature of the different dynamical states were determined unambiguously in the whole parameter range that was explored . in fig .
[ diag34 ] , we report the computed limits of the bistability domain ( full lines ) in the plane ( @xmath110 ) and the experimental points .
, @xmath1 ) .
experimental points : @xmath112 monostable with non excitable f state , @xmath113 monostable with excitable f state , @xmath114 bistable with oscillatory ft state , @xmath115 bistable with stationary ft state , @xmath116 the cstr switches to the thermodynamic state . computed results : the full lines are the limits of the bistability domain , the dotted line is the limit of the oscillory ft state within the bistability region , + marks are points for which the f state has been shown to be excitable.,width=321 ] gel annuli with @xmath117 mm could not be technically achieved , so that this part of the diagram can not be explored experimentally .
numerically , the upper limit @xmath118 of the ft state is slightly lower than the limit @xmath119 computed for the f state in the cstr .
actually , in the ft state , the concentration gradients of the substrates at @xmath14 are large as seen in fig .
[ numprof ] , so that , although @xmath120 , the last term in eq . ( [ eqcstr ] ) is not completely negligible and the contents of the cstr are entrained into the t state slightly before the limit @xmath121 is reached .
experimentally , this minor deviation is too small to be detected .
there is a first major difference with the standard cdi reaction @xcite : the spatial bistability persists at large values of @xmath1 , which seems in contradiction with our theoretical presentation asserting that the f state can not exist when the width of the gel is large .
actually , this results from the extreme slowness of the reaction when the system is in the basic f state .
this leads to a quasi infinite induction time .
thus , one can not switch spontaneously from the f state to the ft state by a continuous change of @xmath1 , if one remains within a realistic range . nevertheless , one reaches this state by an appropriate perturbation or choice of the initial condition in the gel , but the bistability domain extends to huge values of the depth @xmath1 .
contrary to the standard case of the cdi reaction , the limit of the ft state when @xmath81 decreases at constant @xmath1 is complex and exhibits dynamical phenomena both numerically and experimentally , provided that @xmath1 is large enough as seen in fig .
[ diag34 ] .
when the value of @xmath81 decreases and crosses the dotted line , the position of the front begins to oscillate and @xmath122 becomes a function of time .
this domain of oscillatory dynamics is narrow and difficult to explore precisely in experiments .
however , one dimensional numerical simulations allow for a precise exploration of this oscillatory domain . in fig .
[ oscil34 ] , we represent the numerical oscillations for decreasing values of @xmath81 at @xmath1=1 mm , but similar sequences are observed for other widths of the gel .
( @xmath1=1 mm ) : oscillation 1 : @xmath81=0.700 ; oscillation 2 : @xmath81=0.692 ; period 2 : @xmath81=0.6895 ; period 4 : @xmath81=0.6894 ; period 8 : @xmath81=0.6892 ; chaos : @xmath81=0.6885.,width=321 ] when @xmath81 decreases , the amplitude of the oscillations increases and they take a more and more relaxation character . when the @xmath81 values are drawn still closer to the stability of the ft state , one observes , over a very narrow range of @xmath81 values , a period doubling sequence leading to chaotic oscillations .
when @xmath81 decreases from the onset of oscillations , the period of the oscillations slowly increases . on the contrary , the period strongly increases when the width @xmath1 increases . since the experimental conditions are never strictly uniform all along the outer rim of the gel and since the width @xmath1 can not be perfectly constant , the oscillations of the front lead to the formation of pacemakers that emit travelling waves with periods comparable to those found in the computations . beyond the stability limit of the ft front ,
dynamical phenomena persist in the form of excitability . in a large domain of parameters , an acid perturbation , made at a localized place of the gel annulus , propagates under the form of two opposite travelling pulses along the inner rim .
after a short transient , each pulse takes a stationary shape and travels without deformation along the annulus ( fig .
[ excit ] ) .
= 0.3 mm , @xmath81=0.48.,width=302 ] the length of the pulse tail decreases as @xmath81 decreases until the excitability is eventually lost .
when the pulses meet , they anihilate each other , but by adding a basic perturbation at the right place and time , it is possible to kill one of the two pulses .
then , the surviving pulse can rotate indefinitely along the annulus as previously done with the belousov - zhabotinski reaction , the paradigm of excitable reactions @xcite .
this behavior is identical to that observed with the simple model and is also reproduced with the extended model for the two different sets of parameters that were checked for excitability .
it is known that , in homogeneous bistable systems , the addition of a negative feedback that introduces a new time scale can induce oscillations . in the same way
, excitability can arise from the coupling of a bistable subsystem that exhibits a s - shaped nulcline with the feedback dynamics of a recovery variable .
our simple model has clearly shown that the additional time scales introduced by long range activation can play a similar role and are able to induce excitability and spatio - temporal oscillations in a system which can only be bistable in homogeneous conditions .
when the diffusion of the activator is not significantly higher than the diffusion of the other species , the system can only exhibit spatial bistability .
but , when the ratio @xmath59 becomes significantly larger than 1 , the system exhibits a domain of oscillations connected to the bistable region ; the diagram has a typical cross - shaped topology @xcite . in a narrow band within the bistability region ,
the ` acid ' state can also become oscillatory .
this is analog to the experimental observations and to the results produced by the extended model .
moreover , an excitability region develops along the stability limit of the ft state . actually , there is no discontinuity between a stable ft state and an excitable pulse with an infinite recovery tail within the f state .
one must be aware that the recovery is not controlled by a slow kinetic process imbedded in the reaction mechanism as for the belousov - zhabotinskii reaction , but by the diffusion process that rules the feed from the cstr boundary .
thus , the excitability properties do not only depend on intrinsic quantities , i.e. the reaction rates and diffusion coefficients , but also on a geometrical characteristic of the reactor , namely the depth of the gel .
nevertheless , although the simple model predicts that differential diffusion can induce oscillations of the ft state in the spatial bistability region and excitability of the f state , the global agreement with experimental data , as well as with the extended model , is not very good . except for some analogies in the left upper part of fig .
[ diagvl ] , the shape of the diagrams is different .
a first reason is that the regions located on the right side of the bistable and the ` basic ' domains , including the oscillation pocket , are not physically reachable .
the corresponding values of @xmath58 could not be maintained in the cstr with realistic residence times .
the cstr would switch into a thermodynamic state . by the way , with this model
, it is difficult to speak of a true ft state since the concentration profiles do not exhibit any front .
this is why we prefer to use the terms ` basic ' or ` acid ' rather than f and ft when speaking of the simple model .
the second major reason is that oh@xmath44 and the autocatalytic species @xmath108 are related by a strong constraint independent of the c - t reaction itself , namely the fast equilibrium of water dissociation .
this leads to buffering processes and excludes that @xmath58 takes values close to zero , as permitted by the simple model .
thus this toy model , which allows for extensive numerical calculations , is an excellent tool to demonstrate the mechanisms of long range activation phenomena , but it should not be used for a precise analysis of the properties of the real system . in particular , it would be illusive to try to connect quantitatively the model parameters with realistic quantities . in the extended model , where the cstr dynamics is included , a direct comparison with experiments is possible .
spatial bistability limits are in good agreement .
there is nevertheless a point that deserves special attention .
species @xmath111 and @xmath123 are the natural substrates , but since the system is fed with a basic solution , @xmath124 must be considered as another substrate in spite of its special relation with @xmath108 .
since the diffusion coefficients of these substrates are not equal , there is an ambiguity in the choice of the species to be considered in the basic theory that leads to eq .
( [ valdelta ] ) . fig .
[ numprof ] clearly shows that the relevant species is @xmath124 .
its concentration drops linearly down to very low values at the exact position of the front , whereas the other substrates do not exhibit a linear drop all the way down and their concentration is far from minimal at the position of the front .
a natural explanation is that , although @xmath124 diffuses fast , @xmath108 reacts much faster with @xmath124 than with the other substrates .
this defines the smallest @xmath122 value and controls the overall process .
however , the tail of the concentration profiles of these other subtrates plays a role in the anticipation of the ft@xmath125f transition that is actually found in the computations . like in the simple model , oscillations induced by the long range activation are observed .
if one decreases @xmath81 within the bistability domain , the ft state becomes oscillatory when its stability limit is approached .
the diagram is in good quantitative agreement with experimental observations .
since oscillations are restricted to a narrow region of control parameters , the unavoidable small inhomogeneities along the annulus , make impossible to perform a fine analysis of this front dynamics . in the model , the complex oscillations and the period doubling sequence observed in the 1-d simulations
can be understood using the following arguments .
actually , coupled oscillators may present the same complex dynamical behavior .
close to the onset of the oscillatory regime , the motion of the front exhibits simple periodicity .
the amplitude of the oscillations @xmath126 is small and the front remains located close to the cstr boundary .
but when @xmath81 decreases , approaching the limits of existence of the ft state , this amplitude becomes larger and larger .
when @xmath127 is of the order of @xmath1 , the impermeable wall acts as a reflector for the diffusive concentration fields .
then , the front dynamics couples with itself , with delay and partial damping . this dynamics of coupled oscillators explains the observed dynamical complexity . in spatial bistability , it is possible to prepare two regions of the reactors in the two different states f and ft side by side and to study the dynamics of the f / ft interface . in the standard case , the direction of propagation of this interface depends on the control parameters and changes within the bistability domain according to the relative stability of the two states f and ft as was experimentally shown with the cdi reaction @xcite . in the c - t reaction , this situation is radically different .
because the fast diffusion of the autocatalytic species excites the medium ahead of the interface , the ft state propagates always into the f state , i.e. within the whole domain of spatial bistability .
moreover , this behavior can persist beyond the stability limit of state ft .
the f / ft interface still propagates into state f , but , since the ft state is now unstable , the system ultimately returns to the f state .
this gives rise to a recovery tail that follows the interface .
if the ratio of the diffusion of the activator to the diffusion of the main substrates is significantly larger than 1 ( here @xmath128 ) , the propagation of the ft state balances the return to the f state at the level of the interface . thus , the process goes on and a pulse can propagate without damping in time . if @xmath59 is smaller , or if the system is too far from the stability limit of the ft state , the return to the stable state dominates and the pulse dies .
this explains the domain of excitability observed in the experiments and accounted for by the models . despite the good agreement of the extended model results with experimental data
, this model still presents some serious limitations .
a major drawback of the extended model is the use of effective diffusion coefficients which neglects the ionic interactions .
the main effect of such interactions is more likely to give different effective diffusion coefficients in the flow state and in the thermodynamic state . as a result
, this should change the position of the front and the exact limits of the oscillatory and excitability domains , but we believe that our main conclusions should hold . from an experimental point on view , a severe limitation is the control of parameter uniformity along the annular gel / cstr boundary which makes the front oscillations to appear in the form of travelling waves .
finally , it has been recently found experimentally that the kinetic mechanisms summarized by eq .
( [ kin ] ) breaks down when the concentration of chlorite or tetrathionate are significantly increased @xcite . beyond these first experimental observations
, one can expect to control the front motion by a continuous change of the geometrical control parameter @xmath1 .
experiments along these lines are presently in progress both numerically and experimentally . | we show both experimentally and numerically that the time scales separation introduced by long range activation can induce oscillations and excitability in nonequilibrium reaction - diffusion systems that would otherwise only exhibit bistability .
namely , we show that in the chlorite - tetrathionate reaction , where the autocalytic species h@xmath0 diffuses faster than the substrates , the spatial bistability domain in the nonequilibrium phase diagram is extended with oscillatory and excitability domains .
a simple model and a more realistic model qualitatively account for the observed dynamical behavior .
the latter model provides quantitative agreement with the experiments .
dynamics of spatial and spatio - temporal reaction - diffusion patterns in chemical systems kept far from equilibrium by a permanent feed of fresh reactants is an active domain of nonlinear science @xcite .
these patterns include travelling waves in oscillating or excitable media , turing patterns , self - replicating spots , labyrinthine patterns and dynamically structured fronts .
they result from the destabilisation of a trivial state of high symmetry by a positive feedback ( ` activation ' ) , caused by autocatalysis or substrate inhibition , competing with a negative feedback on appropriate time scales .
although some patterns are transiently observed in some systems , these dissipative structures can only be sustained in permanently fed open reactors . among these
, the one side fed reactor ( osfr ) made of a thin slab of a chemically inert gel fed on one face and described further on in detail has become popular .
recently , a new type of dynamical behavior , namely spatial bistability , has been evidenced in this type of reactor @xcite . up to now , two classes of spatial patterns have effectively been observed . in the first class
, the primary source of the patterns is the dynamical properties of the reaction in homogeneous conditions .
reactions which oscillate or are excitable in homogeneous conditions , i.e. in well stirred reactors , give rise to travelling waves in unstirred media .
the role of diffusion is to convert the temporal dynamics into spatio - temporal dynamics @xcite .
the phenomenon may occur whatever are the relative values of the diffusion coefficients of the different species . in the second class ,
the source of patterns is the so - called ` long range inhibition ' .
an activatory species , i.e. a species which controls the positive feedback , like an autocatalytic species , diffuses much slower than the inhibitory species .
this favors the growth of the instability at localized positions by creating two different time scales for spatial interactions @xcite .
the prototypes of such patterns are the stationary turing patterns @xcite , but long range inhibition is also involved in other phenomena like self replicating spots @xcite , labyrianthine patterns @xcite or morphological instabilities of fronts @xcite . since all small molecules in solution have almost the same diffusion coefficients , it was necessary to slow down the effective diffusion of the activatory species by complexation with immobile sites to meet these conditions experimentally @xcite . actually , ` long range activation ' where an activating species diffuses faster than an inhibitory species defines a third class of systems with several diffusive time scales .
like with long range inhibition , one expects this differential diffusion be responsible for new instabilities that are not present in the sole chemical kinetics .
however , rather than inducing the growth of the perturbations more or less locally , the fast diffusing activator tends to excite perturbations at large distance from the initial inhomogeneities and to transform a stationary state into an excitable or an oscillatory one . in a recent study of spatial bistability in the chlorite - tetrathionate reaction , which exhibits long range activation ,
we have reported a preliminary observation of such a behavior @xcite .
here , we report more extensive experimental data and detailed numerical studies of these phenomena . in a first part
, we summarize former basic results on spatial bistability in osfrs , and present the reaction , the models and the experimental systems . in a second part ,
we report the results obtained with a simplified model to illustrate the general ideas . in a third part
, we report both experimental results and numerical simulations with a more complete model .
the last part is devoted to the discussion of the results . |
the main point to prove the result stated in the abstract is this : if we are dealing with a polynomial @xmath17 that is a polynomial in @xmath21 s and @xmath1 s that are @xmath4-free and we want to compute its norm , we observe that if @xmath17 is of ( joint ) degree at most @xmath22 then @xmath23 will be of degree at most @xmath24 .
the ricard - xu non - commutative khintchine inequality ( @xcite ) gives @xmath25 where @xmath26 is a certain expression ( actually a norm depending on @xmath27 ) that we will need to analyse below .
+ fix @xmath28 .
the last inequality gives us that if @xmath29 is fixed but chosen large enough so that @xmath30 then we have @xmath31^{1/2m}\le \|p\|\le ( 1+{\varepsilon})[kh((p^*p)^m)]^{1/2m}.\ ] ] thus to show the strong convergence of @xmath32 to @xmath33 it suffices to show that for @xmath27 fixed and @xmath23 we have @xmath34^{1/2 m } \to [ kh(q({x^{}},y)]^{1/2 m } , \ ] ] or merely @xmath35 but now a closer look at @xmath36 in [ sec3 ] will show that this holds .
it will be convenient to use ultraproducts , but we only need very basic and elementary facts that are recalled below . [ [ ul ] ] let @xmath37 be a non trivial ultrafilter on @xmath38 . given a sequence @xmath39 of banach spaces , we will usually denote by @xmath40 their ultraproduct ( see e.g. @xcite for more background information ) .
the elements @xmath41 are equivalence classes of bounded sequences @xmath42 with @xmath43 for all @xmath44 . by definition ,
two such sequences @xmath42 , @xmath45 are equivalent if @xmath46 . we will sometimes write @xmath47_\u$ ] to denote that @xmath42 is a representative of @xmath48 . whenever this holds we have @xmath49 .
[ [ wo ] ] as is well known , when all the spaces in @xmath39 are hilbert spaces @xmath40 is also a hilbert space .
+ it may be worthwhile to remind the reader that if @xmath50 is another family of hilbert spaces , we have a canonical isometric embedding of @xmath51 into the ultraproduct of the family @xmath52 .
of course this extends to an arbitrary finite number of factors .
moreover , if @xmath53 , then this embedding is an isomorphism .
[ [ wo2 ] ] let @xmath54 be sequences of banach spaces ( resp .
unital @xmath10-algebras ) .
let @xmath55 be a bounded sequence of linear mappings ( resp .
unital @xmath4-homomorphisms ) then the mapping @xmath56 defined whenever @xmath47_\u$ ] by @xmath57_\u$ ] is bounded ( resp .
a unital @xmath4-homomorphism ) with @xmath58 .
+ given two banach spaces @xmath59 , we will view a finite sum @xmath60 as an operator from @xmath61 to @xmath21 , and we denote by @xmath62 the associated operator norm .
+ consider now a sequence @xmath63 in the algebraic tensor product with uniformly bounded rank @xmath64 .
let us assume more precisely that there are , for each @xmath65 , bounded sequences @xmath66 and @xmath67 such that @xmath68 let @xmath69_\u$ ] and @xmath70_\u$ ] , and let @xmath71 then @xmath72 incidentally , this says that the injective tensor product @xmath73 embeds canonically isometrically into the ultraproduct of the sequence @xmath74 .
[ [ u ] ] let @xmath75 be a bounded sequence .
let @xmath76 be the associated operator .
then it is easy to see that @xmath77 more generally , let @xmath78 be a fixed integer and @xmath79 a hilbert space .
we denote by @xmath80 the space of @xmath81 matrices with entries in @xmath82 with the usual norm .
+ let @xmath83 be a bounded sequence .
then clearly @xmath84\|_{m_n(b(h^{\u } ) ) } = \lim\nolimits_\u \|s^{(k ) } \|_{m_n(b(h^{(k)}))}.\ ] ] [ [ b ] ] let @xmath85 be a unital @xmath4-algebra . by a state on @xmath85 we mean a linear functional such that @xmath86 and @xmath87 for all @xmath88 .
given this , the classical gns construction produces a hilbert space denoted by @xmath89 and a @xmath4-homomorphism @xmath90 equipped with a distinguished cyclic unit vector @xmath91 , such that @xmath92 for any @xmath93 .
let @xmath94 .
then @xmath95 is a unital @xmath10-algebra .
let @xmath96 for any @xmath97 .
then @xmath98 is a state on @xmath95 , @xmath99 and the representation @xmath100 can be identified with the result of the gns construction applied to @xmath101 . [
[ e ] ] given a sequence of states @xmath102 on a unital @xmath4-algebra @xmath103 , let @xmath104 be the associated gns construction and let @xmath105 be the associated @xmath10-algebra .
let @xmath106 be the representation defined for any @xmath107_\u\in h^\u$ ] and @xmath108 by @xmath109_\u)= [ \pi ^{(k)}(b)(z^{(k)})]_\u.\ ] ] obviously , @xmath110 let @xmath111 relative to _ pointwise _ convergence on @xmath103 , let @xmath112 be the associated gns representation and let @xmath113 .
let also @xmath114_\u$ ] .
then @xmath115 therefore , the correspondence @xmath116 extends to an isometric isomorphism from @xmath89 onto the subspace @xmath117 that is the closure of @xmath118 .
more precisely , the restriction of @xmath119 to @xmath120 , i.e. @xmath121 , is unitarily equivalent to the representation @xmath122 . +
now let us assume moreover that @xmath111 _
strongly_. this means ( see below ) that @xmath123 for any @xmath108 .
then the mapping @xmath124 defines an isometric embedding @xmath125
we now turn to a more formal description of our main result .
+ a more abstract ( but equivalent ) version of the statement in the abstract can be given in terms of convergence of states .
we use the notation in [ b ] .
let @xmath126 ( resp .
@xmath102 , @xmath127 ) be states on a unital @xmath4-algebra @xmath85 with associated gns hilbert spaces denoted by @xmath89 ( resp .
@xmath128 ) .
let @xmath129 ( resp .
@xmath130 ) be the associated gns representations of @xmath85 on these hilbert spaces .
we say that @xmath102 tends to @xmath126 strongly and we write @xmath131 if @xmath102 tends to @xmath132 pointwise on @xmath85 and moreover if @xmath133 for any @xmath134 in @xmath85 . in @xcite
the notion of free product of a family of states is defined .
it can be described as follows .
consider a family of states @xmath135 with gns hilbert space @xmath136 , gns representation @xmath137 and distinguished unit vector @xmath138 .
let @xmath139 be the associated @xmath10-algebra .
we denote by @xmath140 the inclusion map .
let @xmath141 be the ( algebraic ) free product of unital @xmath4-algebras .
following voiculescu ( see @xcite ) one defines a hilbert space free product @xmath142 and a representation @xmath122 of @xmath103 acting on @xmath143 .
let @xmath144 ( resp .
@xmath98 ) be the vector state on @xmath145 ( resp .
@xmath146 ) associated to @xmath138 ( resp .
@xmath147 ) . the unital @xmath10-subalgebra @xmath148 , equipped with @xmath98 , is called the reduced free product of @xmath149 we will denote by @xmath150 the vector state on @xmath103 defined by @xmath151 .
we call it the free product of the states @xmath135
. then we can reformulate the main result like this : [ t3 ] let @xmath152 ( @xmath153 ) be a family of unital @xmath4-algebras .
let @xmath154 ( @xmath155 ) be a sequence of families of states , each @xmath156 being a state on @xmath152 .
assume that we have states @xmath157 on @xmath152 such that , for each @xmath153 , when @xmath158 we have @xmath159 then @xmath160 [ [ le0 ] ]
let @xmath145 be associated to @xmath161 as above .
we view each @xmath145 as a subalgebra of @xmath162 . by a monomial of length
@xmath22 we mean a product of the form @xmath163 with @xmath164 such that @xmath165 and such that @xmath166 . by a homogeneous element of length @xmath22 in @xmath162 we mean a finite sum of monomials of length @xmath22 .
an element is called of length @xmath167 if it is a sum of homogeneous elements each of length @xmath167 .
note that the elements of finite length are dense in @xmath95 .
let us denote by @xmath168 ( resp .
@xmath169 ) the space of homogeneous elements of length @xmath22 ( resp .
@xmath170 ) in the preceding sense .
we also set @xmath171 and denote by @xmath172 the closure of @xmath168 in @xmath95 .
then the ricard - xu inequality we will use is this ( note that the assumption in @xcite that all the gns constructions are faithful is here automatic ) : there are constants @xmath173 and @xmath174 such that @xmath175 where we set @xmath176 and where @xmath177 are defined as follows : consider an element @xmath178 , of the form @xmath179 with @xmath180 monomial of degree @xmath22 as described above . let us denote by @xmath147 the distinguished unit vector in @xmath181 .
we define when @xmath182 @xmath183 and also @xmath184 and @xmath185 we will identify an element @xmath186 with the linear map @xmath187 defined by @xmath188 . in this way
we will view @xmath189 as an element of @xmath190 , and we denote by @xmath191 its norm .
assuming @xmath192 for simplicity .
let us denote by @xmath193 ( @xmath153 ) the canonical embedding .
assuming that @xmath194 for @xmath195 , we set + @xmath196 \in b(h^*,h ) \otimes
[ a_1 \oplus a_2 ] .\ ] ] we then denote by @xmath197 its norm in the minimal tensor product @xmath198 $ ] .
equivalently this is the maximum of two norms , one in @xmath199 ( induced by @xmath200 ) and one in @xmath201 ( induced by @xmath202 ) .
+ [ [ le00 ] ] by @xcite there is a ( completely ) bounded projection @xmath203 with @xmath204 , defined so that if @xmath205 with @xmath206 for all @xmath207 , then @xmath208 .
obviously this implies @xmath209 it will convenient for us to extend the above definition of @xmath210 as follows : for any @xmath211 and any @xmath212 we define @xmath213 combining with we now obtain that there are constants @xmath214 and @xmath215 such that @xmath216 [ [ cor ] ] for any @xmath153 we set @xmath217 .
the elements of @xmath218 will be sometimes called centered " ( with respect to @xmath126 ) .
let @xmath219 . by elementary ( free ) algebra , one can show that the algebraic free product @xmath103 is linearly isomorphic to the direct sum @xmath220 where the subspaces @xmath221 ( @xmath222 , @xmath223 ) are formed of all products of the form @xmath224 recall that , for each @xmath44 , we are given a state @xmath225 on @xmath152 .
we will denote by @xmath226 the linear mapping that transforms centering with respect to @xmath126 into centering with respect to @xmath225 .
more precisely , @xmath227 is defined by @xmath228 and @xmath229 .
using the above direct sum decomposition of @xmath103 , and viewing @xmath230 we can extend the mappings @xmath227 to a single mapping on @xmath103 .
more precisely , using the freeness of the product , there is a unique linear map @xmath231 that coincides with @xmath227 on @xmath152 for each @xmath153 and is such that for any element of the form we have @xmath232 we note that if we equip @xmath103 with the maximal @xmath10-norm then we clearly have @xmath233 let us denote by @xmath234 the linear projection ( relative to ) from @xmath103 to the subspace @xmath235 defined by @xmath236 fix @xmath44 .
again let @xmath237 and @xmath238 .
+ suppose now that we replace @xmath126 by @xmath225 so that we have a direct sum decomposition as above but now associated to @xmath225 .
this leads to subspaces @xmath239 defined exactly like @xmath240 but with respect to @xmath225 .
let @xmath241 denote the linear projection from @xmath103 to the subspace @xmath239 in the said direct sum decomposition relative to @xmath225 .
it is easy to check that we have for any @xmath44 @xmath242 [ [ f ] ] returning to the situation of theorem [ t3 ] , let @xmath243 with distinguished vector @xmath244 , gns representation @xmath245 with @xmath246 .
we define @xmath247 , @xmath248 and let @xmath249 be the corresponding representation .
let @xmath250 ( resp .
@xmath251 ) denote the ultraproduct of @xmath252 ( resp .
@xmath253 ) . +
we will use [ e ] when @xmath254 and @xmath255 .
let @xmath256 with @xmath113 .
we denote by @xmath257 the gns representation relative to @xmath126 .
we have natural identifications @xmath258 in this case , the correspondence @xmath259_\u$ ] defines a unitary operator from @xmath79 to @xmath120 .
we will denote by @xmath260 the corresponding isometry .
+ assume now that @xmath261 .
then ( see [ e ] ) we have an isometric embedding @xmath262 such that @xmath263 for any @xmath264 .
[ [ le ] ] consider now a monomial of length @xmath22 in @xmath265 as defined in [ le0 ] above , of the form @xmath266 with @xmath164 such that @xmath267 and such that @xmath166 .
we can choose @xmath268 such that @xmath269 .
we may also view @xmath270 as an element of @xmath141 , then with this abuse of notation we may write also @xmath271 .
+ let @xmath272 and @xmath273 . then by @xmath274 .
+ it will be useful to record here also that , a fortiori , for any @xmath275 we have by @xmath276_\u\quad{\rm and}\quad v(x_{r+1}\cdots x_d\xi)=[y ^{(k)}_{r+1}\cdots y ^{(k)}_d\xi^{(k)}]_\u.\ ] ] the pointwise convergence of @xmath277 to @xmath278 is obvious by definition of the free product of states . to show the strong convergence it suffices to show that @xmath279 for any @xmath108 and any non trivial ultrafilter @xmath37 on @xmath38 .
recall @xmath141 ( algebraic free product ) .
note that @xmath280 .
the main point is that , by the ricard - xu inequality , there are constants @xmath214 and @xmath215 such that @xmath281 if we accept this result , the proof is immediate : we just note that @xmath282 is of degree at most @xmath24 , therefore @xmath283 and @xmath284 .
so we find @xmath285 and letting @xmath286 yields the equality @xmath287 .
we now turn to the proof of . by the ricard - xu inequality
, we have @xmath288 and @xmath289 thus to conclude , it suffices to show that @xmath290 .
+ we claim that it suffices to prove this for @xmath291 , i.e. such that @xmath292 for some @xmath22 .
+ let @xmath293 be the analogue of @xmath294 with respect to @xmath295 . by the very definition of @xmath210 in it suffices to show @xmath296 for any @xmath22 .
at this point it is worthwhile to observe that @xmath297 and @xmath298 .
thus it suffices to show equivalently that for any @xmath22 we have @xmath299 note that by , since @xmath241 is idempotent , we have @xmath300 and by this implies @xmath301 .
then the equivalence , applied to each @xmath302 , implies @xmath303 , and hence @xmath304 therefore , to prove we may assume that @xmath305 for some @xmath22 , proving our claim . to conclude it suffices to show , when @xmath305 , that we have for any @xmath306 : @xmath307 in order to verify this , let @xmath308 , and let @xmath309 .
we may assume that @xmath310 is a finite sum of the form @xmath311 with @xmath312 , and @xmath313 with @xmath314 and @xmath315 .
+ then @xmath316 with @xmath317 , with @xmath318 , and we have @xmath319 with @xmath320 with @xmath321 .
we remind the reader that @xmath147 denotes the distinguished unit vector in @xmath322 , and + @xmath323 is viewed as an element of @xmath190 . moreover , in the case @xmath192 , @xmath324 $ ] . +
recall also that @xmath325 denotes the hilbert space ultraproduct of the free products defined by @xmath326 .
we will compare @xmath189 with @xmath327 we can naturally associate to the sequence @xmath328 an element @xmath329 .
moreover , since the sum over @xmath330 is a finite sum , the ranks of the tensors @xmath331 are _ uniformly bounded _ ,
and hence it is easy to check ( see [ wo2 ] ) that @xmath332 consider now ( see [ f ] ) the isometry @xmath333 defined by @xmath334_\u$ ] .
we have then , by @xmath335 from which @xmath336 follows . similarly , assuming @xmath192 for simplicity , we will compare @xmath337 with @xmath338 .\ ] ] now again , since @xmath330 runs over a fixed finite set , the sequence @xmath339 defines an element @xmath340 in @xmath341,$ ] where @xmath342 is viewed as embedded in @xmath343 and the norm is the minimal norm ( see [ le0 ] ) .
+ using [ u ] and the uniform bound on the ranks of the tensors involved , it follows that ( see [ u ] ) @xmath344 we now again use [ f ] .
we have then , by @xmath345 ( s _ r(x))\ ] ] from which @xmath346 follows
. thus we obtain by and that + @xmath347 and @xmath348 , and this concludes the proof of .
we now turn to the situation considered in the abstract . given a family @xmath349 in a @xmath10-probability space @xmath350 ,
let @xmath85 be the unital @xmath4-algebra of all polynomials in the non - commutative variables @xmath351 .
we will say that @xmath350 is _ adapted _ to @xmath21 if @xmath85 is dense in @xmath95 and the gns - representation associated to @xmath126 ( or equivalently to the restriction of @xmath126 to @xmath85 ) is isometric ( it suffices for this that @xmath126 be faithful on @xmath95 ) .
let @xmath15 , @xmath16 and @xmath352 , @xmath353 be sequences of families of non - commutative random variables in some adapted @xmath10-probability spaces .
assume that @xmath15 tends strongly to @xmath16 and that @xmath352 tends strongly to @xmath353 when @xmath354 .
assume moreover that for each @xmath44 @xmath15 and @xmath352 are @xmath4-free and also ( a fortiori ) that @xmath16 and @xmath353 are @xmath4-free .
then the joint family @xmath355 tends strongly to @xmath356 .
the assumption that @xmath15 tends strongly to @xmath16 means that for any polynomial @xmath17 in the non - commutative variables @xmath18 we have @xmath357 ( this is the convergence of moments ) and moreover @xmath20 .
we also assume this with @xmath1 in place of @xmath21 .
let @xmath358 ( resp .
@xmath359 ) be the unital @xmath4-algebra generated by non - commutative variables @xmath351 ( resp .
@xmath360 ) .
if @xmath16 is given as realized inside a @xmath10-algebra equipped with a state @xmath361 , we define the associated state @xmath362 on @xmath358 by @xmath363 and similarly for @xmath364 on @xmath359 . repeating this for each @xmath44 , this leads to states @xmath156 on @xmath152 for @xmath365 .
we may then identify the free product @xmath366 with the unital @xmath4-algebra generated by non - commutative variables @xmath367 .
the corollary then appears as a particular case of the preceding theorem .
[ [ rem1 ] ] recall @xmath368 . going back to the definition of the free product , a moment of thought ( recall [ wo ] ) shows that we have a canonical isometric embedding @xmath369 that respects the distinguished vectors .
+ assuming @xmath192 , the mapping @xmath370 can be described like this : first we have @xmath371 , then whenever we consider an element @xmath372 in @xmath373 ( resp .
@xmath374 ) we can choose representatives @xmath375 of @xmath372 with @xmath376 in @xmath377 ( resp .
@xmath378 ) , so that given an element @xmath379 of degree @xmath22 in @xmath380 , with alternating factors in @xmath377 and @xmath378 , we then define @xmath381 as the element of @xmath382 admitting as representative the sequence @xmath42 with @xmath383 .
with the notation of the corollary , whenever @xmath384 on @xmath11 converges in moments to @xmath385 on @xmath350 , it is well known that for any non - trivial ultrafilter @xmath37 and any polynomial @xmath17 , we have @xmath386 indeed , for any @xmath28 there are polynomials @xmath387 with unit norm in @xmath89 so that @xmath388 from which follows .
the converse inequality is the essence of strong convergence . with the same notation as [ rem2 ] , if the variables @xmath389 all commute ( i.e. form a family of commuting normal operators ) then the ( commutative ) @xmath10-algebra @xmath95 to which @xmath21 is adapted is isometric to the @xmath10-algebra @xmath390 of all continuous functions on a compact set @xmath391 .
similarly , if all @xmath392 commute , the @xmath10-algebra @xmath295 to which @xmath393 is adapted is isometric to the ( commutative ) @xmath10-algebra @xmath394 for some subset @xmath395 . when the family @xmath21 ( resp .
@xmath393 ) is reduced to a single normal operator , we can adjust so that this operator corresponds to the function @xmath396 on @xmath397 ( resp .
@xmath398 ) . in that case , @xmath397 ( resp .
@xmath398 ) is the spectrum of the operator @xmath21 ( resp .
@xmath393 ) and it is an easy exercise to check that @xmath399 iff it converges in moments and moreover @xmath400 wigner s classical theorem about the convergence of the eigenvalues of gaussian random matrices was strengthened in @xcite as follows : let @xmath401 be a random @xmath402-matrix the entries of which are independent complex gaussian @xmath403 , then for almost all @xmath404 , @xmath401 tends strongly to a circular random variable .
a similar result is valid for the classical gaussian wigner ( hermitian ) matrices ( model for the so - called gue ) now with a semi - circular limit .
see @xcite for more general results .
+ _ acknowledgement .
_ i am very grateful to mikael de la salle , ric ricard and paul skoufranis for useful communications .
brown and n. ozawa , _
@xmath10-algebras and finite - dimensional approximations _ , graduate studies in mathematics , 88 , american mathematical society , providence , ri , 2008 . b. collins and c. male .
the strong asymptotic freeness of haar and deterministic matrices .
u. haagerup and s. thorbjrnsen . a new application of random matrices : @xmath405 is not a group . ,
162(2):711775 , 2005 . | we give a different proof of paul skoufranis s very recent result showing that the strong convergence of possibly non - commutative random variables @xmath0 is stable under reduced free product with a fixed non - commutative random variable @xmath1 . in fact we obtain a more general fact : assuming that the families @xmath2 and @xmath3 are @xmath4-free as well as their limits ( in moments ) @xmath5 and @xmath6 , the strong convergences @xmath0 and @xmath7 imply that of @xmath8 to @xmath9 .
phrased in more striking language : the reduced free product is continuous " with respect to strong convergence .
the analogue for weak convergence ( i.e. convergence of all moments ) is obvious . by a @xmath10-probability space , we mean a unital @xmath10-algebra equipped with a state .
+ let @xmath11 and @xmath12 be @xmath10-probability spaces .
let @xmath13 , @xmath14 be families of non - commutative random variables .
we say that @xmath15 tends strongly to @xmath16 if for any polynomial @xmath17 in the non - commutative variables @xmath18 we have @xmath19 ( this is the convergence of moments ) and moreover @xmath20 .
the notion of strong convergence that was introduced in @xcite , was inspired by haagerup and thorbjrnsen s paper @xcite .
more recent examples of strong convergence are obtained in @xcite .
the results below are motivated by work by camille male @xcite , who first considered the question of the stability of strong convergence , and by d. shlyakhtenko s proof ( see the appendix of @xcite ) that the reduced free product with the @xmath10-algebra generated by free creators on the fock space satisfies the desired stability property . very recently
, this was generalized by p. skoufranis @xcite to essentially all reduced free products . in this note
we give a different more direct proof based on an inequality due to .
ricard and q. xu that is a generalization to arbitrary reduced free products of results proved previously by voiculescu , haagerup and buchholz ( see @xcite ) for free products of groups .
our proof yields actually a stronger stability than the one appearing in @xcite , as described in the abstract , but p. skoufranis informed us that the original proof of @xcite also yields that improvement . |
the first photometric evidence for the presence of multiple stellar populations in globular clusters came from the discoveries of the discrete distributions of stars on the red giant - branch ( rgb ) and main - sequence ( ms ) of the most luminous globular cluster @xmath0 cen @xcite .
the fact that the distribution of rgb stars is not just showing a spread but is discrete is a compelling evidence for the heavy elements enrichment and the formation of successive metal - enhanced generations of stars in proto-@xmath0 cen .
the similarity of this feature to that of the m54 , suggested as a nuclear star cluster of the sagittarius dwarf galaxy @xcite , together with other peculiarities of @xmath0 cen , has led the community to conclude that @xmath0 cen was once part of a more massive dwarf galaxy that merged with the milky way , as the sagittarius dwarf galaxy is in the process of doing now @xcite . in recent years
, more and more evidence is reported for the presence of double or multiple stellar populations in other globular clusters , such as ngc 2808 @xcite , ngc 1851 @xcite , ngc 6388 @xcite , and m22 @xcite . for most of these peculiar globular clusters ,
however , the evidence for the discrete distribution of heavy elements as observed in the rgb of @xmath0 cen is hitherto lacking , although spreads in some lighter elements ( * ? ? ? * and references therein ) and helium @xcite are reported .
therefore , the case of @xmath0 cen , and now that of m22 @xcite , are still viewed as exceptional , and the presence of chemical inhomogeneity in other globular clusters is largely considered as due to the pollution mechanisms expected in normal globular clusters , such as the winds from the intermediate mass asymptotic giant branch ( agb ) stars or fast - rotating massive stars @xcite . the purpose of this letter is to report that one of these globular clusters , ngc 1851 , is showing a clear split in the rgb from our ctio 4 m @xmath1 photometry .
this is compared with the @xmath4-@xmath5 photometry @xcite to confirm that the two distinct rgb populations are different in the abundance of heavy elements .
stellar population models are then constructed to show that , when the metal - rich second generation stars are also enhanced in helium abundance , the split of the rgb discovered in @xmath2 color would not be detected in usual optical colors , such as @xmath3 used in the @xmath6 survey .
our observations were performed using the ctio 4 m blanco telescope during 2007 november 13@xmath716 and 2008 october 27@xmath731 .
the telescope was equipped with the mosaic ii ccd imager consisting of eight 2k @xmath8 4k site ccds providing a plate scale of 0.27 arcsec pixel@xmath9 and a field of view of 36 @xmath8 36 arcmin@xmath10 .
all of our science frames were obtained under photometric conditions . the total exposure times for @xmath1
were 3990 , 676 , and 573s , respectively , split into short , intermediate , and long exposures in each band .
ngc 1851 was placed on chip 6 , approximately 4.5 arcmin south and 6 arcmin east from the ccd center .
several standard fields @xcite were also observed during our observing runs .
the iraf mscred package was used for preprocessing including bias correction and flat fielding .
psf photometry was then carried out using daophot ii / allstar@xcite , and daogrow was used for aperture corrections @xcite .
figure 1 shows color - magnitude diagrams ( cmds ) of ngc 1851 in ( @xmath11 , @xmath2 ) and ( @xmath12 , @xmath3 ) planes .
stars within 3.6 arcsec and outside of 4.5 arcmin from the cluster center are excluded from the cmds to reduce blending effects and the field star contaminations , respectively . to examine the cmd features
more carefully , we adopted the `` separation index '' @xcite for selecting stars that are relatively less affected by adjacent starlights .
all the stars in our cmds lie well within chip 6 , and therefore our cmds are not subject to any uncertainty stemming from the possible chip to chip variations of the mosaic ccds .
open circles denote rr lyrae variables in our program field among those identified by @xcite , plotted at random phase of pulsation .
the most remarkable feature of figure 1 is the presence of two distinct rgbs in the ( @xmath11 , @xmath2 ) cmd .
the discrete distribution is clear from the sub giant - branch ( sgb ) to the tip of rgb where the mean separation on the rgb is @xmath130.27 mag in @xmath2 .
when measured at given i magnitude , this value is reduced to @xmath130.20 mag . for the bright rgb stars , some early hints for this feature
were noted by @xcite from the strmgren ( @xmath14 , @xmath15 ) cmd , and by @xcite from the @xmath4-@xmath5 photometry ( e.g. see their figure 1 ) .
note that @xcite only discovered a split in the sgb , and the split of rgb we discovered in this paper was not detected in their @xmath6 photometry employing f606w and f814w passbands .
similarly , the rgb split is not apparent in our ( @xmath12 , @xmath3 ) cmd ( figure 1b ) .
this is most likely because the @xmath11-band is more sensitive to metal abundance variation than other passbands , as more metal atomic and molecular lines are located in the @xmath11-band ( see section 3 ) . given the small foreground reddening value of e(@xmath16 ) @xmath17 0.02 @xcite toward ngc 1851 , it is very unlikely that the differential reddening has caused the double rgbs .
the color difference between the two rgbs in @xmath2 color , at given @xmath18 mag of the horizontal - branch ( hb ) level , is @xmath130.20 mag , which is about three times larger than the maximum color difference expected in the extreme situation where one group of stars are all reddened by e(@xmath16 ) @xmath17 0.02 , while the other group has e(@xmath16 ) @xmath17 0.00 . also , if the observed color difference in @xmath2 color is due to the differential reddening , it would result in the @xmath3 color difference of 0.10 mag , which would have been detected along the rgb in our ( @xmath12 , @xmath3 ) cmd .
similar spatial distributions of stars on the bluer and redder rgbs also indicate that the differential reddening , if any , is not likely the cause of the double rgbs .
we have shown that the rgb of ngc 1851 is split into two distinct subpopulations .
the fact that the distribution of rgb stars does not just show a spread but is discrete can naturally eliminate the possibilities such as ( 1 ) star formation from not well mixed inhomogenous interstellar matter , ( 2 ) differential reddening ( see also section 2 ) , and ( 3 ) photometric errors , as all of these would produce a @xmath19 in color rather than a discrete distribution .
consequently , the most plausible interpretation for the double rgbs is that the ngc 1851 underwent metal enrichment in its early stage of evolution and successively formed metal - enhanced second generation stars .
spectroscopic observations show star - to - star abundance variations of the lighter elements ( elements lighter than si , such as n , o , na , & al ) in ngc 1851 @xcite .
this is generally interpreted as a result of pollution from winds of intermediate - mass agb stars @xcite or fast - rotating massive stars @xcite .
based on the elemental abundances observed by @xcite and @xcite for eight bright rgb stars , we can estimate the differences in the lighter elements between the two rgb populations .
analysis of these data by ( * ? ? ?
* see their figure 3 ) indicates that , while n , na , and al are all enhanced in redder rgb ( @xmath20[n / fe ] @xmath21 0.47 , @xmath20[na / fe ] @xmath21 0.53 , @xmath20[al / fe ] @xmath21 0.20 ) , o is depleted ( @xmath20[o / fe ] @xmath21 @xmath70.45 ) , and no significant variation is shown in [ c / fe ] . in order to investigate the effect of these elemental variations in the ( @xmath2 ) color , in figure 2 , we have computed the differences in fluxes between two synthetic spectra using the spectral - synthesis program spectrum @xcite , one without and the other with these enhancements ( and depletion for o ) in lighter elements ( panels a & b ) . also shown in figure 2 are for the two additional cases where ( 1 ) cno and na are enhanced by 0.3 dex while heavier elements are fixed ( panels c & d ) , and ( 2 ) elements heavier than al are enhanced by 0.3 dex while lighter elements are fixed ( panels e & f ) .
these simulations demonstrate that small variations either in lighter or heavier elements could cause significant change in ( @xmath2 ) color .
the observed variations in the lighter elements alone , however , would cause relatively small line blanketing equivalent to @xmath20(@xmath2 ) @xmath22 0.079 in terms of color difference . despite uncertainties , taken at face value ,
this is only @xmath1341% of the observed color difference [ @xmath20(@xmath2 ) @xmath17 0.195 @xmath23 0.011 ] at given @xmath18 magnitude and suggests that other effects may also be responsible for the rgb split in ( @xmath2 ) color .
indeed , recent @xmath4-@xmath5 photometry @xcite shows that besides lighter elements variations , rgb stars of ngc 1851 also show bimodal distribution in ca , which can only be supplied by type ii supernovae ( sne ii ; * ? ? ?
* ) . in figure 3
, we show in our ( @xmath11 , @xmath2 ) cmd the `` ca - normal '' and `` ca - strong '' stars from @xcite .
the ca - strong stars lie well on the redder rgb sequence , whereas the ca - normal stars are on the bluer rgb . according to @xcite , the difference in ca abundance
is estimated to be @xmath20[ca / h ] @xmath21 0.15 dex , and other heavy elements , albeit small , are similarly enhanced in redder rgb population .
we conclude therefore that the redder rgb population is richer than the bluer rgb population not only in lighter elements ( n , na , & al ) but also in heavy elements ( such as ca , si , ti , & fe ) . in order to better understand the origin of the rgb color difference in ( @xmath2 ) , and to place stronger constraints on the chemical combinations of the two subpopulations , we have constructed stellar population models based on the updated version of the yonsei - yale ( @xmath24 ) isochrones ( yi et al . , in preparation ) and hb evolutionary tracks ( han et al . , in preparation ) .
readers are referred to @xcite and references therein for the details of our model construction .
figure 4 presents our synthetic cmds for ngc 1851 in ( @xmath11 , @xmath2 ) and ( @xmath12 , @xmath3 ) planes .
our models are constructed under two different assumptions regarding the chemical enrichment in ngc 1851 .
first , we assumed that the second generation population is more enhanced in metallicity , but not in helium ( figure 4a ; hereafter @xmath20z - only model ) .
note that , while @xcite and @xcite suggested the difference in only cno abundance , here we are assuming the difference in overall metallicity of 0.15 dex as discussed above .
second , in figure 4b , we then assumed that both metal and helium abundances are enhanced ( hereafter @xmath20z+@xmath20y model ) .
the @xmath11 flux is very sensitively affected by cn and nh bands , and therefore by n abundance . in order to reflect this effect in the ( @xmath11 , @xmath2 ) cmd of the second generation population ( redder rgb )
, we are also including the effects of the additional line absorptions from the enhancements in n and other lighter elements as discussed above , again by using spectrum .
more rigorous modeling should include the effects of these lighter elements enhancements in the construction of stellar evolutionary tracks , but we note that @xcite found the enhancement in n has only negligible effect in the hr diagram morphology at globular cluster ages .
comparison of the number ratio between the two subpopulations suggests that bluer rgb , brighter sgb , and redder hb are associated with one subpopulation , which we refer to as `` pop-1 '' , while the redder rgb , fainter sgb , bluer hb are associated with the other subpopulation , which we refer to as `` pop-2 '' .
pop-1 comprises about 75% of the total population , while pop-2 takes about 25% of the whole population . the @xmath20z - only model in figure 4a ( @xmath20z = 0.0004 and @xmath20age = 0.1 gyr ) matches well with the observed cmd from the ms through the rgb in ( @xmath11 , @xmath2 ) cmd . yet
, the model fails to reproduce the blue hb and the narrow rgb in ( @xmath12 , @xmath3 ) cmd ( figure 4a , inset ) .
this is because metal enhancement in pop-2 moves hb to the opposite direction , i.e. , makes hb morphology redder ( see for example * ? ? ?
at the same time , the increase in metallicity , albeit small , will create a gap in ( @xmath3 ) color between the two rgbs which is not shown in the observed cmd @xcite .
therefore , the @xmath20z - only model is in conflict with the observed cmds of ngc 1851 .
figure 4b shows that the @xmath20y+@xmath20z model ( @xmath20z @xmath17 0.0004 , @xmath20age @xmath17 0.1gyr , and @xmath20y @xmath17 0.05 ) is in good agreement with the observation from the ms to the hb .
note that , in our modeling , we are just employing the standard @xcite mass - loss law , and the same mass - loss parameter @xmath25 for both normal helium and enhanced helium populations .
input parameters adopted in our @xmath20y+@xmath20z model are listed in table 1 .
the enhanced helium in pop-2 easily overcomes the effect of metallicity on hb ( see * ? ? ?
* ) , shifting hb morphology toward blue and reproducing the observed hb bimodality . at the same time , the increase in helium abundance in pop-2 moves rgb slightly bluer and this effect cancels out the metallicity effect in ( @xmath12 , @xmath3 ) cmd , making apparently single and narrow rgb . also , the model ( figure 4b , inset ) is in better agreement with the observed sgb split@xcite .
the @xmath11-band is much more sensitive to metal line blanketing , and therefore the increased helium has relatively small effect on @xmath2 color of rgb .
one caveat to this helium enhanced scenario is that the blue hb in our model appears to be slightly brighter ( @xmath130.05 mag ) than the observation in @xmath11-band , although this could be due to the uncertainty in bolometric correction of the @xmath11-band .
the apparent helium enhancement in the second generation subpopulation in ngc 1851 is reminiscent of the cases of other globular clusters with multiple populations , including @xmath0 cen @xcite , ngc 2808 @xcite , ngc 6388 and ngc 6441 @xcite .
although the origin of this helium enhancement is currently not fully understood , general consensus is that it can be supplied either by sne ii @xcite or by the intermediate mass agb stars or fast - rotating massive stars @xcite . in the case of ngc 1851 ,
given the enhancements both in the lighter ( such as n ) and heavy ( such as ca ) elements in the second generation population ( pop-2 ) , all of the mechanisms seem to be responsible for the helium enhancement .
a possible scenario is that , soon after the formation of the first generation stars ( pop-1 ) , numerous sne ii explosions enriched both metal and helium of the leftover gas in the proto - ngc 1851 .
winds and ejecta from the intermediate mass agb stars may have added more helium and simultaneously enhance lighter elements .
the second generation stars ( pop-2 ) would have then formed from the gas enriched in overall metallicity , helium , and lighter elements .
we note that the present mass ( @xmath13@xmath26@xmath27 ; * ? ? ?
* ) of ngc 1851 is too small to retain the ejecta from numerous sn explosions @xcite .
therefore , as in @xmath0 cen @xcite , we suggest that ngc 1851 is most likely the remaining core of more massive primeval dwarf galaxies that merged and disrupted to form the proto - galaxy , as recently proposed by @xcite for the origin of globular clusters with the extended hb .
our result presented here is calling a new survey of globular clusters employing @xmath28 and/or calcium filters in order to detect small difference in metallicity expected in the yet to be identified globular clusters with multiple populations .
high resolution spectroscopy of stars in the two distinct groups are also needed to confirm the small difference in the abundance of heavy elements .
baumgardt , h. , kroupa , p. , & parmentier , g. 2008 , , 384 , 1231 bedin , l. r. , piotto , g. , anderson , j. , cassisi , s. , king , i. r. , momany , y. , & carraro , g. 2004 , , 605 , l125 bekki , k. , & freeman , k. 2003 , , 346 , 11 calamida , a. , et al .
2007 , , 670 , 400 caloi , v. , & dantona , f. 2007 , , 463 , 949 carretta , e. , bragaglia , a. , gratton , r. g. , & lucatello , s. 2008 , astro - ph 0811.3591 cassisi , s. , salaris , m. , pietrinferni , a. , piotto , g. , milone , a. p. , bedin , l. r. , & anderson , j. 2008 , , 672 , l115 da costa , g. s. , held , e. v. , saviane , i. , & gullieuszik , m. 2009 , astro - ph 0909.5265 dantona , f. , bellazzini , m. , caloi , v. , pecci , f. f. , galleti , s. , & rood , r. t. 2005 , , 631 , 868 decressin , t. , meynet , g. , charbonnel , c. , prantzos , n. , & ekstrm , s. 2007 , , 464 , 1029 dopita , m. a. , & smith , g. h. , 1986 , , 304 , 283 dotter , a. , chaboyer , b. , ferguson , j. w. , lee , h .- c . , worthey , g. , jevremovi , d. , & baron , e. 2007 , , 666 , 412 gray , r. o. , & corbally , c. j. , 1994 , , 107 , 742 green , e. m. , demarque , p. , & king , c. r. 1987 , the revised yale isochrones & luminosity functions ( new haven , ct : yale univ .
harris , w. e. 1996 , , 112 , 1487 hesser , j. e. , bell , r. a. , harris , g. l. h. , & cannon , r. d. 1982 , , 87,1470 landolt , a. u. 2007 , , 133 , 2502 landolt , a. u. 1992 , , 104 , 340 layden , a. , & sarajedini , a. 1997 , , 486 , l107 lee , j .- w . , lee , j. , kang , y .- w . , lee , y .- w .
, han , s .-
j . , rey , s .- c . , &
yong , d. 2009a , , 695 , l78 lee , j .- w . ,
kang , y .- w . ,
lee , j. , & lee , y .- w .
2009b , nature , in press lee , y .- w . , gim , h. b. , & casetti - dinescu , d. 2007 , , 661 , l49 lee , y .- w . ,
2005 , , 621 , l57 lee , y .- w . ,
m . , sohn , y .-
j . , rey , s .- c . ,
lee , h .- c . , & walker , a. r. 1999 , nature , 402 , 55 lee , y .- w . , demarque , p. , & zinn , r. 1994 , , 423 , 248 marino , a. f. , milone , a. p. , piotto , g. , villanova , s. , bedin , l. r. , bellini , a. , & renzini , a. 2009 , astro - ph 0905.4058 milone , a. p. , stetson , p. b. , piotto , g. , bedin , l. r. , anderson , j. , cassisi , s. , & salaris , m. 2009 , , 503 , 755 milone , a. p. , et al .
2008 , , 673 , 241 moretti , a. , et al .
2009 , , 493 , 539 norris , j. e. 2004 , , 612 , l25 piotto , g. , et al .
2007 , , 661 , l53 piotto , g. , et al .
2005 , , 621 , 777 pryor , c. , & meylan , g. 1993 , structure and dynamics of globular clusters ( asp conf .
50 ) , ed .
s. g. djorgovski & g. meylan ( san francisco , ca : asp ) , 357 reimers , d. 1977 , , 57 , 395 salaris , m. , cassisi , s. , & pietrinferni , a. 2008 , , 678 , l25 stetson , p. b. 1990 , pasp , 102 , 932 stetson , p. b. 1987 , pasp , 99 , 191 stetson , p. b. , bruntt , h. , & grundahl , f. 2003 , pasp , 115 , 413 timmes , f. x. , woosley , s. e. , & weaver , t. a. 1995 , , 98 , 617 ventura , p. , & dantona , f. 2009 , , 499 , 835 ventura , p. , & dantona , f. 2008 , , 385 , 2034 walker , a. r. 1998 , , 116 , 220 yong , d. , & grundahl , f. 2008 , , 672 , l29 yong , d. , grundahl , f. , dantona , f. , karakas , a. i. , lattanzio , j. c. , & norris , j. e. 2009 , , 695 , l62 yoon , s .- j . , joo , s .-
ree , c. h. , han , s .-
i . , kim , d .-
, & lee , y .- w . 2008 , , 677 , 1080 cccccc parameter & population 1 & population 2 + @xmath29 & 0.0012 & 0.0016 + @xmath30 & 0.2324 & 0.282 + @xmath31@xmath32/fe@xmath33 $ ] & 0.3 & 0.3 + age & 10.7 gyr & 10.6 gyr + @xmath25 & 0.59 & 0.59 + @xmath20@xmath34 & 0.2337 & 0.2218 + @xmath35 & 0.015 & 0.015 + population ratio & 0.7 & 0.3 + | there is a growing body of evidence for the presence of multiple stellar populations in some globular clusters , including ngc 1851 . for most of these peculiar globular clusters ,
however , the evidence for the multiple red giant - branches ( rgbs ) having different heavy elemental abundances as observed in @xmath0 centauri is hitherto lacking , although spreads in some lighter elements are reported .
it is therefore not clear whether they also share the suggested dwarf galaxy origin of @xmath0 cen or not .
here we show from the ctio 4 m @xmath1 photometry of the globular cluster ngc 1851 that its rgb is clearly split into two in the @xmath2 color .
the two distinct rgb populations are also clearly separated in the abundance of heavy elements as traced by calcium , suggesting that the type ii supernovae enrichment is also responsible , in addition to the pollutions of lighter elements by intermediate mass asymptotic giant branch stars or fast - rotating massive stars .
the rgb split , however , is not shown in the @xmath3 color , as indicated by previous observations .
our stellar population models show that this and the presence of bimodal horizontal - branch distribution in ngc 1851 can be naturally reproduced if the metal - rich second generation stars are also enhanced in helium . |
nearest neighbor searching has applications in knowledge discovery , data mining , pattern classification , machine learning , data compression , multimedia databases , and document retrieval . from the perspective of worst - case performance , an ideal solution for
the nearest neighbor search would be to preprocess the points in @xmath22 time , into a data structure requiring @xmath23 space so that any query can be answered in @xmath24 time . in @xmath25 ,
this is possible by sorting @xmath0 , and then using binary search to answer queries . in @xmath26 ,
this is possible by computing the voronoi diagram for @xmath0 and then using an algorithm for planar point location to find the cell containing the query point .
however , in higher dimensions , the worst - case complexity of the voronoi diagram grows as @xmath27 .
specifically , no known method achieves the simultaneous goals of roughly linear space and logarithmic query time in higher dimensional spaces .
this is the primary reason in resorting to approximation algorithms .
given a set @xmath0 of @xmath1 points , a point @xmath28 is an @xmath4-approximate nearest neighbor ( ann ) of a point @xmath29 , if the distance between @xmath16 and @xmath12 is at most @xmath4-times the euclidean distance between @xmath16 and its nearest neighbor in @xmath0 .
however , even in approximation schemes , usually the space , time , and query time complexities grow exponential in @xmath4 .
netanyahu et al .
, @xcite presents an algorithm to find an ann of a query point @xmath16 in time @xmath30 ( @xmath31 relies on @xmath2 and @xmath4 ) , with a data structure of size @xmath32 constructed in @xmath33 time .
their data structure , bbd tree , is based on the hierarchical decomposition of space : subdivide space into a collection of cells , each of which is either a d - dimensional rectangle or a set - theoretic difference of two rectangles , one enclosed within another .
the bbd - tree is based on a spatial decomposition that achieves both exponential cardinality decrease ( like kd - trees ) and the geometric size reduction ( like quadtrees ) as one descends the tree . after finding the leaf cell containing the query point @xmath16 in @xmath24 time by a simple descend through the tree
, one can find @xmath16 s approximate nearest neighbors : enumerate the leaf cells in increasing order of distance from @xmath16 until distance to the point @xmath12 associated with a cell exceeds @xmath34 where @xmath35 is the closest point to @xmath16 in the output so far .
the algorithm from arya et al .
, @xcite minimizes the number and complexity of cells by constructing an approximate voronoi diagram of @xmath0 during the preprocessing phase .
this yields a data structure of @xmath36 space that can answer @xmath4-ann queries in time @xmath37 for a space - query time trade - off parameter @xmath38 , where @xmath39 . the principal reduction in space arises from a form of deterministic sampling over bbd - tree .
the approach also creates cells more economically using well - separated pair decomposition ( wspd ) of the points by generating them along the bisector between well - separated pairs . in the case
@xmath40 , the additional @xmath41 factor in space complexity can be avoided to have a data structure that answers queries in @xmath42 time with @xmath43 space ( refer arya et al . ,
@xcite and chan @xcite ) .
har - peled @xcite and sabharwal et al .
, @xcite suggests an approach to reduce the query version of ann problem to point location in balls . in this paper
, we are interested in querying for @xmath4-anns but the definition of @xmath4-ann is bit varied from the literature .
to our knowledge , this seems to be the first paper to give an algorithm for the following variant definition of @xmath4-ann . for a query point @xmath16 ,
the output of our algorithm is a set @xmath5 wherein : @xmath8 , @xmath9 a point @xmath12 in @xmath44 such that @xmath11 and the nearest neighbor of @xmath12 is @xmath13 , and @xmath9 a @xmath14 such that @xmath15 is a nearest neighbor of @xmath16 . as shown in fig .
[ fig : def ] , let @xmath45 be the set @xmath0 . consider a query point @xmath16 that belongs to the voronoi region of @xmath46 .
suppose @xmath47 . the site @xmath48 is in @xmath7 as there exists a point @xmath49 such that @xmath50 and @xmath48 is the nearest neighbor of @xmath49 . similarly , the site @xmath46 is in @xmath7 becuase of @xmath51 and @xmath46 is the nearest neighbor of @xmath16 .
consistent with the definition of @xmath4-ann , not all sites that correspond to regions intersecting with the @xmath4-radius ball centered at @xmath16 are in @xmath7 ( for example , @xmath52 ) .
we first construct the voronoi diagram @xmath53 of @xmath0 .
we do this by space partitioning @xmath53 ( contained in a large bounding hyperbox ) with smaller hyperboxes and organizing hyperboxes into a box tree data structure , termed as _
main box tree_. we also cover specific regions of @xmath53 with hyperboxes whose bounding planes are not parallel to coordinate - planes and construct a set of corresponding box tree data structures , each of which is termed as an _ auxiliary box tree_. the advantage is that the query phase does not require data structures corresponding to @xmath53 .
it only requires main and auxiliary box tree data structures .
although our algorithm does not improve time or space bounds of previous algorithms , it seems to provide another direction ( detailed in the section [ sect : conclu ] ) for approaching query version of ann finding problem .
this is the main motivation in listing this result .
our algorithm first constructs the voronoi diagram @xmath53 corresponding to the input set @xmath0 of points in @xmath2-dimensional euclidean metric space . before deleting data structures corresponding to @xmath53
, we construct a set of box trees from @xmath53 .
we enclose a sufficiently large space of @xmath53 in a bounding hyperbox @xmath54 so that every site in @xmath0 is an ann to any point exterior to @xmath54 .
we spatially partition @xmath54 into hyperboxes of volume @xmath19 ( a real number that relies on @xmath4 ) , and with each such hyperbox @xmath55 we associate sites corresponding to voronoi cells with which @xmath55 intersects .
we organize these hyperboxes into a box tree , termed as the _ main box tree _ ( refer fig . [
fig : datastr ] ) .
whenever the hyperbox @xmath55 stored at any leaf of main box tree is associated with multiple sites , we cover the region enclosed by @xmath55 with multiple hyperboxes , termed as _
auxiliary hyperboxes_. every such hyperbox is obtained by rotating @xmath55 w.r.t .
its center by a distinct discrete angle . by partitioning the region enclosed by each of these auxiliary hyperboxes
, we construct a box tree , termed as an _ auxiliary box tree _ ( refer fig . [
fig : datastr ] ) . like in the main box tree ,
leaf node @xmath56 of every auxiliary box tree is associated with voronoi sites corresponding to voronoi cells that intersect with the hyperbox at @xmath56 .
each internal / leaf node of main / auxiliary box tree implicitly associated with the hyperbox that corresponds to union of hyperboxes that are associated with its children . also , every internal node @xmath57 has information necessary to locate @xmath16 in the hyperbox associated with a specific child of @xmath57 .
given a query point @xmath16 for which the user is interested in finding anns , we descend through the height of the main box tree in finding the leaf node @xmath56 whose hyperbox @xmath55 contains @xmath16 . if @xmath55 is associated with one and only one voronoi site , we have determined the exact neighbor of @xmath16 .
otherwise , similar to traversing the main box tree , we traverse each of the auxiliary box trees associated with leaf node @xmath56 so as to confine the cardinality of set @xmath7 output by the algorithm , as detailed below . essentially , during the query phase we rely on main and auxiliary box trees rather than on data structures that store voronoi structure .
the preprocessing phase starts by constructing a voronoi diagram .
the box tree data structure is a trivial generalization of quadtree data structure to higher dimensions .
it is a rooted tree in which every internal node has @xmath58 children .
the root node of the main box tree corresponds to a hyperbox that is the union of all the hyperboxes into which @xmath54 is partitioned .
the children of any internal node @xmath57 are obtained by subdividing hyperbox @xmath55 corresponding to @xmath57 with the hyperplanes parallel to coordinate planes , which bisect each dimension of @xmath55 . in other words ,
every internal node @xmath57 is partitioned into hyperboxes , and each of them is implicitly stored at a distinct child of @xmath57 .
note that since the box is symmetric the direction of rotation is not important .
each leaf node @xmath56 of the main box tree stores the sites corresponding to voronoi cells that intersect with the hyperbox at @xmath56 as satellite data .
a point in @xmath0 belongs to which half - space defined by a hyperplane is decided by trivial means .
this elementary operation assists in associating sites with the leaf nodes .
the recursive definition of the box tree immediately translates into a recursive algorithm : partition the hyperbox associated with a node @xmath57 into @xmath58 hyperboxes using hyperplanes parallel to coordinate - planes .
the recursion stops at a leaf node @xmath56 of the main box tree whenever the volume of hyperbox @xmath55 at @xmath56 is less than a real number @xmath19 ( which is defined in terms of @xmath4 ) .
consider the case in which the hyperbox @xmath55 at a leaf node @xmath56 is associated with multiple sites but the volume of @xmath55 is less than @xmath19 . in this case , we associate a linked list of auxiliary box trees to leaf node @xmath56 .
the root of each of these auxiliary box trees correspond to an auxiliary hyperbox which is constructed by rotating @xmath55 by a solid angle @xmath4 w.r.t .
its center .
however , since a query point that belongs to @xmath55 does not necessarily contain in any such auxiliary hyperbox , we double the size of @xmath55 before orienting it .
let @xmath59 be the hyperbox that is obtained by scaling @xmath55 by two w.r.t . its center ( refer fig . [
fig : algo ] ) .
the root of the first auxiliary box tree in the linked list is associated with a hyperbox @xmath60 , that is obtained by rotating @xmath59 w.r.t its center by @xmath4 solid angle .
the next auxiliary box tree in the linked list is obtained by rotating @xmath60 by @xmath4 etc .
, so that no two of the auxiliary hyperboxes corresponding to two distinct auxiliary box trees are oriented with the same angle . in other words ,
the length of the linked list associated with @xmath56 is @xmath61 .
each of the auxiliary box trees are defined similarly as in main box tree : partition the hyperbox associated with a node @xmath57 into @xmath58 hyperboxes using hyperplanes parallel to coordinate axes .
the recursion stops at a leaf node @xmath56 whenever the volume of a hyperbox corresponding to a node is less than @xmath19 .
given a query point @xmath16 in @xmath2-dimensional euclidean space , we intend to find its anns .
first , we find the leaf node in main box tree that contains @xmath16 .
this is accomplished by traversing one node at each level of the main box tree .
the satellite data associated with internal nodes assists in branching to a particular child of @xmath57 . the same could be accomplished in @xmath62 time using hashing i.e. , by maintaining a hash table rather than a box tree .
however , to maintain the consistency between primary and secondary level data structures , we organize primary level data structure as a box tree .
let @xmath16 belongs to a hyperbox @xmath55 associated with a leaf node @xmath56 of the main box tree .
if @xmath55 is associated with a single site @xmath13 , then it is immediate to conclude that the exact neighbor of @xmath16 is @xmath13 .
otherwise , we traverse through the linked list @xmath63 referred by node @xmath56 .
as mentioned earlier , each node of the linked list refers to an auxiliary box tree .
similar to the traversal of main box tree , we traverse each of the auxiliary box trees referred by @xmath63 , say @xmath64 .
let @xmath65 be the set that comprises of hyperboxes that contain @xmath16 .
also , let @xmath66 be the sets of sites respectively associated with hyperboxes in @xmath67 .
the algorithm returns the set @xmath68 as the set of anns to @xmath16 .
from the analysis mentioned in section [ sect : analysis ] , it is immediate that in both the preprocessing phase and in the query phase algorithm always terminates .
we have chosen the size of bounding box @xmath54 so that the algorithm is correct for all the query points exterior to @xmath54
. let @xmath16 belongs to a hyperbox @xmath55 that corresponds to a leaf node @xmath56 of the main box tree . if @xmath55 is associated with a single site , according to the definition of voronoi diagram , the algorithm outputs the nearest neighbor of @xmath16 .
otherwise , the proof of correctness follows .
we are ensuring that the hyperbox corresponding to the root of every auxiliary box tree contains @xmath16 by doubling the size of @xmath55 .
the intersection of all the auxiliary hyperboxes associated with any leaf node @xmath56 together yield a ( approximate ) hypersphere @xmath69 whose boundary consists of hyperboxes of size @xmath4 in @xmath70 .
given the volume of the byperbox is @xmath19 , the length of the main diagonal of a hyperbox is @xmath71 .
hence , the diameter of hyperball @xmath69 is less than or equal to @xmath72 .
let @xmath7 be the set of sites whose voronoi regions intersect with the hyperball @xmath69 .
suppose @xmath16 lies in @xmath69 . in this case
, though the output of the query algorithm does not have any site from the set @xmath73 , no auxiliary box tree eliminates any site in @xmath7 . by choosing the diameter of @xmath69 as @xmath4
, every site in @xmath7 is an ann of @xmath16 .
hence , we choose @xmath19 as @xmath20 .
consider the other case in which @xmath16 does not lie in @xmath69 . in this case ,
set intersection of sites associated to multiple auxiliary hyperboxes together define the anns of @xmath16 .
since the main diagonal lengths of every hyperbox that is implicitly stored at any leaf node is upper bounded by @xmath4 , the correctness is ensured again .
[ ] assuming uniform distribution of @xmath0 in @xmath44 euclidean metric space , the expected cardinality of set @xmath7 is @xmath74 . here , @xmath75 is the radius of the smallest hyperball that contains @xmath0 .
first note that the hyperball of radius @xmath75 has volume @xmath76 , where @xmath77 . due to uniform distribution of @xmath0 ,
the expected number of sites per unit volume are @xmath78 . since the diameter of the hyperball @xmath69 is @xmath4 , the number of sites in @xmath69 are @xmath79 .
[ depthmainboxtree ] the depth of the main box tree is at most @xmath80 , where @xmath21 is the volume of the initial bounding box and @xmath19 is the volume tolerance of the smallest possible box .
the box volume goes down by @xmath58 for every level , therefore the volume of the box at depth @xmath81 is @xmath82 .
we know that the volume @xmath19 is possible at one less than the maximum depth . therefore , the smallest possible box volume is , @xmath83 .
hence , for any hyperbox in the main box tree @xmath84 .
[ maxmb - nodes ] the maximum possible number of nodes in the main box tree are @xmath85 if we consider the complete box tree , the number of boxes in the box tree at the @xmath81th level are @xmath86 .
therefore , the total number of nodes in the main box tree are @xmath87(@xmath88 terms ) , summing over this geometric progression yields the result .
[ reordepth ] the maximum depth of any auxiliary box tree is @xmath62 .
we know that the volume of any hyperbox @xmath55 at a leaf node of main box tree is @xmath19 . since we scale @xmath55 by two , the volume of the resultant hyperbox at the root of auxiliary box tree is @xmath89 .
then from lemma [ depthmainboxtree ] , the depth of the auxiliary box tree is at most @xmath90 , which is @xmath62 .
[ reorspace ] the maximum number of nodes in any auxiliary box tree are @xmath91 .
this is immediate by substituting @xmath92 in the statment of lemma [ maxmb - nodes ] .
given the voronoi diagram of @xmath0 , the space complexity in constructing all box trees together is @xmath17 .
the main box tree is a complete tree and the number of leaf nodes are @xmath93 .
also , each leaf may have a list of length @xmath61 . from lemma [ reorspace ]
, the space complexity is as stated .
the preprocessing time in constructing all the box trees is @xmath94 .
since only constant time is spent in creating each node of every box tree , the time complexity is same as the space complexity except that we need to consider the following .
the cost in associating sites to main box tree leaves involve @xmath95 point location searches per each level of the main box tree .
the same is applicable to auxiliary box trees .
including the time complexity of constructing the voronoi diagram of @xmath0 yields the stated .
the query time complexity is @xmath96 . from lemma [ depthmainboxtree ] , the worst - case depth of the main box tree is @xmath97 .
the worst - case length of any list at a leaf node of the main box tree is @xmath61 . from lemma [ reordepth ] ,
the worst - case height of the auxiliary box tree is o(1 ) .
therefore , the worst - case query time is @xmath98 .
note that each occurrence of @xmath19 in the above analysis can be expressed in terms of @xmath4 , as @xmath99 .
arrows indicate the direction of expansion of hyperboxes into voronoi cell so that they together cover that cell .
this paper proposed an algorithm that builds data structures of size @xmath17 during the preprocessing phase which facilitate in querying anns of an input query point in @xmath100 time . here
@xmath19 equals to @xmath20 , and @xmath21 is the volume of the large bounding box that contains set @xmath0 of points .
the expected cardinality of output set @xmath7 corresponding to a given query point is shown to rely on @xmath0 and @xmath4 .
although this algorithm does not improve either space or query - time complexity of existing algorithms , it suggests an approach to explore the problem under the following setting : independently cover each voronoi cell with @xmath101 hyperboxes ( refer fig .
[ fig : conclu1 ] ) ; this leads to a space complexity of @xmath102 ( given that there are at most @xmath1 voronoi cells ) .
organize all the hyperboxes covering the entire voronoi diagram into a data structure , similar to the one proposed herewith , to achieve query time efficiency . | given a set @xmath0 of @xmath1 points in @xmath2-dimensional euclidean metric space @xmath3 and a small positive real number @xmath4 , we present an algorithm to preprocess @xmath0 and answer queries that require finding a set @xmath5 of @xmath4-approximate nearest neighbors ( anns ) to a given query point @xmath6 .
the following are the characteristics of points belonging to set @xmath7 : * @xmath8 , @xmath9 a point @xmath10 such that @xmath11 and the nearest neighbor of @xmath12 is @xmath13 , and * @xmath9 a @xmath14 such that @xmath15 is a nearest neighbor of @xmath16 . during the preprocessing phase , from the voronoi diagram of @xmath0
we construct a set of box trees of size @xmath17 which facilitate in querying anns of any input query point in @xmath18 time . here
@xmath19 equals to @xmath20 , and @xmath21 is the volume of a large bounding box that contains all the points of set @xmath0 .
the average case cardinality of @xmath7 is shown to rely on @xmath0 and @xmath4 . |
the outbreak of ebola virus disease ( evd ) in west africa caused 28,646 cases and 11,323 deaths as of march 30 , 2016 @xcite .
the current outbreak is of the ebov ( zaire ebola virus ) strain , the most fatal strain @xcite .
the largest previous outbreaks of the ebov strain occurred in the democratic republic of the congo in 1976 and 1995 , causing 318 and 315 cases @xcite .
clinical progression includes two broad stages of infection , often characterized as early and late @xcite . in the first stage ,
approximately five to seven days , symptoms include fever , weakness , headache , muscle / joint pain , diarrhea , and nausea @xcite . in some patients
, the disease progresses to a second stage , with symptoms including hemorrhaging , neurological symptoms , tachypnea , hiccups , and anuria @xcite .
mortality rates are higher among those exhibiting second - stage symptoms @xcite .
evd is transmitted through direct contact with an infected individual @xcite .
transmission risk factors include contact with bodily fluids , close contact with a patient , needle reuse , and contact with cadavers , often prepared for burial by the family @xcite .
this outbreak was primarily in the contiguous countries of guinea , sierra leone , and liberia , which experienced widespread and intense transmission @xcite .
interventions have included quarantine , case isolation , additional treatment centers , border closures , and lockdowns , restricting travel within a region ( as in a military - enforced curfew ) @xcite .
public health authorities must allocate resources effectively , focusing personnel and funds to respond best to outbreaks .
however , not all response measures are equally beneficial or cost - effective , and testing the relative benefits of each is often impossible or unethical .
modeling thus provides a valuable tool for comparing interventions and identifying areas where interventions are most effective .
modeling analyses can inform public health policy regarding ongoing and future outbreaks .
dynamic spatial modeling has been proposed as a useful approach to understand the spread of evd and evaluate response strategies benefits @xcite .
the recent outbreak has sparked an increase in evd modeling @xcite .
the spread between contiguous countries in the current evd outbreak highlights the spatial element to its proliferation @xcite ; however , as previous evd outbreaks were more localized than the 2014 - 2015 epidemic @xcite , there is little historical data on the geospatial spread of ebola .
mobility data , which may help inform spatial evd modeling , is limited , although some studies have highlighted its usefulness and extrapolated based on mobility data from other regions @xcite . to address the need for spatial evd models , models have been developed to examine local spatial spread within liberia @xcite and evaluate the potential risk of international spread using data such as airline traffic patterns@xcite .
our models add to existing spatial models by incorporating intervention comparison and an exploration of the dynamics between different regions , using an uncomplicated model structure . between - country mobility
is important in this epidemic because the borders between countries are porous ; borders are drawn across community or tribal lines , resulting in frequent border - crossing to visit family , conduct trade , or settle disputes @xcite .
in addition , santermans and coauthors recently demonstrated that the outbreak is heterogeneous : transmission rates differ between locales @xcite .
our district - level model offers a mechanistic explanation of observed spatial heterogeneities , adding to the existing literature by explicitly simulating the interactions between districts . in this study , we present spatial models of evd transmission in west africa , using a gravity model framework , which captures dynamics of local ( within - region ) and long - range ( inter - region ) transmission .
gravity models are used in many population - mobility applications , especially to relate spatial spread of a disease to regional population sizes and distances between population centers @xcite .
gravity models have been applied to other diseases , including influenza in the us and cholera in haiti @xcite , and have been used to examine general mobility patterns in west africa@xcite .
we demonstrate that the gravity modeling approach fits and forecasts case and death trends in each country , and describes transmission between countries .
the district - level model successfully simulates local geospatial spread of evd , indicating that the gravity model framework captures much of the local spatial heterogeneity . using the model structures , we examined interventions at country and district scales , evaluating the relative success of intervention types as well as the most responsive regions .
we developed a compartmental gravity model using ordinary differential equations to model spatiotemporal progression of evd ( figure [ fig : modelmap ] , model equations and details given in the supplementary information ) .
each country s compartmental structure includes susceptible ( @xmath0 ) , latent ( @xmath1 ) , two - stage infection ( @xmath2 , @xmath3 ) , funeral ( @xmath4 ) , recovered ( r ) , and deceased ( d ) , based on previous compartmental models @xcite . patients in stage
@xmath2 can recover or transition to @xmath3 , where they may recover with lower probability or transition to @xmath4 , based on clinical observations of symptom progression and mortality @xcite .
the latent period and two - stage infection are based on the clinical progression of evd , wherein patients become contagious in @xmath2 , with increasing contagiousness in @xmath3 @xcite .
funerals play a role in transmission because of the high viral load of the deceased and frequent contact with susceptible persons due to cultural burial practices @xcite .
the precise relative magnitude of this contribution is unknown ; previous models have shown that the relative contributions of each stage are unidentifiable from early data @xcite .
thus , patients in @xmath4 were assumed to be as contagious as those in @xmath3 .
the spatial component of the model is a three - patch gravity model .
each country is one patch , with a compartmental model within it and the capital as the population center , similar to previous gravity models , which used political capitals / large cities as centers @xcite .
after evd appeared in western guinea , subsequent cases were in the capital , conakry , suggesting that the capital acts as the central population hub .
similar progression occurred in each country : after evd cases appeared in border regions , cases soon appeared in the capital @xcite .
the force - of - infection term for patch @xmath5 consists of transmission from within the region and transmission from other patches into patch @xmath5 .
each long - range transmission term is determined by a `` gravity '' term , proportional to the population sizes , and inversely proportional to the squared distance between them .
cumulative local cases were measured by the number of infections in a patch due to local transmission .
cumulative long - range cases were measured by number of cases due to long - range transmission .
we developed a more detailed gravity model of west africa incorporating the 14 districts / areas of sierra leone , 15 counties of liberia , and 34 prefectures of guinea .
this model is structured the same way as the country - level model : a compartmental model in each administrative unit , linked to all other patches by gravity terms .
transmission is separated into local ( within - district ) and long - range ( between - district ) .
case and death incidence in each country was collected from world health organization ( who ) situation reports on evd from march 29 to october 31 , 2014 @xcite .
the country - model simulations used data from may 24 through september 30 for fitting model parameters .
data from october was used for validation , to compare model projections to data unused in fitting .
road distance between centers was used for the gravity component @xcite .
we evaluated direct distance between capitals in the country - level model , yielding similar results .
for further verification of the model , we also tested the model s ability to fit the outbreak when data was incorporated from march 29 .
the fits and forecasts from these simulations were similar to those incorporating data from may onward ( supplemental figures 1 - 2 ) .
the district model was simulated from march 31 , 2014 to january 31 , 2015 .
the model was compared to data on the outbreak s local geospatial progression from each district , using incidence of cases and deaths from who updates and the un ocha database @xcite .
model parameters were determined from clinical literature and fitting to available data , similarly to previous models @xcite . in the country - level model
, nine parameters were fitted to the data : transmission ( @xmath6 ) and death ( @xmath7 ) as well as three gravity - term constants , @xmath8 , were separately fitted for each country . .
the parameter ( @xmath8 ) adjusts the `` gravity '' terms to reflect the balance of local and long - range transmission in each country .
parameters , definitions , units , ranges , and sources are given in supplementary table s1 . in the country - level model ,
500 sets of initial values for all parameters were selected with latin hypercube ( lh ) sampling from realistic ranges for parameters , determined from who , cdc , and literature data , given in supp .
@xcite . for each of the 500 parameter sets , only transmission rates , death rates , and @xmath8
were fitted by least - squares , using nelder - mead optimization in matlab @xcite .
all other parameters were held constant to the values from the lh sample . in the district - level model ,
best - fit parameters from the country - level model were used for all parameters except @xmath9 .
a different @xmath9 was fitted for each district to reflect differing at - risk populations and reporting rates . since fitting 63 parameters
is highly complex , slow , and possibly unidentifiable , @xmath9 parameters were calculated according to final size , then adjusted by hand to generate curves matching the progression of the outbreak , both overall and in each district .
while all other initial conditions were determined from early data and @xmath9 , in conakry , guinea , the initial conditions were fitted independently from @xmath9 , reflecting possible reporting rate differences between initial situation reports and subsequent data .
interventions were compared by the number of cumulative cases the model forecasted on october 31 for different types and levels of intervention .
reduction of local transmission represents interventions that limit contact of an evd patient with others in his or her home country , including quarantine , isolation , hospitalization , and local lockdowns .
reduction of long - range transmission represents border closures , large - scale lockdowns , or other intervention measures that reduce the transmission between countries . in the district - level model
, interventions were simulated by eliminating the outbreak in one patch ( both local and long - range intervention ) , and simulating to january 31 .
this is analogous to effective case isolation and quarantine measures within a district , which would limit the transmission of evd within and from that district .
the effectiveness of intervention in one district was measured by calculating the cumulative number of cases reduced in all other districts .
the percent reduction from intervention in a district and percent reduction relative to size of the outbreak in that district were calculated .
the best - fit accurately fit and forecasted the outbreak data and trends for each country ( figure [ fig : modelfits ] ) .
the best - fit model forecast of cases on october 29 , 2014 , was 14,070 ; the who case data on that date was 13,540 @xcite .
cumulative cases in each country due to local and long - range transmission are shown in figure [ fig : local_longrange ] .
liberia had more local than long - range transmission .
early on , guinea had more local transmission , due to initial local cases . as the epidemic progressed , the transmission ranges overlapped , although the best - fit trajectory for long - range transmission remained smaller than the local - transmission contribution . in sierra leone ,
the best - fit trajectory of long - range transmission was more significant than local transmission , although forecasted ranges were similar .
the district model was successful in fitting outbreak data for cases and deaths in each district .
the model captured the final size of the outbreak in each patch with an @xmath10 value of 0.96 ( figure [ fig : districtr2 ] ) as well as matching the progression of the outbreak ( figure [ fig : districtmapprog ] ) .
the model captured the temporal dynamics of districts with negligible case counts or high levels of transmission .
the time - course plots of all 63 district - level fits are included in the supplement , as is a scatter plot of all data vs model values ( @xmath10 = 0.83 ) .
the district - level model accurately captured the trend of spread of evd through second - level administrative units .
while overall patterns at the district level are captured and show the correct ordering , the actual speed of disease spread in the model was faster in some districts than in the data , likely due to a combination of stochastic introductions and reporting delays .
the district - level model was able to forecast local spread of evd , matching data on spatial progression of evd to different locales as the outbreak intensified ( figure [ fig : districtmapprog ] ) .
an animation of the district - level simulation is given in the web supplement ( video s1 ) .
according to our intervention simulations ( figure [ fig : interventions ] ) , reduction in local transmission , from interventions such as isolation or improved case - finding and ebola treatment unit ( etu ) capacity , is most effective in liberia . in the model , eliminating local transmission in liberia reduced the outbreak by up to 11,000 cases in all countries by october 31 , 2014 , a 76% reduction .
reduction of long - range transmission was most effective in sierra leone . eliminating long - range transmission into sierra
leone reduced the outbreak by up to 9,600 cases in all countries by october 31 , a 66% reduction .
intervention analysis identified regions in which intervention was most successful ( figure [ fig : districtinterventions ] ) .
overall , the most successful regions for intervention were conakry , coyah , and dubreka , guinea , and lofa , and monrovia / montserrado , liberia . to evaluate whether intervention effects were disproportionately significant compared to cases within the target region , we defined the intervention - amplifying region ( iar ) level by the ratio of cases prevented in other patches to cases within the patch , for regions with more than 0.05% of the epidemic s cases .
dubreka , coyah , and conakry were the most effective iars ( figure [ fig : districtinterventions ] ) .
our results demonstrate that outbreak dynamics in all three countries can be accurately captured using a gravity - model approach . moreover , country - level model forecasts successfully predicted cumulative cases and deaths one month ahead ( figure [ fig : local_longrange ] ) .
the models were able to capture simultaneously the dynamic interactions within and between each region at both country and district levels .
the model simulations suggest a form of `` spatial herd protection '' , wherein interventions in one region benefit surrounding regions as well , by reducing the spatial transmission between them . in the country - level model , we compared the effects of local and long - range interventions in each country .
local interventions include more strict quarantine and isolation procedures , increased case - finding and etu capacity , safer burial practices , and other interventions that reduce contact of susceptible persons with infected persons in the local community , as well as behavioral changes that reduce local transmission .
the model indicates that these interventions were most effective in liberia , both in reducing the outbreak in liberia and mitigating the whole epidemic .
indeed , liberia s outbreak had the fastest initial growth rate , it was the first to turn over and end @xcite .
our results suggest that liberian interventions may have had significant indirect effects on the epidemic dynamics in guinea and sierra leone as well .
the most effective long - range transmission reduction was in sierra leone , with 66% case reduction across all countries when long - range transmission was completely eliminated .
this reflects the porous border in sierra leone .
based on the district - level intervention analysis , no single district in sierra leone acts as a source of long - range transmission to other locales , rather all districts play some combined role .
thus , the country - level impact of long - range transmission is possibly a summative combination of two factors across all districts : long - range transmission into sierra leone and early introduction of cases from sierra leone into liberia . in the district - level model , the capitals of montserrado , liberia and conakry , guinea were highly effective intervention sites .
their large population size makes them important in dictating the dynamics of the outbreak in surrounding regions .
dubreka and coyah , guinea , near conakry , were effective intervention sites despite low numbers of cases .
this indicates that they influenced the larger outbreaks surrounding them , especially conakry .
intervention in these districts was especially significant in reducing the scope of the outbreak , not just within the district but also in surrounding regions . in some regions
, intervention produced unexpectedly significant reductions in the overall outbreak : these districts amplify the effects of intervention to reduce disproportionately many cases outside the region
these are denoted intervention - amplifying regions ( iars ) .
the most significant iars include conakry , coyah , and dubreka in guinea and montserrado in liberia ( figure [ fig : districtinterventions ] ) .
for instance , dubreka made up only 0.13% of the outbreak but intervention in dubreka alone reduced the size of the outbreak in all regions by 21% , drastically higher than its case contribution would suggest .
coyah and dubreka were the most dramatic iars because they had relatively small numbers of cases .
their proximity to conakry , which had a large number of cases and also functioned as an iar , implies that coyah and dubreka could have influenced outbreak dynamics in conakry .
the outsize importance of these districts in the dynamics of the current outbreak makes them a possible target for increased intervention during future outbreaks in west africa .
several papers have noted that guinea s peculiar outbreak curve is difficult to fit using simple models , due to plateaus in cumulative incidence between growth periods @xcite .
however , by including spatial interaction between countries , we successfully captured guinea s outbreak dynamics ( figure [ fig : modelfits ] ) .
this suggests that spatial interactions , such as local die - outs of evd followed by long - range reintroductions , may be responsible for the unusual incidence patterns in guinea .
the country - level model s ability to capture the singular epidemic progression in guinea , due to its spatial transmission component , demonstrates that long - range dynamics do not just affect introduction , but also later stages of the outbreak .
the district - level model , which separates the affected countries into 63 patches based on secondary administrative units , was successful in capturing the overall dynamics of spatial transmission .
it captured the spread of evd from the initial sites of the outbreak to other locales in the region in an order and magnitude similar to the actual progression of the epidemic .
this suggests that the county - level model captures the underlying transmission patterns that led to spread of evd between different locales in the affected countries ( figure [ fig : districtmapprog ] ) .
these results are perhaps surprising given that the district - level model uses the same transmission parameters for all patches within the same country this suggests that even though there are likely to be significant local heterogeneities from patch to patch @xcite , most of these variations can be captured using the relatively simple framework afforded by the gravity model .
in fact , a model that incorporates local heterogeneities often requires a large amount of case data at the local scale @xcite .
this gravity model provided a reasonable approximation of the early dynamics of the outbreak , based simply on initial conditions from march 30 , soon after surveillance began : it can capture local differences in transmission without complex ( difficult - to - simulate ) models or extensive datasets .
thus , it could be applicable for early - outbreak forecasting by providing an indication of the areas at greatest risk for ebola cases , in order to guide resources and aid to those locations .
our models focus on early proliferation throughout west africa : interactions between regions , which led to the spatial spread of ebola . regional evd introductions
are inherently stochastic ; thus the model did not capture the precise temporal pattern in west africa in every district .
stochastic model simulations might better characterize the randomness of regional introduction , but a stochastic implementation of the 63-patch regional structure would be computationally demanding .
therefore , a deterministic model offers an elegant method to provide insight into the role of spatial dynamics .
in addition , we considered the time scales that allowed us to capture the spatial dynamics of interest . in the country - level model , the turnover and re - ignition of the epidemic in guinea was of particular interest , since simple compartmental models provide little insight into its mechanism @xcite .
our time - course included the first ignition , turnover , and re - ignition ; since simpler compartmental models fail to capture the dynamics of the outbreak , our model s highly accurate fit implies that the spatial structure captures the dynamics affecting guinea .
our time - course ends in november - december 2014 , at which time interventions were increasing throughout west africa .
this massive intervention scale - up affected the parameters for the outbreak : in order to simulate the progression of the outbreak after this time , time - variant parameters could be used .
the late dynamics were not the focus of this research ; however , the model could be implemented with time - variant parameters , potentially providing insight into late dynamics of the outbreak .
while this model captures epidemic dynamics without incorporating explicit movement patterns , further mobility data ( e.g. from cell phone carriers ) could elucidate dominant movement patterns in west africa .
this data could provide a validation of the gravity model s success , if the gravity model accounts for these patterns .
the model does not account for a detailed population structure : an organization of individuals into households , villages , or other communities .
the gravity model could be applied with a structured - population ( network ) model , such as kiskowski s model @xcite .
this model has been applied to study community - based interventions @xcite ; it could provide a granular simulation of ebola dynamics , with possible insights into community and differing levels of local intervention . during the course of the outbreak , parameters such as reporting rates , etu availability , and intervention rates changed .
models that use a broader time - scale than our model see more benefits to incorporating time - variant parameter @xcite .
in addition , transmission changes as behavior of infected and susceptible individuals changes @xcite .
thus , models that incorporate behavioral changes could be successful in capturing that aspect of the epidemic .
we developed dynamic transmission models that account for spatial spread by considering the `` gravitational pull '' of larger populations and shorter distances in west africa .
the country - level model accurately captures epidemic dynamics and successfully forecasts cases and deaths for all three countries simultaneously .
the district - level model captures the progression of evd between and within districts in west africa .
our models suggest that reduction of local transmission in liberia and reduction of long - range transmission in sierra leone were the most effective interventions for the outbreak .
the models also reveal differences in transmission levels between the three countries , as well as different at - risk population sizes .
our gravity spatial model framework for evd provides insight into the geographic spread of evd in west africa and evaluate the relative effectiveness of interventions on a large , heterogeneous spatial scale .
ultimately , gravity spatial models can be applied by public health officials to understand spatial spread of infectious diseases and guide interventions during disease epidemics .
this work was supported by the national institute of general medical sciences of the national institutes of health under award number u01gm110712 ( supporting mce ) , as part of the models of infectious disease agent study ( midas ) network .
the content is solely the responsibility of the authors and does not necessarily represent the official views of the national institutes of health .
_ country - level model_. as discussed above , the model consists of @xmath0 , @xmath1 , @xmath11 , @xmath12 , @xmath4 , and @xmath13 compartments in each country .
the model equations are given by : @xmath14 where @xmath15 , @xmath16 , and @xmath17 indicates each patch ( guinea , sierra leone , liberia ) , and @xmath18 and @xmath19 the other two patches ( with @xmath20 ) .
the risk of new infections from within each patch was represented by the @xmath21 term for the `` home '' patch .
tthe risk terms for evd transmission from outside the patch , @xmath22 and @xmath23 , include @xmath24 , the gravity term .
the distance exponent of the gravity term , @xmath25 , is fixed at two .
a range of values for @xmath25 were tested , however we found that changes in @xmath25 were compensated by changes in the fitted value of @xmath8 .
thus , @xmath25 was fixed for clarity , based on values in previous gravity models @xcite .
there are separate @xmath8 for each country , so the rate of transmission into each country is different .
this reflects the differing border porosities and rates of travel between countries .
+ _ district - level model_. similarly , the district - level model equations for patch @xmath17 are given by : @xmath26 this model structure is similar to that of the country - level model .
the main difference is that the long - range transmission is grouped into one @xmath22 term , which sums the long - range transmission of every other patch into patch @xmath17 .
the model parameters were estimated by simulating total cumulative cases , defined as the integral of the incidence ( @xmath29 ) multiplied by @xmath9 , a correction factor for underreporting and the fraction of the population at risk , among other factors @xcite .
initial conditions for both models were determined based on the initial values of the data as follows : the number of susceptible persons in each patch was estimated based on the population size scaled by @xmath9 .
the total population of each patch was determined using data from the world bank and national censuses @xcite .
the number of exposed persons was determined as twice the initial number of infected ( @xmath2 and @xmath3 ) persons , based on the @xmath30 for evd , which has been estimated to be approximately 2 @xcite .
the number of infected persons in i1 was determined based on the number of new cases in the previous nine days , based on the incubation period for evd , and the number of infected persons in i2 was determined by subtracting the number of deaths within the next four days from the number of currently infected persons at the starting date @xcite .
the number of infected persons from local versus long - range transmission was estimated based on the origin of the outbreak and the location of cases up until the starting point for the data : since the outbreak began in guinea , the initial cases in guinea were considered from local transmission , while the initial cases in sierra leone and liberia were considered to be from long - range transmission from guinea .
the number of recovered persons was estimated based on the number of infected persons who did not die within the nine - day period that preceded the starting date .
the number of persons in the f class was based on a fraction of the number of persons who died within the period before the starting date , estimated using the burial rate in table 1 .
the initial conditions were all scaled to fractions of the population using the parameter @xmath9 .
we tested the country - level model s ability to fit the outbreak when data was incorporated using a start date of march 30 . in the simulations starting on march 30 , local transmission in liberia
was turned off until may ( when the first case appeared in liberia ) , as the ode framework used here can not capture the stochastic process which leads to emergence of an outbreak in a new locale .
the resulting fits and forecasts from these simulations were similar to those incorporating data from may onward ( supplemental figures [ fig : modelfits_sup ] , [ fig : local_longrange_sup ] ) . for the district - level model , parameters
were determined using sampling from within plausible ranges and from the best - fits of the three - patch models .
the 1 transmission rates were determined at the country level , reflecting the differing risks of transmission in the patches in each country .
each @xmath9 was calculated to match final outbreak size , then adjusted by hand .
the initial conditions were determined based on data from the who reports and from data compiled by the un .
the initial conditions were determined similarly to the country - level model , based on data from the who reports and from data compiled by the un .
the district capitals were considered to be the population centers for all districts except montserrado , liberia , and bonthe , sierra leone .
montserrado county is the location of the national capital and largest city , monrovia , which was considered the population center .
the capital of bonthe is on a coastal island , so yagoi , the closest city to bonthe on the coast , was considered the population center . in the country - level model
, interventions were simulated by reducing transmission parameters for either local or long - range transmission .
local transmission reductions were simulated by reducing 1 , the rate of local infections , in increments of one percent of the original value .
long - range transmission reductions were simulated by reducing 2 and 3 , the rates of infections from outside the patch , in one - percent increments . in the district - level model , interventions were simulated by eliminating the outbreak in one district ( local transmission , long - range transmission into the district , and long - range transmission out of the district ) .
the effect on the 62 surrounding patches was measured ( difference in number of cases in those 62 patches in the normal model simulation versus number of cases in the 62 patches in the intervention simulation ) .
in addition , the iar score was determined as follows : best fit & range & sources + @xmath32 & transition rate from @xmath1 to @xmath2 & days@xmath33 & 0.1059 & 0.1 - 0.125 & @xcite + @xmath34 & transmission rate for @xmath2 in guinea & persons@xmath33 days@xmath33 & 0.0950 & initial : 0 - 0.5 , fitted : 0.0000 to 0.0676 & fitted to data @xcite + @xmath35 & transmission rate for @xmath2 in sierra leone & persons@xmath33 days@xmath33&0.0504&initial : 0 - 0.5 , fitted : 0.0005 to 0.1858&fitted to data @xcite + @xmath36 & transmission rate for @xmath2 in liberia&persons@xmath33 days@xmath33&0.1555&initial : 0 - 0.5 , fitted : 0.0727 to 0.1656&fitted to data @xcite + @xmath37&ratio of @xmath2 transmission to @xmath3 and @xmath4 transmission&none&2.4565&1.5 - 3.0 & @xcite + @xmath38 & transmission rate for @xmath3 and @xmath4 in guinea&persons@xmath33 days@xmath33&0.2335&calculated from @xmath39 and @xmath37&see @xmath39 , @xmath37 + @xmath40&transmission rate for @xmath3 and @xmath4 in sierra leone&persons@xmath33 days@xmath33&0.1238&calculated from @xmath39 and @xmath37&see @xmath39 ,
@xmath37 + @xmath41 & transmission rate for @xmath3 and f in liberia&persons@xmath33 days@xmath33&0.3820&calculated from @xmath39 and & see @xmath39 , @xmath37 + @xmath42 & mortality for infected persons in guinea&none&0.6643&initial : 0.5 - 0.9 , fitted : 0.3601 to 1.4023&fitted to data @xcite + @xmath43 & mortality for infected persons in sierra leone&none&0.3910&initial : 0.5 - 0.9 , fitted : 0.2547 to 0.6159&fitted to data @xcite + @xmath44 & mortality for infected persons in liberia&none&0.6710&initial : 0.5 - 0.9 , fitted : 0.5453 to 0.8236&fitted to data @xcite + @xmath45&fraction of persons in @xmath3 that dies&none&0.9732&0.9
- 1.0&@xcite + @xmath46&death rate for persons in @xmath3&days@xmath33&0.7597&calculated from @xmath45 and @xmath47&see @xmath45 and @xmath47 + @xmath48&burial rate&days@xmath33&0.7425&0.3333 - 1.0 & @xcite + @xmath47&1/duration of @xmath3&days@xmath33&0.7806&0.3 - 0.8 & @xcite + @xmath49&1/duration of @xmath2&days@xmath33&0.1822&0.1429 -0.2&@xcite + @xmath25&distance exponent&unitless&2&&fixed + @xmath50&between - patch coupling strength for guinea&km@xmath51 persons@xmath52&@xmath53&initial : @xmath54-@xmath55,fitted : @xmath56 - @xmath57&fitted to data @xcite + @xmath58&between - patch coupling strength for sierra leone&km@xmath51 persons@xmath52&@xmath59&initial : @xmath54-@xmath55,fitted : @xmath60-@xmath61&fitted to data @xcite + @xmath62&between - patch coupling strength for liberia&km@xmath51 persons@xmath52&@xmath63&initial:@xmath54-@xmath55,fitted : @xmath64-@xmath65&fitted to data @xcite + @xmath66 & recovery rate of persons in @xmath2 in guinea&days@xmath33&0.0578&calculated from @xmath67 , @xmath45 , @xmath49&fitted to data @xcite ( see @xmath67 , @xmath45 , @xmath49 ) + @xmath68 & recovery rate of persons in @xmath2 in sierra leone&days@xmath33&0.1090&calculated from @xmath67 , @xmath45 , @xmath49&fitted to data @xcite ( see @xmath67 , @xmath45 , @xmath49 ) + @xmath69 & recovery rate of persons in @xmath2 in liberia&days@xmath33&0.0566&calculated from @xmath67 , @xmath45 , @xmath49&fitted to data @xcite(see @xmath67 , @xmath45 , @xmath49 ) + @xmath70 & recovery rate of persons in @xmath3 & days@xmath33&0.0209&calculated from @xmath45 , @xmath47&@xcite + @xmath71&rate and proportion of movement of persons from @xmath2 to @xmath3 in guinea & days@xmath33&0.1243&calculated from @xmath67 , @xmath45 , @xmath49&fitted to data @xcite ( see @xmath67 , @xmath45 , @xmath49 ) + @xmath72 & rate and proportion of movement of persons from @xmath2 to @xmath3 in sierra leone&days@xmath33&0.0732&calculated from @xmath67 , @xmath45 , @xmath49&fitted to data @xcite(see @xmath67 , @xmath45 , @xmath49 ) + @xmath73&rate and proportion of movement of persons from @xmath2 to @xmath2 in liberia&days@xmath33&0.1256&calculated from @xmath67 , @xmath45 , @xmath49&fitted to data @xcite(see @xmath67 , @xmath45 , @xmath49 ) + @xmath21 & transmission rate from s to @xmath2 for locally transmitted cases of evd&days@xmath33&varies&calculated from @xmath39 , @xmath74 , @xmath2 , @xmath2 , @xmath4&see @xmath39 ,
@xmath74 + @xmath75 & transmission rate from s to @xmath2 for long - range transmitted cases of evd&days@xmath33&varies&calculated from @xmath39 , @xmath74 , @xmath2 , @xmath2 , @xmath4&see @xmath39 , @xmath74 + @xmath24&influence of long range cases on home patch&unitless & & calculated from @xmath76 , @xmath77 , @xmath25 , @xmath8&see @xmath76 , @xmath77 , @xmath25 , @xmath8 + @xmath78 & population in patch @xmath17&persons&&fixed for each country&@xcite + @xmath79 & distance between patch @xmath17 and @xmath18 & km & &fixed for each combination of patches&@xcite + @xmath9 & correction factor for reporting rate , population at risk , and other factors & unitless&0.0030&0.001 - 0.1&sampled + sources + @xmath32 & transition rate from @xmath1 to @xmath2 & days@xmath33 & 0.1059 & @xcite + @xmath34 & transmission rate for @xmath2 in guinea & persons@xmath33 days@xmath33 & 0.0950 & fitted to data @xcite + @xmath35 & transmission rate for @xmath2 in sierra leone & persons@xmath33 days@xmath33&0.0504 & fitted to data @xcite + @xmath36 & transmission rate for @xmath2 in liberia&persons@xmath33 days@xmath33&0.1555&fitted to data @xcite + @xmath37&ratio of @xmath2 transmission to @xmath3 and @xmath4 transmission&none&2&@xcite + @xmath42 & mortality for infected persons in guinea&none&0.6643&fitted to data @xcite + @xmath43 & mortality for infected persons in sierra leone&none&0.3910&fitted to data @xcite + @xmath44 & mortality for infected persons in liberia&none&0.6710&fitted to data @xcite + @xmath45&fraction of persons in @xmath3 that dies&none&0.97&@xcite + @xmath48&burial rate&days@xmath33&0.9 & @xcite + @xmath47&1/duration of @xmath3&days@xmath33&0.8 & @xcite + @xmath49&1/duration of @xmath2&days@xmath33&0.2&@xcite + @xmath71&rate and proportion of movement of persons from @xmath2 to @xmath3 in guinea & days@xmath33&0.1370&fitted to data @xcite ( see @xmath67 , @xmath45 , @xmath49 ) + @xmath72 & rate and proportion of movement of persons from @xmath2 to @xmath3 in sierra leone&days@xmath33&0.0806&fitted to data @xcite(see @xmath67 , @xmath45 , @xmath49 ) + @xmath73&rate and proportion of movement of persons from @xmath2 to @xmath2 in liberia&days@xmath33&0.1384&fitted to data @xcite(see @xmath67 , @xmath45 , @xmath49 ) + @xmath25&distance exponent&unitless&2 & fixed + @xmath8&between - patch coupling strength & km@xmath51 persons@xmath52 & @xmath80 & fitted to data @xcite + @xmath24&influence of long range cases on home patch&unitless & calculated from @xmath76 , @xmath77 , @xmath25 , @xmath8&see @xmath76 , @xmath77 , @xmath25 , @xmath8 + @xmath78 & population in patch @xmath17&persons&fixed for each patch&@xcite + @xmath79 & distance between patch @xmath17 and @xmath18 & km & fixed for each combination of patches&@xcite + @xmath9 & correction factor for reporting rate , population at risk , and other factors & unitless&varies by patch&fitted to data + caption for supplementary animation : + * video s1 . * a gif representing the progression of ebola according to the model from march 30 , 2014 , to january 31 , 2015 . color intensity of district represents number of cumulative cases in that district : darker color represents higher number of cases . | the 2014 - 2015 ebola virus disease ( evd ) epidemic in west africa was the largest ever recorded , representing a fundamental shift in ebola epidemiology with unprecedented spatiotemporal complexity .
we developed spatial transmission models using a gravity - model framework to explain spatiotemporal dynamics of evd in west africa at both the national and district - level scales , and to compare effectiveness of local interventions ( e.g. local quarantine ) and long - range interventions ( e.g. border - closures ) . incorporating spatial interactions
, the gravity model successfully captures the multiple waves of epidemic growth observed in guinea .
model simulations indicate that local - transmission reductions were most effective in liberia , while long - range transmission was dominant in sierra leone .
the model indicates the presence of spatial herd protection , wherein intervention in one region has a protective effect on surrounding regions .
the district - level intervention analysis indicates the presence of intervention - amplifying regions , which provide above - expected levels of reduction in cases and deaths beyond their borders .
the gravity - modeling approach accurately captured the spatial spread patterns of evd at both country and district levels , and helps to identify the most effective locales for intervention . this model structure and
intervention analysis provides information that can be used by public health policymakers to assist planning and response efforts for future epidemics .
+ + * keywords * : ebola virus disease , transmission modeling , spatial modeling , interventions + * abbreviations * : evd , ebola virus disease ; etu , ebola treatment unit ; , who , world health organization , iar , intervention - amplifying region |
the excitation of long - lived solitary waves during the interaction of high - intensity laser pulses with plasmas is a topic with applications and of theoretical interest . as multi - dimensional particle in cell ( pic ) simulations
have shown , these waves form behind the laser pulse and they consist of electron density depressions with a trapped intense electromagnetic field oscillating at a frequency well below the laser frequency @xcite .
solitary waves can propagate towards the plasma - vacuum interface where the stored electromagnetic energy is radiated away in the form of low - frequency electromagnetic bursts @xcite , a process recently detected in the laboratory @xcite .
bright spots in optical plasma images with the same polarization as the laser pulse have been attributed to the formation of such solitons @xcite .
these waves can evolve to a state named postsoliton , that has also been observed in the laboratory with proton imaging techniques @xcite .
the propagation of electromagnetic solitary waves has been intensively investigated within the cold , relativistic , one - dimensional fluid approximation @xcite . restricting attention to circularly polarized travelling wave solutions , with velocity @xmath0 and frequency @xmath1 , yields a pair of second order differential equations that governs the dynamics of the electrostatic potential @xmath3 and the amplitude of the vector potential @xmath4 .
hence , the system describes a langmuir and an electromagnetic wave coupled by the nonlinear terms arising from the perturbation of the density and the relativistic mass .
it admits solitary waves with vanishing ( vbc ) and nonvanishing ( nvbc ) boundary conditions and , even though the governing equations are not completely integrable , they are commonly referred to as solitons . for vbc , @xmath5 and @xmath6 as @xmath7
, a soliton is interpreted as a light wave which is trapped in a self - generated plasma wave @xcite .
they have been classified according to the number of zeros of the vector potential profile , @xmath8 , see ref .
they are commonly referred to as bright solitons and their stability has been analyzed too @xcite . in @xcite , the existence of such solutions is discussed in the parametric @xmath2 plane and @xmath9 families of solitons are identified .
some of these families end at certain velocity values where the ion density profile shows a cusp at the center of the soliton .
the soliton breaking has been proposed as a mechanism for particle acceleration in high - intensity laser plasma interaction @xcite and it was observed in pic simulations for solitons with @xmath10 and overcritical amplitude @xcite .
solitons with vbc in warm @xcite and magnetized plasmas @xcite have been studied too . on the other hand , three different solutions with nvbc
are possible @xcite : ( i ) grey solitons ( @xmath11 and @xmath6 as @xmath12 ) , ( ii ) dark solitons ( @xmath13 and @xmath6 as @xmath12 ) and ( iii ) shock waves ( @xmath11 and @xmath6 as @xmath14 and @xmath5 and @xmath15 as @xmath16 ) .
the asymptotic values @xmath17 and @xmath18 are related to the two parameters @xmath0 and @xmath1 .
the branch of shock waves splits the @xmath2 plane in two different regions where either dark or grey solitons exist .
branches of solutions are found and they break down at increasing @xmath17 due to divergence of the electron density @xcite . as we will see , the grey and dark solitons with nvbc can be interpreted as a localized modulation in a long circularly polarized electromagnetic wave coupled with a plasma wave . even though the circularly polarized wave is susceptible to the relativistic raman and the modulational instabilities @xcite , the analysis of these solitary structures
is fully justified .
first , the present work extends the discussion about solitary waves with nvbc , that until now was limited to the narrow velocity range , @xmath19 @xcite .
second , the solitary waves could play a role in processes that are faster than the inverse of the parametric instability growth rate .
we also point out that any damping mechanism , not included in the present analysis for simplicity , could reduce or suppress these instabilities . from the mathematical point of view , the problem of finding solitons reduces , through the traveling wave ansatz , to that of finding orbits of an associated hamiltonian system for @xmath4 and @xmath3 consistent with the boundary conditions of the fluid model for each type of soliton .
the hamiltonian system is four dimensional , time - reversible and autonomous .
it admits four fixed points that , for brevity , will be denoted by the letter @xmath20 and the value of @xmath3 and @xmath4 inside brackets .
only three of them , @xmath21 and @xmath22 , play a role when discussing the existence of solitary waves .
orbits that are asymptotic to an equilibrium as @xmath7 are called _ homoclinic connections _ , while orbits that connect two distinct equilibria in the limits @xmath7 are termed _ heteroclinic connections _ ( see @xcite for a review of homoclinic orbits in reversible systems ) .
therefore , grey solitons , dark solitons and shock waves correspond to the connecting orbits @xmath23 , @xmath24 and @xmath25 respectively .
existence and robustness under variations in @xmath0 and @xmath1 of such connecting orbits can be directly inferred by simple geometrical arguments from the theory of dynamical systems , with a critical role played by the properties of the equilibria of the system and in particular their linear stability .
previous works @xcite studied the parametric domain where @xmath26 is a saddle - center , with eigenvalues @xmath27 , @xmath28 , and they found solutions in the velocity range @xmath19 . here , we extend the analysis to velocities up to @xmath29 , and specifically in regions of the parametric domain where @xmath26 is a saddle - focus , i.e. its eigenvalues are of the form @xmath30 ) .
it is well known @xcite that homoclinic connections of a saddle - focus in hamiltonian systems do generally exist and are robust against small perturbations of the hamiltonian .
therefore one expects them to appear in continuous families in the @xmath2 plane .
moreover , existence of such a homoclinic orbit , for instance one that corresponds to a one - hump soliton , implies existence of infinitely many of them , corresponding to solitons with a different number of humps @xcite .
taking into account these insights from the theory of dynamical systems , we numerically identify new families of single and multi - hump solitons of the grey and dark varieties that can be interpreted as counterparts to the branches reported for vbc @xcite .
we also report , for the first time , asymmetric solitons in this system and explain the disappearance of certain families of solutions as parameters are varied .
the organization of the paper is as follows . in section [ sec : dynamical ]
the system of equations that governs the dynamics of the solitons is revisited , while its fixed points and their possible connecting orbits are discussed in section [ sec : fp : orbits ] .
section [ sec : algorithm ] describes the spectral algorithm @xcite that we have used to find homoclinic and heteroclinic orbits , as well as some optimizations carried out for our specific problem .
the numerically computed families of dark and grey , symmetric and asymmetric solitons are presented in section [ sec : results ] . in section [ sec : stability ]
the stability of certain types of grey solitons is investigated with the aid of particle - in - cell ( pic ) simulations .
the conclusions are summarize in section [ sec : conclusions ] where the similarities and differences with the vbc waves are stressed .
this section briefly summarizes the theory of one - dimensional circularly polarized solitons ( see @xcite for a detailed discussion ) .
the plasma is assumed to be cold and composed of electrons and ions , denoted by the subscript @xmath31 respectively .
it is convenient to use length , time , velocity , momentum , vector and scalar potential , and density normalized over @xmath32 , @xmath33 , @xmath34 , @xmath35 , @xmath36 and @xmath37 respectively .
here @xmath37 , @xmath38 and @xmath39 are the unperturbed density , the rest mass and the electron plasma frequency . using this notation the maxwell ( in the coulomb gauge ) and plasma equations read @xmath40 @xmath41 @xmath42 @xmath43[sys : general ] where @xmath44 and @xmath3 are the vector and scalar potential , @xmath45 , @xmath46 and @xmath47 and @xmath48 are the kinetic momentum and the fluid velocity respectively . for convenience the dimensionless parameter @xmath49 has been introduced ( @xmath50 and @xmath51 are species charges ) . taking @xmath52 , the coulomb gauge immediately gives @xmath53 where @xmath54 denotes the direction perpendicular to @xmath55 .
the transverse component of yields @xmath56 .
we also take all the variables to be functions of @xmath57 and assume a circularly polarized vector potential @xmath58 here , @xmath0 is the group velocity of the solitary wave , @xmath59 is the wavevector and @xmath1 is the frequency in a frame moving with the solitary wave . with the above assumptions , equation and the longitudinal component of can be integrated . imposing the boundary conditions @xmath60 , @xmath61 , @xmath62 and @xmath63 as @xmath64 , the kinetic momentum , energy and the density of each species are functions of just the potentials @xmath3 and @xmath4 .
for instance , the densities and the @xmath65 factors are given by @xmath66 @xmath67 with @xmath68 , @xmath69^{1/2}$ ] and @xmath70 .
for brevity we write @xmath71 and for the numerical calculation we set @xmath72 . substituting in and yields @xcite @xmath73a\label{sys : hamiltonian : a}\ ] ] @xmath74 and @xmath75 , where the prime denotes derivative with respect to @xmath76 .
we remark that , for solutions described by equation , the group velocity of the solitary wave @xmath0 should be equal to the phase velocity @xmath77 .
solutions with different velocities require the addition of a relative phase in equation @xcite .
system - describes the dynamics of localized electromagnetic modulations trapped by a self - generated langmuir wave embedded in a infinitely long electromagnetic wave . introducing the momenta @xmath78 and @xmath79 , it can be written as a fourth order hamiltonian system with hamiltonian @xmath80-\frac{1}{2}p_\phi^2 + v\left[r_e(a,\phi)+\frac{r_i(a,\phi)}{\epsilon}\right]\label{eq : inv}\ ] ]
we also note that the conditions @xmath81 imply the following restriction on @xmath3 @xcite @xmath82
the fixed points are obtained by setting the righthand side of - equal to zero and , for brevity , they will be denoted by @xmath83 ( assuming @xmath84 as @xmath85 ) .
we note that only orbits connecting the fixed points @xmath86 are consistent with the previously imposed boundary conditions . hence , @xmath26 and their stability properties play an important role when discussing the presence of solitary waves in the full fluid system .
one readily verifies that @xmath21 is a fixed point of system - if the following dispersion relation is satisfied @xmath88 its physical interpretation is evident when we express as a function of the wavevector @xmath89 and the frequency in the laboratory frame @xmath90 @xcite @xmath91 that is the normalized dispersion relation of a pure transverse electromagnetic wave with relativistic effects and the ion motion .
note that equation has a solution within the range @xmath92 .
the stability of @xmath26 is determined by the eigenvalues of the jacobian of system - . in the @xmath2 plane there
exist 3 different regions of stability ; for any given @xmath1 , @xmath26 is a saddle - center if @xmath93 , a center if @xmath94 and a saddle - focus if @xmath95 . the velocities @xmath96 and @xmath97 are ( see appendix [ sec : appendix ] and figure [ fig : parametric ] ) @xmath98 ^ 2}{(\gamma_i^3+\epsilon\gamma_e^3)^2-\left[\gamma_i^3(\gamma_e^2+a_0 ^ 2)+\epsilon\gamma_e^3(\gamma_i^2+\epsilon^2a_0 ^ 2)\right]^2}}\nonumber\\ & & \label{eq : vsc}\\ v_{sf } & \equiv & \sqrt{1-\left[\frac{2a_0(1-\epsilon^2)\gamma_e\gamma_i}{\gamma_i^3(\gamma_e^2+a_0 ^ 2)+\epsilon\gamma_e^3(\gamma_i^2+\epsilon^2a_0 ^ 2)}\right]^2}\label{eq : vsf}\end{aligned}\ ] ] note that @xmath99 as @xmath100 ( @xmath101 ) and @xmath102 and @xmath103 as @xmath104 ( @xmath105 ) .
system - also admits the fixed point @xmath107 . since orbits must connect @xmath106 to @xmath26 , we need to enforce the same value of the invariant ( equation ) at both fixed points .
this yields @xmath108^{-1/2}\frac{\epsilon a_0 ^ 2}{2\gamma_{e}\gamma_{i}}\label{eq : vs}\ ] ] which gives an explicit relation between the velocity of the wave and the asymptotic value @xmath17 , plotted in figure [ fig : parametric ] . as @xmath104 @xmath109 one has @xmath110 .
appendix [ sec : appendix ] shows that @xmath106 is a saddle - center for velocities given by and , in principle , shock waves are possible in the range @xmath111 .
shock waves belong to the parametric domain where @xmath26 is a saddle - center , see figure [ fig : parametric ] , and will not be studied in the present work ( see @xcite and @xcite ) .
the system has an additional fixed point @xmath112 for velocities @xmath0 greater than @xmath113 ( see appendix [ sec : appendix ] ) .
as @xmath26 and @xmath112 must share the same value of @xmath114 for a connection to exist , a condition similar to can be obtained , which can not be fullfilled for any value of the parameters and therefore heteroclinic connections involving @xmath112 and @xmath26 are not possible ( see appendix [ sec : appendix ] ) .
characteristic velocity curves and @xmath26 stability regions in the @xmath2 plane .
the velocities @xmath115 and @xmath116 ( solid lines ) split the plane in three regions where @xmath26 has different stability . along the curve @xmath117 (
dashed line ) shock waves are possible .
the inset shows a detail close to @xmath118 where both @xmath97 and @xmath119 approach 1 as @xmath101 . ]
we now turn to the discussion of the conditions under which heteroclinic and homoclinic connections , and therefore solitons of the fluid system - are expected to exist .
we will need to recall well known facts from the theory of dynamical systems ( see @xcite , for example ) and introduce the notion of the _ stable ( unstable ) manifold _ @xmath120 ( @xmath121 ) of a fixed point @xmath122 , as the set of forward ( backward ) in @xmath76 trajectories that terminate at @xmath122 .
a heteroclinic ( homoclinic ) connection from @xmath123 to @xmath124 ( to itself ) lies on the intersection of the unstable manifold of @xmath123 and the stable manifold of @xmath124 ( @xmath123 ) .
stable and unstable manifolds are complicated objects with intriguing structure and their visualization can become a formidable task ( see @xcite for a review of methods ) . here it will be enough to consider the dimensionality of stable and unstable manifolds , as this dictates whether or not they are in general expected to intersect .
we note that dynamics of the @xmath125-dimensional system - are constrained to a @xmath126-dimensional _ energy - manifold _ by conservation of the invariant @xmath114 , equation .
the intersection of stable and unstable manifolds has to take place within this energy - manifold .
we first discuss homoclinic connections involving @xmath26 .
when @xmath26 is a saddle - center , both its stable and unstable manifolds are @xmath127-dimensional and are , in general , not expected to intersect in @xmath126-dimensional space . if they do intersect , they necessarily have to be the same from uniqueness of solutions .
a homoclinic connection is therefore not generic , in the sense that it requires the condition @xmath128 which can only be fullfilled for specific @xmath0 and @xmath1 values .
hence the homoclinic connections in the saddle - center case are expected to occur in branches in the @xmath2 plane .
however , the existence of a continuous spectrum of single - hump solitons when @xmath26 is a saddle - center has been suggested for vbc @xcite and nvbc @xcite boundary conditions . unless further restrictions are imposed ,
such a continuous spectrum has to be interpreted as a numerical artifact , only valid within the accuracy of long - time integration of system - .
nevertheless , singular cases could arise .
for instance , at the @xmath10 case where the dynamics is @xmath129-dimensional , it can be proved analytically that standing solitary wave solutions exist within a continuous @xmath1 range @xcite . in the saddle - focus case on the other hand , @xmath130 and @xmath131
are @xmath129-dimensional and are in general expected to intersect transversally along a @xmath127-dimensional curve in the @xmath126-dimensional energy - manifold .
therefore we expect homoclinic solutions to exist generically in the @xmath2 plane .
furthermore it is well known that , for given parameters , if one such transverse intersection exists then there exist infinitely many , which have been shown to form a local , complete horseshoe structure @xcite . in our case this implies that for any one - hump soliton we can find an associated family of multi - hump solitons for any given @xmath0 and @xmath1 @xcite .
the symmetry properties of - help us deduce more properties of the connecting orbits .
equations - are invariant under the reflection symmetry , @xmath132 . as a result , for any homoclinic @xmath133 connection there exists an identically shaped @xmath134 one .
note that , since we impose inhomogeneous boundary conditions , we can not have reflection invariant connections .
moreover , reflection symmetry allows us to carry many of the results available for homoclinic orbits over to heteroclinic orbits involving @xmath135 and @xmath136 ( dark solitons ) ; the two equilibria can be considered as a single equilibrium in a reduced system in which @xmath4 is identified with @xmath137 and in which heteroclinic connections become homoclinic @xcite
. therefore we can expect heteroclinic orbits in the saddle - focus parametric regime to occur generically .
furthermore , the @xmath24 connection has an identically shaped @xmath138 counterpart .
system - is also time - reversible , that is invariant under simultaneous change of sign of @xmath76 and the generalized momenta @xmath139 , @xmath140 . as a result ,
homoclinic orbits can either be symmetric ( self - dual , @xmath141 ) or come in pairs of asymmetric orbits related by time - reversal .
similarly , heteroclinic connections @xmath24 have to be either antisymmetric functions of @xmath76 ( @xmath142 ) or come in asymmetric pairs , related by @xmath76-reversal .
note that the antisymmetric heteroclinic solitons are associated with the combined action of @xmath76-reversal and reflection .
asymmetric homoclinic and heteroclinic orbits in @xmath76-reversible hamiltonian systems are well studied @xcite , but to our knowledge they appear here for the first time in the context of relativistic solitons , see section [ sec : results ] .
the addition of a perturbation that breaks the conserved quantity @xmath114 but preserves the reversibility would destroy asymmetric solutions @xcite . for later use , we introduce the so - called _ symmetric section _
@xmath143 , which is in a sense a symmetry hyperplane of the system , as any point in it is left invariant under @xmath76-reversal .
symmetric homoclinic orbits have to intersect @xmath144 @xcite .
as we will see in section [ sec : results ] symmetric homoclinic orbits can disappear through a mechanism referred to as _ coalescence _
@xcite , as a result of which the stable and unstable manifolds fail to intersect on the symmetric section @xmath144 .
due to time reversibility , the computation of symmetric homoclinic or antisymmetric heteroclinic orbits of - involves the solution of a boundary value problem on the semi - infinite interval ( @xmath145 ) . within the regime where @xmath26 is a saddle - center and the unstable
manifold is one - dimensional , previous works truncate this interval and consider the linearized dynamics close to the fixed points , solving for the parameter values in the @xmath2 plane in which the boundary value problem is satisfied @xcite . for the saddle - focus case , where the local unstable manifold is a @xmath129-dimensional plane , one would also need to solve for the initial condition in this plane , for instance by parametrizing it by a polar angle @xcite . here
we implement the rational spectral collocation algorithm of ref .
@xcite that avoids both the truncation of the domain and the introduction of an additional parameter .
let us consider the system @xmath146 with @xmath147 and the components of @xmath148 given by the right - hand sides of - .
assuming a fast enough approach of the solutions to the asymptotic values @xmath149 as @xmath85 @xcite , the variables can be expanded as a sum of orthogonal rational functions @xmath150 , \ \ \ \ \ \ i=1,\ 2 \label{eq : expansion1}\end{aligned}\ ] ] the basis functions suggest the following choice for the m collocation points @xmath151 that are complemented by the two following collocation points @xmath152 and @xmath153 .
the coefficients @xmath154 are given by @xmath155 with @xmath156 if @xmath157 or @xmath158 and @xmath159 if @xmath160 .
computing the second derivative from and substituting the results in @xmath161 yield @xmath162 nonlinear algebraic equations @xmath163=0\ \ \ \ \ \
1\leq j\leq m,\ \ \ \ \ \ i=1,2\label{eq : collocation : eqs}\ ] ] where @xmath164\sin^4\left(\frac{j\pi}{m+1}\right)\ ] ] we also have 4 boundary conditions @xmath165 @xmath166 with @xmath167 .
since the system is autonomous , every translation on @xmath76 of its solution is also a solution .
this indeterminacy can be removed by adding the phase condition @xmath168 where @xmath169 denotes the @xmath170 inner product and @xmath171 the previous orbit on a branch .
solutions occuring at branches @xmath172 need this extra condition to adequately vary the parameter .
this would be the case of the branches found for @xmath93 @xcite .
however , as we will see , for @xmath95 the solutions exist continuously in the @xmath2 plane . in this case equations and - constitute a set of @xmath173 equations for the @xmath173 coefficients @xmath154 that can be solved for fixed values of @xmath1 and @xmath0 .
we used a newton - raphson algorithm with its jacobian calculated analytically to speed up the convergence . since system -
does not include the first derivatives of the variables @xmath174 on the right - hand side , we have directly computed the second derivative .
this is a difference from the general scheme given in reference @xcite that allows us to divide by two the number of coefficients .
note also that for asymmetric solutions all the coefficients must be calculated whereas in the case of symmetric or antisymmetric solutions just one half of the coefficients are needed ( for instance symmetric grey solitons are even solutions and @xmath175 ) .
the variable @xmath76 can be stretched according to @xmath176 .
here @xmath177 is a scaling factor that can be used to optimize accuracy of the solution .
although some strategies have been discussed to find a proper value of @xmath177 @xcite , in practice it is chosen by experimentation with different @xmath177 for a given value of @xmath178 ( as suggested in reference @xcite ) .
we found @xmath179 to be an adequate value for our numerical computations and a number of collocation points @xmath180 .
the initial guess for the newton - raphson algorithm can be obtained from the intersection of stable and unstable manifolds , as discussed in section [ sec : results ] . for asymmetric solutions
it can also be built up from pieces of symmetric solutions of different parity , smoothed by dropping the high order coefficients .
similarly , guesses for multi - hump solutions can be built up from single hump solutions of appropriate shape .
once we have a good initial guess , the spectral algorithm gives an accurate solution that can be used as initial guess for other parameter values .
we finally remark that the accuracy of the solutions can be checked by testing the spectral convergence : a plot of the logarithm of the coeficients @xmath154 versus the order @xmath59 should decay linearly .
this is an advantage as compared to the truncation of the interval methods where the error is controlled by the initial distance to the fixed point which is fixed arbitrarily .
such a criterion allows to separate true from false solutions , specially when the fixed point has eigenvalues with small real part ( in absolute value ) and the solutions approach slowly to their asymptotic values .
figure [ fig : homo1 ] depicts our numerically computed grey soliton solutions , belonging to a certain family . in panel
[ fig : homo1](a ) we also plot the curves @xmath96 and @xmath97 .
the panels ( d)-(f ) display a representative solution at the parameter values @xmath181 and @xmath182 , just above the frequency @xmath183 where these solutions disappear for @xmath181 .
the densities and the @xmath65 factors are computed from equations and respectively ( note the different scales for @xmath184 and @xmath185 ) .
we point out that for this particular family of solutions the ion density and the potential @xmath3 exhibit one hump at the center of the solution .
the physical meaning of the solitary waves with nvbc can be undestood with the aid of figure [ fig : grey : physical ] where we plotted electric field and electron momenta components .
the longitudinal electric field @xmath186 and momentum @xmath187 vanish outside the solitary wave that , as the plots of @xmath188 and @xmath189 show , can be intepreted as a localized modulation of an infinitely long circularly polarized electromagnetic wave ( @xmath190 and @xmath191 are not presented ) . even though our fluid description of the plasma can not take into account discrete particle effects like acceleration or heating , it is interesting to compute the maximum value of the electron and ion kinetic energy within the solitary wave @xmath192\end{aligned}\ ] ] for the previously discussed family , these quantities are plotted in figure [ fig : homo1](b ) and [ fig : homo1](c ) ( z - axis is in a logarithmic scale ) . the electron fluid can reach energies of the order of hundreds of mev and ions several tens of mev .
we also note that , for a fixed value of @xmath1 , the largest energy is reached at the lowest admissible @xmath0 value .
( color online ) a family of one - hump symmetric grey solitons . ] electric field and electron momenta at a given time for a grey soliton with @xmath193 and @xmath182 ] figure [ fig : homo2 ] corresponds to a different family of grey solitons . in this case both potentials present a minimum at the center of the soliton and the electron density has one central peak .
the panel [ fig : homo2](a ) shows the existence domain found by randomly varying the parameters @xmath0 and @xmath1 ( as we also did for figure [ fig : homo1 ] ) .
however , by fixing one of the parameters , using a small step for the other and initializing the algorithm with the solution obtained at the previous iteration , we were able to find solitary waves of the same type for parameter values outside the region indicated in [ fig : homo2](a ) ( see figure [ fig : homo2](d)- [ fig : homo2](f ) ) .
as the frequency decreases , the potential develops a cusp shape at the center of the wave and the algorithm requires an initial condition closer to the real solution to ensure convergence .
hence , the boundary exhibited in figure [ fig : homo2](a ) is a numerical artifact as we will also confirm in section [ sec : sub : coalescence ] .
we remark that potential profiles with a cusp shape have also been reported in solitary waves with vbc for parameter values close to the wavebreaking @xcite .
a family of one - hump symmetric grey solitons overlapping in the parameter space with the family of figure [ fig : homo1 ] ] a representative family of dark solitons , or heteroclinic connections @xmath194 , is shown in figure [ fig : hete1 ] .
note that its existence domain encompasses that of the grey solitons of figure [ fig : homo1 ] .
panels [ fig : hete1](d)-[fig : hete1](f ) display a particular solution of this family of dark solitons at the parameter values @xmath195 and @xmath181 .
the potential @xmath3 has one central hump and , as opposed to the grey solitons , the vector potential is an antisymmetric function .
the peak electron and ion kinetic energies are of the order of hundreds and tens of mev , respectively .
these three families of solitary waves provide some of the simplest examples of solitary solutions admitted by the system - . however , due to the fact that the fixed point is a saddle - focus and the system is hamiltonian and reversible , multi - hump and asymmetric solutions can exist too .
a few examples of grey multi - hump solutions with @xmath196 are displayed in figure [ fig : homo3](a)- [ fig : homo3](d ) .
panels [ fig : homo3](e ) and [ fig : homo3](f ) correspond to the density and @xmath65 factor of the solution with 5 humps .
the potential @xmath3 exhibits one central hump , while the vector potential has multiple humps .
these solutions are characterized by a cavity with a depression of the electron density in which an electromagnetic wave can be trapped . on the other hand ,
grey and dark asymmetric solitons are shown in figure [ fig : multi](a ) and [ fig : multi](b ) , respectively . for completeness ,
figure [ fig : multi](c ) also shows a multi - hump dark soliton ( heteroclinic connection ) .
solutions in panels [ fig : multi](a ) and [ fig : multi](c ) are plotted in panel [ fig : multi](d ) , where a projection on the phase space @xmath197 together with the fixed points @xmath198 is shown .
note that the asymmetric grey soliton connects @xmath199 , although it passes close to @xmath200 ( see the inset ) . some examples of symmetric multi - hump grey solitons .
] some examples of different types of solutions : ( a ) homoclinic asymmetric , ( b ) heteroclinic asymmetric and ( c ) heteroclinic with 2-humps .
panel ( d ) shows the projection of orbit ( a ) and ( b ) on the @xmath197 plane . ]
a striking feature of the families of solitary waves we identified is that they fill in regions with a well defined , within our numerical precision , boundary in the @xmath2 plane . as we have seen in section [ sec : fp : orbits ] , homoclinic and heteroclinic orbits lie on the interesection of stable and unstable manifolds of fixed points .
this suggests that certain families of orbits can cease to exist when the corresponding manifolds fail to intersect in the neighborhood of the solitary solution .
the mechanism responsible for loss of intersection in our case is known as _ coalescence _ @xcite and we now turn to its detailed description , through a numerical experiment involving the disapperance of a member of the family of grey solitons shown in figure [ fig : homo1 ] .
an approximation to the unstable manifold of @xmath135 can be computed by integrating equations - with initial condition @xmath201 = [ 0\ 0\ a_0\ 0]+\sigma\mathbf{v}_1 \ ] ] where @xmath202 is the real part of one of the unstable eigenvectors of @xmath135 , associated with an eigenvalue @xmath203 , @xmath204 . here
@xmath205 is a small parameter that controls the position of the initial condition on the plane spanned by the unstable eigenvectors .
computing one thousand outward spiraling orbits with @xmath206 $ ] , @xmath207 then results in an approximation of the unstable manifold .
the stable manifold was computed similarly , sprinkling initial conditions along the stable eigendirection and integrating backward in @xmath76 .
the stable and unstable manifolds are best visualized by keeping track of their intersection with a _ poincar ( surface of ) section _
@xmath208 , a @xmath126-dimensional surface in our @xmath125-dimensional phase space . here
we will be interested in symmetric homoclinic connections , which always intersect the symmetric section @xmath144 .
it will therefore be important to choose a poincar section that contains @xmath144 , for instance @xmath209 .
figure [ fig : intersection ] shows the first four intersections of our numerical approximation to the stable and unstable manifolds of @xmath135 with @xmath208 , for three different values of @xmath1 with fixed @xmath181 . for frequency value @xmath210 ( panel [ fig : intersection]a )
the stable and unstable manifolds intersect transversely and two homoclinic orbits ( grey solitons ) exist , marked by the points of intersection @xmath211 ( belonging to the homoclinic orbit in figure [ fig : homo1](d)- [ fig : homo1](f ) ) and @xmath212 ( belonging to the orbit in figure [ fig : intersection](d ) ) . at @xmath213 , @xmath211 and @xmath212
coalesce into a single solution and the manifolds become tangent . for @xmath214 the unstable and stable manifolds do not intersect and the family of homoclinic orbits shown in figure [ fig : homo1 ] ceases to exist for @xmath181 .
note however that different families of solitary waves do exist for @xmath214 , @xmath181 ; in figure [ fig : intersection ] we only plot a part of the stable and unstable manifolds and therefore more intersections can still take place .
this becomes apparent when comparing , for instance , figures [ fig : homo1 ] and [ fig : homo2 ] showing that a family of grey solitons can extent beyond the range of existence of another family .
a similar analysis was carried out for the second type of grey solitons , figure [ fig : homo2 ] .
we computed the stable and unstable manifolds for parameter values close to the the boundary exhibited in figure [ fig : homo2](a ) and we did not observed a tangency close to the homoclinic orbit , confirming that the existence boundary is a numerical artifact .
therefore , the poincar analysis is a useful tool to distinguish real boundaries from numerical artifacts .
note also that visualization of the stable and ustable manifolds of the fixed points on a poincar surface of section is a very effective way of generating initial guesses for the numerical computation of homoclinic and heteroclinic connections through the spectral method of section [ sec : algorithm ] .
panels ( a)-(c ) show the intersection of the stable ( black ) and the unstable ( grey ) manifolds with the poincar section @xmath209 .
panel ( d ) is the grey soliton corresponding to the point labelled @xmath212 . ]
even though an exhaustive analysis of the stability of the solitary waves is beyond the scope of the present work , we sketch out here some relevant aspects . as previously mentioned , the electromagnetic circularly polarized wave is susceptible to the relativistic raman and the modulational instabilities that would ultimately destroy the solitary wave .
however , since the growth rates of these instabilities are controlled by the amplitude of the electromagnetic wave ( @xmath17 in our dynamical system ) and the plasma density ( related to the parameter @xmath1 ) , certain parameter values could allow long distance propagation of the solitons . to test the stability of the solitary waves we show a couple of 1-dimensional simulations with the pic code calder @xcite .
this method does not prove stability of the solutions in a strict mathematical sense but it provides an insight into the dynamics .
we took a computational domain equal to @xmath215 ( in @xmath32 units ) that is large enough to assume periodic boundary conditions and we used one million of cells with ten particles per cell .
the code was initialized with a grey solitary wave of the type presented in figure [ fig : homo2 ] .
figure [ fig : stability1 ] shows the evolution of the electric field component @xmath186 of a grey solitary wave with @xmath216 and @xmath217 . for this high value of the frequency
the ions are almost immobile and the asymptotic value of the amplitude of the wave is @xmath218 .
it propagates undistorted during a few tens of @xmath33 , until the part of the solution with vanishing @xmath186 ( corresponding to the infinitely long circularly polarized wave ) becomes unstable due to the raman instability
. however , a solitary wave with a lower frequency value @xmath219 ( @xmath220 ) presents different dynamics ( see figure [ fig : stability2 ] ) . for this second case
the solitary wave develops an instability at its trailing edge and it radiates part of its energy away .
a similar instability appears in multi - hump solutions with vbc @xcite .
these examples reveal that , without collision and for a cold plasma model , our solutions present an unstable character
. however , a dissipation mechanism could reduce or even supppress the raman instability , thus allowing the propagation of the solitary waves for longer distances .
pic simulation initialized with a grey soliton of the type shown in figure [ fig : homo2 ] with @xmath221 . ]
pic simulation initialized with a grey soliton of the type shown in figure [ fig : homo2 ] with @xmath222 and @xmath223 . ]
solitary waves excited by the interaction of a high - intensity laser with a plasma have been observed in laboratory experiments @xcite and particle - in - cell ( pic ) simulations @xcite . although many theoretical works have been carried out on these structures ( see @xcite for a review ) , the parametric domain where the fixed point that controls the nvbc is a saddle - focus was unexplored . in our study of this regime
we were able to exhibit new ranges of solutions including grey and dark solitons .
we recall that , for vbc , the solutions are organized in the @xmath224 plane on a set of infinitely many branches @xcite .
each branch is characterized by the number of humps ( or number of nodes ) of the vector potential and it ends at a certain point due to the wavebreaking of the soliton .
the potential is always a symmetric function whereas the vector potential can be either symmetric or antisymmetric . for nvbc
we have shown that there is a continuum of solutions in the @xmath224 plane .
grey solitary waves with a symmetric potential and vector potential and dark solitary waves with symmetric potential and antisymmetric vector potential can be found for a wide range of parameters .
this is a natural extension of the vbc case with the number of nodes being even or odd , respectively .
multi - hump solutions are also possible for any value of parameters for which a single - hump orbit exists .
further , asymmetric single and multi - hump solutions exist , consistent with symmetry breaking in a conservative , time - reversible system .
in addition to being an important channel of laser - pulse energy transformation , solitary waves have been propossed as interesting candidates for photon and particle acceleration schemes @xcite .
pic simulations with an overcritical amplitude soliton showed electron acceleration during the nonlinear wavebreaking @xcite whereas ion acceleration has been detected during the postsoliton expansion @xcite . in @xcite ,
the authors reported wavebreaking of the solitary wave at the critical velocity determined by the end of the branch of the solutions in the @xmath224 plane .
it was estimated that ions could reach an energy value of the order of 70 mev .
similarly , in the present work we have presented domains of existence in the @xmath224 plane and we have shown that coalescence of solitary waves ( in parameter space ) leads to disappeance of families of solutions .
coalescence was visualized by keeping track of the stable and unstable manifolds of a fixed point by means of a poincar surface of section .
we point out that the stability of solitons with vbc has been studied in the past @xcite .
in particular , 1-dimensional numerical fluid simulations with immobile ions showed that single hump solutions are stable whereas the multi - hump solutions suffer the raman instability @xcite . on the other hand
, 2-dimensional simulations revealed that all solutions are unstable and the tranverse dynamics always dominates the longitudinal one @xcite .
our pic simulations initialized with a grey solitary wave showed that some of them could propagate undistorted during a few tens of @xmath33 , just before the circularly polarized wave suffers the raman instability .
however , other grey waves radiate a portion of their energy from the trailing edge , similarly to the multi - hump solutions with vbc @xcite .
since these are just a few examples , a complete stability analysis would requiry the study of other types of solutions ( dark waves , asymmetric , multi - humps etc ) in the whole @xmath224 plane . adding warm plasma effects or a collision term would be relevant too .
besides existence and stability , the question about how to excite solitary waves with nvbc remains open .
however , soliton - like electromagnetic modes with vbc have been observed during laser plasma interaction @xcite .
our preliminary pic simulations on solitary wave excitation show that the interaction between a solitary wave with vbc and a long circularly polarized laser pulse can produce a solitary wave with nvbc .
these simulations will be presented in a future work .
g. snchez - arriaga is supported by anr under the gospel project , grant reference anr-08-blan-0072 - 03 .
we thank d. bnisti and l. gremillet for helpful discussions .
let us write - as @xmath225 with @xmath226 $ ] .
fixed points are given by @xmath227 and the solutions of @xmath228a&=&0\label{eq : fp1}\\ v\left(\frac{\psi_e}{r_e}-\frac{\psi_i}{r_i}\right)&=&0\label{eq : fp2}\end{aligned}\ ] ] whereas the stability depends on the eigenvalue of the jacobian matrix @xmath229 the four eigenvalues of @xmath230 at the fixed points @xmath26 can be written as @xmath231 with @xmath232 @xmath233 ^ 2 -\frac{(1-v^2)(1-\epsilon^2)^2a_0 ^ 2}{(v\gamma_{e}\gamma_{i})^4}\ ] ] fixed point @xmath234 ( or @xmath235 ) is a saddle - focus if @xmath236 , a saddle - center if @xmath237 and a center if @xmath238 .
the conditions @xmath239 and @xmath240 yield the velocities and respectively .
the eigenvalue of the jacobian matrix at @xmath106 for @xmath241 are @xmath242 @xmath243^{3/2}}}\ ] ] and clearly the fixed point @xmath106 is a saddle - center ( note that @xmath244 ) . on the other hand , the existence and stability analysis of @xmath112 requires some auxiliar operations . from and assuming @xmath245 , one gets @xmath246 and by substituting in .
@xmath247 where we introduced the subscript @xmath248 to denote that we are dealing with a fixed point .
note that restriction together with @xmath249 in show that solutions of must lie on the intervals @xmath250 the solutions of can be discussed taking into account some properties of the function @xmath251 and its derivative : @xmath252(\psi_i-\epsilon\psi_e)}\right]\ ] ] in particular we are interested in the zeros of this derivative .
the factor inside the squared root vanishes at @xmath253 , that it is outside the physical domain . on the other hand ,
the roots of the term inside the braces are given by the zeros of the cubic equation @xmath254 for discussing the solutions of within the domain @xmath255 , we first note that @xmath256 , corresponding with the fixed points @xmath26 .
one also has the asymptotic behaviours @xmath257 as @xmath258 and @xmath259 ( @xmath260 ) for velocities less ( greater ) than @xmath261 ( see ) .
the analysis of its derivative shows that @xmath262 has 0 or 1 extreme for @xmath263 and therefore has one solution in this regime ( @xmath26 ) . on the other hand , for @xmath264
it always has one extreme , at say @xmath265 , with @xmath266 .
therefore , if @xmath267 , has two solutions and , in addition to @xmath26 , there is another fixed point that we call @xmath112 .
one also checks that @xmath112 and @xmath26 lie in different manifolds given by .
hence , heteroclinic connections among @xmath112 and @xmath26 are not possible .
10 url # 1#1urlprefix[2][]#2 s v , esirkepov t z , naumova n m _ et al .
_ 1999 _ phys . rev .
lett . _ * 82 * 34403443 d and bulanov s v 2002 bright and dark relativistic solitons in plasmas _ superstrong fields in plasmas _ ( _ american institute of physics conference series _ vol 611 ) ed m lontano , g mourou , o svelto , t tajima pp 151156 a ,
pakter r and rizzato f b 2007 nonlinear dynamics of electromagnetic pulses in cold relativistic plasmas _ proceedings of the 2007 ieee particle accelerator conference ( pac 07 )
. 25 - 29 jun 2007 , albuquerque , new mexico . | this article discusses the existence of solitary electromagnetic waves trapped in a self - generated langmuir wave and embedded in an infinitely long circularly polarized electromagnetic wave propagating through a plasma . from the mathematical point of view
they are exact solutions of the 1-dimensional relativistic cold fluid plasma model with nonvanishing boundary conditions . under the assumption of traveling wave solutions with velocity @xmath0 and vector potential frequency @xmath1
, the fluid model is reduced to a hamiltonian system .
the solitary waves are homoclinic ( grey solitons ) or heteroclinic ( dark solitons ) orbits to fixed points . by using a dynamical systems description of the hamiltonian system and a spectral method
, we identify a great variety of solitary waves , including asymmetric ones , discuss their disappearance for certain parameter values , and classify them according to : ( i ) grey or dark character , ( ii ) the number of humps of the vector potential envelope and ( iii ) their symmetries .
the solutions come in continuous families in the parametric @xmath2 plane and extend up to velocities that approach the speed of light .
the stability of certain types of grey solitary waves is investigated with the aid of particle - in - cell simulations that demonstrate their propagation for a few tens of the inverse of the plasma frequency . |
the effective one- or two - dimensional dirac equation describes surprising variety of physical systems .
it appears in effective models of spontaneous symmetry breaking in the quantum field theory @xcite , @xcite , @xcite , in the andreev approximation of the bogoliubov - de gennes equations of superconductivity @xcite , or in the effective description of the low - energy charge carriers in graphene and related carbon nanostructures @xcite . in all these scenarios ,
the explicit solution of dynamical equations is desirable .
however , it can not be found rather frequently due to analytically intractable form of interaction .
here , the importance of exactly solvable models comes out . despite their family
is rather limited , they can be adjusted to the required physical situation by variety of methods .
unitary transformations in quantum mechanics are convenient tool for making calculations more feasible , however , they are usually not considered to have much physical importance as the unitarily equivalent quantum systems provide the same physical predictions .
nonetheless , the unitary transformation can relate two systems whose physical interpretation is substantially different from each other . as an example , let us mention the free massless dirac particle on the real line whose hamiltonian reads @xmath0 .
the unitary transformation @xmath1 gives rise to the operator @xmath2 that can be interpreted as the hamiltonian of dirac fermion in carbon nanotube in presence of a scalar potential @xmath3 .
the relation between @xmath4 and @xmath5 has a nontrivial physical impact here ; it explains the absence of backscattering of dirac quasiparticles on impurities in carbon nanotubes @xcite .
two quantum systems related by unitary transformation have the same energy spectrum ; they are _
isospectral_. all the relevant properties of the transformed system can be effortlessly obtained from the initial model .
similar advantage is offered by the supersymmetric quantum mechanics , where the unitary equivalence is generalized by the intertwining relation @xmath6 and its conjugate . instead of unitarity
, the intertwining operator @xmath7 , known also as darboux transformation , satisfies @xmath8 where @xmath9 is a ( polynomial ) function of the initial hamiltonian @xcite .
this transformation was discussed in @xcite for the one - dimensional dirac operators where explicit formulas for @xmath5 satisfying the intertwining relations for a fixed @xmath4 and @xmath7 were provided .
the scalar potential is invariant with respect to the darboux transformation @xmath7 , i.e. it has the same form in both @xmath4 and @xmath5 .
the spectra of @xmath5 and @xmath4 are identical up to a finite number of energy levels ; the systems are _ almost isospectral_. the dirac hamiltonians @xmath4 and @xmath5 intertwined by @xmath7 describe different interactions in general .
for instance , @xmath5 can represent the hamiltonian of radially twisted carbon nanotubes whereas @xmath4 is associated with a twist - free nanotubes , see @xcite .
we aim to construct the new , solvable , one - dimensional dirac hamiltonian that would describe dirac fermions in graphene in presence of an inhomogeneous electromagnetic field .
the procedure should allow us to modify the scalar potential and to introduce or maintain a constant mass term .
the initial and the transformed hamiltonians do not need to be isospectral .
we are rather interested in the systems where the isospectrality of @xmath4 and @xmath5 ( @xmath10 ) is generalized by the _ spectral isomorphism _ , defined in terms of a real function @xmath11 ,
@xmath12 the construction will be divided into two independent steps .
the first step comprises of the special unitary transformation whereas the second one allows us to modify the spectrum of the operator globally , introducing spectrally isomorphic systems in a simple way .
dirac fermion in graphene is described by the dirac equation @xcite @xmath13+m \sigma_3+v \right)\psi = e\psi,\ ] ] where @xmath14 is the vector potential corresponding to the external magnetic field whereas @xmath15 corresponds to the pseudo - magnetic field induced by mechanical deformation @xcite .
the pseudo - spin distinguishes two triangular sublattices of the hexagonal lattice .
the mass term @xmath16 makes the two sublattices inequivalent ; it is absent in the suspended graphene but can appear when the graphene sheet is posed on a substrate , e.g. on the crystal of hexagonal boron - nitride @xcite .
we will focus on the settings with translational invariance in one direction .
we suppose that the pseudo - magnetic and magnetic field have nonvanishing @xmath17-component only , @xmath18 @xmath19 , @xmath20 , @xmath21 , @xmath22 . for convenience
, we will work with the rescaled coordinates @xmath23 , @xmath24 , where @xmath25 and @xmath26 are dimensionless quantities and @xmath27 has dimension of length ( it can be identified with the magnetic length @xmath28 for instance ) .
after the substitution and separation of variables @xmath29 , the equation ( [ h ] ) can be written as @xmath30 where @xmath31 @xmath32 , @xmath33 , and @xmath34 are dimensionless quantities .
we used suggestively the tilded notation in ( [ tildehh ] ) as we shall construct solvable hamiltonians of this type .
let us take a generic one - dimensional dirac operator @xmath4 , @xmath35 with real , sufficiently smooth , functions @xmath3 , @xmath36 and @xmath37 , where @xmath36 and @xmath37 do not vanish simultaneously .
we are looking for a `` gauge '' transformation @xmath38 that would change @xmath4 to @xmath5 of ( [ tildehh ] ) , @xmath39 we take @xmath38 as an element of the @xmath40 group , @xmath41 where @xmath42 , @xmath43 are real and continuous functions .
transformation of the kinetic term of @xmath4 results in the following expression @xmath44\partial_{x}+\dots.,\ ] ] where the dots represent the terms without derivative .
we require the coefficient at the derivative to be constant .
it can be satisfied by the following simple ansatz for @xmath38 , @xmath45 substituting this expression into ( [ tildeh ] ) , we get @xmath46 the mass term @xmath47 is required to be a non - negative constant .
it can be achieved by @xmath48 where the functions @xmath49 and @xmath50 are fixed as @xmath51 the functions ( [ gauge ] ) have to be real .
one can check , @xmath52 , @xmath53 . then
the argument of the square root in @xmath54 is real if and only if @xmath55 . the left - hand side is always positive .
whenever @xmath56 , the right - hand side is negative and the inequality is satisfied .
for @xmath57 , the inequality is satisfied again as it reduces to @xmath58 .
the same conclusion can be obtained for @xmath59 . ]
that this requirement is satisfied whenever @xmath60 .
besides , the gauge transformation @xmath38 has to be continuous .
otherwise , it would introduce discontinuities into the wave functions .
as @xmath61 and @xmath62 are continuous , we can analyze the continuity of @xmath49 and @xmath50 in the @xmath63-plane , taking @xmath61 and @xmath62 as two independent variables .
the function @xmath64 has discontinuity at @xmath65 for @xmath66 whereas @xmath67 has discontinuity at @xmath65 for @xmath68 , see fig.[fig01 ] for illustration .
hence , the continuous gauge transformation @xmath38 can be constructed as long as the values of @xmath61 and @xmath62 of the seed hamiltonian avoid one of the indicated discontinuities at least . substituting ( [ gauge ] ) into ( [ potentials ] ) , we get @xmath69 [ cols="^,^ " , ] addition of the mass term into
the two gap hamiltonian @xmath70 opens an additional central gap in the spectrum of @xmath71 that corresponds to the three - gap system , see fig .
[ lame1]a ) .
this is underlined by the fact that its potential term solves the corresponding nonlinear equation of the ablowitz - kaup - newell - segur ( akns ) hierarchy . defining @xmath72
, it solves @xmath73 @xmath74 where @xmath75 for @xmath76 .
let us notice that both the two - gap hamiltonian @xmath70 and the three - gap operator @xmath71 have a nontrivial integrals of motion represented by a higher - order differential operators with matrix coefficients .
their existence emerges as a consequence of ( [ akns ] ) , see e.g. @xcite for details on their generic construction .
the unitary transformation allows to construct dirac hamiltonian with the oscillating electromagnetic field .
the transformation ( [ gauge ] ) is well defined ( the functions @xmath49 and @xmath50 are continuous and real ) since we take @xmath77 and @xmath61 is bounded .
the hamiltonian @xmath5 has the following potential terms @xmath78 in dependence on the parameters @xmath79 , @xmath80 , and @xmath47 , the two potentials can acquire remarkable variety of profiles , see fig .
[ lame1]b)-d ) .
for instance , the vector potential can exhibit only mild oscillations around a non - zero constant that are negligible in comparison to the amplitude of the electrostatic potential , see fig .
[ lame1]b ) . this system can be seen as an approximation of the setting where dirac particle travels in the @xmath81 periodic array of electrostatic barriers with the momentum @xmath82 parallel to the barriers .
the small , periodic fluctuations of @xmath83 in our model facilitate solution of the stationary equation that would nt be analytically solvable for constant @xmath84 .
when @xmath85 , the reflection symmetry of the scalar potential @xmath86 is spoiled , see fig .
[ lame1]d ) .
instead , the potential seems to be odd with respect to the reflection within the finite intervals determined by two consecutive extrema of the potential .
in the article , we focused on the construction of solvable dirac operators with inhomogeneous vector and scalar potential , and constant ( possibly vanishing ) mass .
the core of our results represents the unitary transformation ( [ g ] ) , ( [ gauge ] ) that brings a generic one - dimensional dirac operator into the required form .
we illustrated its application on the special class of solvable models that are based on the mass - less , solvable dirac operator with pure vector potential .
we showed that solvability of these systems is not compromised by addition of the mass term , yet the spectrum changes globally and gives rise to the spectral isomorphism ( [ ee ] ) between the systems .
the constructed models were used in description of dirac fermions in graphene in presence of an inhomogeneous electromagnetic field .
we presented solvable models where the particle experiences localized as well as periodic electromagnetic barrier .
the two settings with localized barrier proved to be good approximation for magnetic field induced either on the edge of the magnetized film ( [ morse ] ) or by a magnetized strip of finite width ( [ scarfspec ] ) . in case of the periodic array of barriers ,
the system with two gaps in the spectrum served as the initial `` seed '' system .
we constructed a spectrally isomorphic hamiltonian with three spectral gaps and discussed some its peculiar properties , e.g. its relation to the akns hierarchy of integrable systems ( [ akns ] ) .
the presented procedure is complementary to the existing methods , namely to the darboux transformation @xcite .
the latter one can transform an initial dirac operator into an ( almost ) isospectral hamiltonian with altered vector potential and mass term .
the presented unitary transformation can set the mass to a constant and induce an inhomogeneous scalar potential .
it can provide the new isospectral partners of the recently discussed dirac systems with reflection - less potential @xcite , @xcite or inhomogeneous magnetic fields @xcite , @xcite .
it is worth noticing that a different approach to construction of solvable systems described by dirac equation was developed in @xcite .
there , a transformation involving analytic continuation of parameters was discussed that allowed to get new solvable dirac operators from the known ones with , in specific manner interchanged , vector and scalar potentials .
it would be interesting to consider spectral isomorphism for generic finite - gap systems .
in particular , it should be understood how the procedure , which maps an @xmath87gap hamiltonian to an @xmath88-gap one , is reflected by the associated equations of the akns hierarchy , or manifested in the structure of ( lax ) integrals of motion associated with the finite - gap hamiltonians . in this context
, the results on of the hidden nonlinear supersymmetry of the finite - gap systems could be particularly useful @xcite . as the operator @xmath89 represents a schrdinger operator with matrix potential ( the antidiagonal terms with first derivative
can be transformed out be an additional unitary transformation ) , the presented procedure can be beneficial for analysis of these nonrelativistic systems that are of growing interest recently @xcite , @xcite , @xcite .
the presented construction was primarily aimed at derivation of explicit exactly solvable models .
however , it can be utilized for qualitative analysis as well .
the hamiltonians @xmath71 were discussed qualitatively in @xcite , where a set of sufficient conditions for existence of bound states was derived . with the use of the unitary transformation ( [ gauge ] ) , those qualitative results are directly applicable on a class of operators @xmath5 with inhomogeneous vector and scalar potential .
a. v. sokolov , phys .
a * 377 * , 655 ( 2013 ) ; + a. a. andrianov , a. v. sokolov , `` minimal realizations of supersymmetry for matrix hamiltonians , '' arxiv:1409.6695 v. jakubsk , d. krejik , annals of physics * 349 * , 268 ( 2014 ) . | we construct the new one - dimensional dirac hamiltonians that are spectrally isomorphic ( not isospectral ) with the known exactly solvable models .
explicit formulas for their spectra and eigenstates are provided .
the operators are utilized for description of dirac fermions in graphene in presence of an inhomogeneous electromagnetic field .
we discuss explicit , physically relevant , examples of spectrally isomorphic systems with both non - periodic and periodic electromagnetic barriers . in the latter case ,
spectrally isomorphic two- and three - gap systems associated with the ablowitz - kaup - newell - segur hierarchy are considered . |
multi - dimensional shift is an important and a highly active area of ongoing research in dynamical system .
it is also closely related to lattice models in the scientific modeling of spatial structure .
relevant investigations have been performed on phase transitions and chemical reactions @xcite , biology @xcite and image processing and pattern recognition @xcite .
lattice models can be better understood if multi - dimensional shifts of finite type are understood .
the most interesting properties of shifts include entropy and various mixing properties , such as topological mixing and strong specification ( or strong irreducibility ) .
these properties enjoy many of the important properties of dynamical systems @xcite .
however , determining whether a given system exhibits topological mixing or strong specification in multi - dimensions is not easy .
the intrinsic difficulty is related to undecidability of multi - dimensional coloring problem @xcite .
nevertheless , this study provides some easily checked sufficient conditions for topological mixing and strong specification of two - dimensional shifts of finite type .
let @xmath0 be the two - dimensional planar lattice .
vertex ( or corner ) coloring is considered first . for any @xmath1 and @xmath2 , the @xmath3 rectangular lattice with the left - bottom vertex @xmath4 is denoted by @xmath5 in particular , @xmath6 let @xmath7 be a set of @xmath8 ( @xmath9 ) colors or symbols . for @xmath1 , @xmath10 is the set of @xmath3 local patterns .
let @xmath11 be a basic set of allowable local patterns .
for any lattice @xmath12 , the set of all @xmath13-admissible patterns on @xmath14 is defined as @xmath15 denote @xmath16 for @xmath17 .
@xmath18 is the set of all global patterns that can be constructed from the local patterns in @xmath13 .
notably , any @xmath0-shift of finite type can be represented by some @xmath11 for some @xmath19 @xcite .
hence , only the case of @xmath11 , @xmath19 , is considered here .
first , topological mixing is introduced . for any shift @xmath20 and any subset @xmath21 ,
@xmath22 is the restriction map .
denote by @xmath23 the euclidean metric on @xmath0 .
a @xmath0 shift @xmath20 is topologically mixing ( mixing , for short ) if for any finite subsets @xmath24 and @xmath25 of @xmath0 , a constant @xmath26 exists such that for all @xmath27 with @xmath28 , and for any two allowable patterns @xmath29 and @xmath30 , there exists a global pattern @xmath31 with @xmath32 and @xmath33 ; see @xcite . on the other hand , @xmath20 has strong specification if a number @xmath34 exists such that for any two allowable patterns @xmath29 and @xmath35 with @xmath36 , where @xmath37 are subsets of @xmath0 , there exists a global pattern @xmath31 with @xmath32 and @xmath38 ; see @xcite .
clearly , strong specification implies topological mixing .
very few results that verify that @xmath39 is mixing or has strong specification are known @xcite .
previously , in studying pattern generation problems @xcite , the authors introduced connecting operators to study the entropy of @xmath39 . in this paper , connecting operators are also used to provide sufficient conditions for the mixing of or strong specification of @xmath39 . first , topological mixing is considered .
given two patterns @xmath40 and @xmath41 defined on @xmath24 and @xmath42 respectively , in general , @xmath24 and @xmath42 are not located horizontally or vertically .
typically , the gluing process is decomposed into three steps , as presented in fig .
1.1 . for clarity , in fig 1.1 ,
the patterns @xmath43 s are presented and the underlying lattices @xmath14 s are omitted . 1 .
extend @xmath41 to @xmath44 such that @xmath40 can connect @xmath44 horizontally .
the combined pattern becomes an @xmath45-shaped pattern @xmath46 .
2 . extend the @xmath45-shaped pattern to a rectangular pattern .
3 . extend the rectangular pattern to a global pattern on @xmath0 .
@xmath47{fig1_1.eps}\ ] ] @xmath48 to ensure that all processes are executable , the following sufficient conditions are proposed in each step : 1 .
the primitivity of horizontal transition matrices @xmath49 and vertical transition matrices @xmath50 , for each @xmath51 .
2 . the corner - extendability for @xmath45-shaped lattices .
3 . the rectangle - extendability to the @xmath0-plane . notably , a matrix @xmath52 is primitive ( or @xmath53-primitive ) if there exists @xmath54 such that @xmath55 for all @xmath56 ; here , @xmath55 means that each entry of @xmath57 is positive except in positions of @xmath52 where a zero row or zero column is present . to find finitely checkable sufficient conditions of ( i ) , two kinds of sufficient conditions for the primitivity of @xmath49 and @xmath50 are introduced . 1 . invariant diagonal cycles , 2 . primitive commutative cycles .
both of these conditions are applied to construct the primitive diagonal submatrices of @xmath58 for some @xmath59 , @xmath51 ; then , they are used to show that @xmath49 is primitive .
furthermore , when either condition applies , only finitely many @xmath49 have to be checked to ensure that @xmath49 is primitive for all @xmath51 .
after the primitivity of @xmath49 and @xmath50 is established , the corner - extendable conditions @xmath60 and crisscross - extendability are introduced ( definitions [ definition:3.2 ] and [ definition:3.11 ] ) to extend the l - shaped pattern and the rectangular pattern into a global pattern , and then to establish that @xmath39 is mixing .
the main theorem is given by theorem [ theorem:3.14 ] .
[ theorem:1.1 ] if 1 .
@xmath61 is crisscross - extendable , 2 .
@xmath13 satisfies three of corner - extendable conditions @xmath62 , @xmath63 , then @xmath64 and @xmath65 are primitive for all @xmath51 if and only if @xmath39 is mixing . next ,
strong specification is considered .
since strong specification is stronger than topological mixing .
apart from the processes in fig .
1.1 which concern the situation in which regions @xmath24 and @xmath42 are far away , the case in which one pattern is enclosed in another pattern , as in fig .
1.2 , must be studied .
@xmath66{fig1_2.eps}\ ] ] @xmath67 notably , in studying topological mixing , fig .
1.2 can be reduced to fig . 1.1 .
however , in studying strong specification , @xmath40 and @xmath41 can not be removed since the relative positions of @xmath24 and @xmath25 are fixed .
now , the gluing of @xmath40 and @xmath41 can be completed by the following two processes . 1 .
extend @xmath40 horizontally and vertically to form a crisscross pattern that touches @xmath41 , as presented in fig .
fill the holes that are surrounded by the rectangularly annular lattice to form a rectangular pattern , as presented in fig .
@xmath68{fig1_3.eps } & & & & & \psfrag{a}{{\small $ ( 5)$ } } \includegraphics[scale=1.35]{fig1_4.eps } \\ \text{figure 1.3 . } & & & & & \text{figure 1.4 . }
\end{array}\ ] ] then , repeat step @xmath69 and extend the rectangular pattern to a global pattern on @xmath0 . the hole - filling condition ( hfc ) in step ( 5 ) is finitely checkable : any hole of size @xmath70 that is surrounded by any admissible annular lattice with width @xmath71 can be filled by admissible local patterns and forms an @xmath72 pattern .
moreover , @xmath73 hole - filling condition ( hfc@xmath74 , @xmath75 , with ( hfc@xmath76hfc is introduced ( see definition [ definition:6.1 ] ) , which is weaker than hfc and is also finitely checkable .
( hfc@xmath74 is closely related to the extension property called square filling @xcite . the main theorem for strong specification
is given by theorem [ theorem:6.4 ] .
[ theorem:1.2 ] given @xmath11 , if there exists @xmath75 such that 1 .
@xmath13 is @xmath73 crisscross - extendable , 2 .
@xmath13 satisfies @xmath77hfc@xmath74 with size @xmath70 for some @xmath78 , 3 .
@xmath79 is @xmath80-primitive and @xmath81 is @xmath82-primitive , then @xmath39 has strong specification .
theorems [ theorem:1.1 ] and [ theorem:1.2 ] are very powerful in verifying mixing properties . the most known results for strong specification and topological mixing in the literature can be checked successfully by them .
their significance in classification of mixing properties is discussed as follows .
previously , boyle et al . @xcite discussed various mixing properties , including strong irreducibility , uniform filling property , corner gluing , block gluing and topological mixing ( see definition [ definition : a.3 ] ) .
the range of ( hfc@xmath74 and these mixing properties for @xmath0 shifts of finite type are listed in figure 1.5 . for brevity ,
we use the following notations . 1 . :
@xmath13 satisfies ( hfc@xmath74 ( i.e. , condition ( ii ) of theorem [ theorem:1.2 ] ) and conditions ( i ) and ( iii ) of theorem [ theorem:1.2 ] , 2 .
: @xmath39 has strong specification , 3 .
: @xmath39 has the ufp , 4 .
: @xmath39 is corner gluing , 5 .
: @xmath39 is block gluing , 6 .
: @xmath39 is topologically mixing .
@xmath83{range.eps}\ ] ] @xmath84 notably , the solid line indicates that the inner property is strictly stronger than the outer one ; the dotted line indicates that inner property is stronger than or equivalent to the outer one .
example [ example:7.8 ] in this paper is an example for @xmath85 , and the examples for @xmath86 and @xmath87 were given by boyle et al .
it is still unsolved whether or not @xmath88 ; see @xcite .
note that strong specification , periodic specification and strong irreducibility are all equivalent ; see remark [ remark:3.14 - 0 ] . in viewing fig .
1.5 , theorem [ theorem:1.1 ] and primitive results in section 4 ensure the weakest case topological mixing holds .
theorem [ theorem:1.2 ] ensures the strongest case strong specification holds . in studying both topological mixing and strong specification , the transition matrices @xmath49 and @xmath50 and the connecting operator @xmath89 or @xmath90 , introduced in section 2 , are extensively used .
indeed , invariant diagonal cycles , primitive commutative cycles and ( hfc@xmath74 can be expressed in terms of transition matrices and connecting operators as the finitely checkable sufficient conditions . all cases with certain extendability conditions can be verified by using transition matrices and connecting operators except strong specification . in conclusion , the mixing properties in fig .
1.5 with certain extendable conditions can be expressed in terms of transition matrices and connecting operators except strong specification .
the related theorems are listed in table 1.1 . on he other hand , the extendability problem is known to be undecidable @xcite . presumably , the gluing problem is also undecidable .
therefore , it is not possible to obtain a necessary and sufficient condition which are finitely checkable for topological mixing and strong specification .
however , the gluing problems in this paper are decomposed into primitivity problems of transition matrices of @xmath91 and @xmath92 , and the extending problems of @xmath93 and @xmath94 .
although extending problem ( 3 ) is generally undecidable , the proposed finitely checkable sufficient conditions ensure that is answered affirmatively under our situation .
@xmath95{table.eps}\ ] ] @xmath96 in many physical problems , edge coloring is very common . the results of vertex coloring can easily be extended to edge coloring .
in particular , the six - vertex and eight - vertex ice models in statistical physics can be shown to exhibit topological mixing and strong specification , respectively .
the other related cases can be treated analogously @xcite .
the rest of this paper is organized as follows .
section 2 introduces ordering matrices of local patterns , transition matrices and connecting operators .
section 3 introduces corner - extendable conditions and crisscross - extendability to study rectangle - extendability and mixing .
section 4 introduces invariant diagonal cycles and primitive commutative cycles to establish sufficient conditions for the primitivity of @xmath49 or @xmath50 .
section 5 introduces @xmath73 hole - filling condition to develop finitely checkable conditions for strong specification .
section 6 presents the theory of edge coloring . in appendix , we present the expressions of ( hfc@xmath74 , uniform filling property and corner gluing with rectangle - extendability by @xmath49 and @xmath89 .
this section reviews the essential aspects of the ordering matrices of local patterns and their associated transition matrices @xcite . then , connecting operators are introduced .
this study depends on more precise properties of connecting operators which were only outlined in the previous paper @xcite .
most of the proofs can be obtained by the arguments similar to those in @xcite and are omitted . as presented elsewhere @xcite , with
@xmath19 fixed , the ordering matrices @xmath97 and @xmath98 are introduced to arrange systematically all local patterns in @xmath99 and @xmath100 , respectively .
since the vertex coloring and face coloring are equivalent on @xmath0 , in the following , the colors of patterns are drawn on faces instead of on vertices .
for an @xmath101-sequence @xmath102 with @xmath103 , @xmath104 , @xmath105 is assigned the number by the @xmath101-th order counting function @xmath106 : @xmath107 the horizontal and vertical ordering matrices @xmath108_{p^{2}\times p^{2}}$ ] and @xmath109_{p^{2}\times p^{2}}$ ] are defined by @xmath110{x_ij.eps } \end{array } & \text{and } & \begin{array}{c } \psfrag{a}{$y_{i_{2},j_{2}}=$ } \psfrag{b}{{\footnotesize $ u'_{0,1}$ } } \psfrag{c}{{\footnotesize $ u'_{1,1}$ } } \psfrag{d}{{\footnotesize $ u'_{0,0}$ } } \psfrag{e}{{\footnotesize $ u'_{1,0}$ } } \psfrag{f } { , } \includegraphics[scale=1.2]{x_ij.eps } \end{array } \end{array}\ ] ] where @xmath111 , @xmath112 , with @xmath113 for instance , when @xmath114 , @xmath115{x_2.eps } \end{array } & \text{and } & \begin{array}{c } \psfrag{c}{$\mathbf{y}_{2}=$ } \psfrag{d}{. } \includegraphics[scale=0.475]{y_2.eps } \end{array } \end{array}\ ] ] now , @xmath116 and @xmath117 are closely related to each other as follows . @xmath118,\ ] ] where @xmath119;\ ] ] the case for @xmath117 is similar .
the higher - order ordering matrices @xmath120_{p^{n}\times p^{n}}$ ] of @xmath99 , @xmath121 , are defined recursively as @xmath122,\ ] ] where @xmath123\ ] ] is a @xmath124 matrix .
notably , the entry @xmath125 is the @xmath126 local pattern @xmath127 with @xmath128 similarly , the higher - order ordering matrix @xmath98 can be defined recursively , as above . given a basic set @xmath11 , the horizontal and vertical transition matrices @xmath129_{p^{2}\times p^{2}}$ ] and @xmath130_{p^{2}\times p^{2}}$ ] are given by @xmath131 according to ( [ eqn:2.5 ] ) and ( [ eqn:2.6 ] ) , the higher - order transition matrices @xmath49
, @xmath121 , can be defined as @xmath132,\ ] ] where @xmath133\ ] ] is a @xmath124 zero - one matrix .
furthermore , for any @xmath51 and @xmath134 , @xmath135 are decomposed by applying ( [ eqn:2.9 ] ) @xmath136 times , as follows . for any @xmath134 and @xmath137 ,
define @xmath138 .
\end{array}\ ] ] therefore , for any @xmath139 , @xmath135 can be represented as a @xmath140 matrix @xmath141_{p^{q+1}\times p^{q+1}}=\left[h_{n+q;\beta_{1};\beta_{2};\cdots ; \beta_{q+1}}\right].\ ] ] in particular , when @xmath142 , @xmath143.\ ] ] more precisely , the relation between @xmath144 and @xmath145 is as follows . given @xmath145 , choose @xmath146 , @xmath147 , such that @xmath148 for @xmath147 , let @xmath149 then , @xmath150 from ( [ eqn:2.10 ] ) , for @xmath134 , @xmath151 can be verified .
hence , let @xmath152_{p^{q+1}\times p^{q+1}}$ ] ; by ( [ eqn:2.11 ] ) and ( [ eqn:2.16 ] ) , @xmath153 where @xmath154 .
actually , we have @xmath155 hence , @xmath156 ( or @xmath157 ) depends only on @xmath158 ( or @xmath159 ) . before showing the formula for reducing @xmath135 to @xmath49 ,
two products of matrices are introduced as follows .
for any two matrices @xmath160 $ ] and @xmath161 $ ] , the kronecker product ( tensor product ) of @xmath162 is defined by @xmath163.\ ] ] next , for any two @xmath164 matrices @xmath165 $ ] and @xmath166 $ ] , where @xmath167 and @xmath168 are numbers or matrices , the hadamard product of @xmath169 is defined by @xmath170,\ ] ] where the product @xmath171 of @xmath167 and @xmath168 may be a multiplication between numbers , between numbers and matrices or between matrices whenever it is well - defined .
now , from ( [ eqn:2.17 ] ) , high - order transition matrices @xmath135 can be reduced to lower order transition matrices @xmath49 as follows .
[ proposition:2.0 ] for any @xmath51 and @xmath134 , @xmath172_{p\times p } \right),\ ] ] where @xmath173 is the @xmath174 full matrix .
notably , the results for @xmath49 are also valid for @xmath50 , as easily determined by exchanging the terms @xmath49 and @xmath50 .
therefore , for simplicity , only the results for @xmath49 are presented herein . in the following
, @xmath175 can be expressed in terms of @xmath176 and connecting operator @xmath90 .
this result is crucial in establishing finitely checkable conditions for the primitivity of all @xmath79 , @xmath75 . from ( [ eqn:2.12 ] ) , for @xmath177 ,
the elementary pattern of @xmath176 is @xmath178 where @xmath179 , @xmath180 .
let @xmath181 where @xmath182 from ( [ eqn:2.2 ] ) and ( [ eqn:2.6 ] ) , the entry @xmath183 is equal to the cardinal number of the set of all @xmath184 @xmath13-admissible local patterns @xmath185 with @xmath186 clearly , @xmath187 and @xmath188 .
therefore , for @xmath177 , @xmath189,\ ] ] where @xmath190 furthermore , denote by @xmath191 a @xmath192 column - vector that consists of all @xmath193 in @xmath194 , which is very useful in deriving the reduction formula .
now , the connecting operator @xmath195 $ ] that was introduced in @xcite is recalled .
first , the connecting ordering matrix @xmath196 $ ] , a different arrangement for @xmath197 from @xmath198 , is introduced .
@xmath196_{p^{2}\times p^{2}}$ ] , where @xmath199 is a @xmath200 matrix of local patterns , is defined as follows . with fixed @xmath201 , for @xmath202 , @xmath203{c_m_ij_st.eps } \end{array}\ ] ] with @xmath204 , @xmath205 , @xmath206 and @xmath207 .
now , @xmath208 can be obtained in terms of @xmath209 as follows .
[ proposition:2.1 ] let @xmath210_{p^{2}\times p^{2}}$ ] . for any @xmath177 and @xmath201 , @xmath211.\ ] ] the matrix multiplication of @xmath199 and @xmath212 can not connect local patterns in the vertical direction
however , @xmath213 does so . by changing the index of @xmath196_{p^{2}\times p^{2}}$ ] , the ordering matrix @xmath214_{p^{2}\times p^{2}}$ ]
is defined by @xmath215 where @xmath216 , @xmath217 , satisfying @xmath218 and @xmath219 . indeed , for @xmath202 , @xmath220{c_m_ij_st.eps } \end{array}\ ] ] with @xmath221 , @xmath222 , @xmath206 and @xmath207 . from ( [ eqn:2.25 ] )
, the matrix multiplication of @xmath213 and @xmath223 represents the vertical connection of the patterns on @xmath224 .
now , given @xmath11 , for @xmath177 , the connecting operator @xmath195_{1\leq i , j\leq p^{2}}$ ] of @xmath196_{1\leq i , j\leq p^{2}}$ ] is defined as follows . for @xmath202 , @xmath225 in the following
, @xmath226 can be obtained explicitly . for @xmath227_{p^{2}\times p^{2}}$ ] , define @xmath228,\ ] ] where @xmath229\ ] ] for @xmath230 .
then , for @xmath201 , @xmath231 is a @xmath232 zero - one matrix . by proposition [ proposition:2.1 ]
, the connecting operator @xmath233 can also be obtained from @xmath90 . for @xmath177 , @xmath234_{1\leq i , j\leq p^{2}}$ ]
satisfies @xmath235.\ ] ] from ( [ eqn:2.24 ] ) , @xmath236_{p^{2}\times p^{2}}$ ] is defined by @xmath237 where @xmath238 such that @xmath218 and @xmath219 .
now , the relation between @xmath239 and @xmath176 is elucidated as follows .
since the sizes of @xmath240 and @xmath241 are different , the elementary pattern @xmath240 can be reduced further as follows .
let @xmath242\ ] ] and @xmath243 as ( [ eqn:2.21 ] ) , the entry @xmath244 can be verified to be the number of the set of all @xmath245 @xmath13-admissible local patterns @xmath246 with @xmath247 in the following proposition , @xmath248 can be obtained as the product of @xmath249 and @xmath250 @xcite , i.e. , @xmath249 reduces @xmath239 to @xmath176 .
[ proposition:2.2 ] for any @xmath17 , @xmath251 furthermore , for @xmath252 , let @xmath253,\ ] ] then @xmath254 furthermore , for @xmath255 , @xmath256-many @xmath249 can reduce @xmath175 to @xmath176 as follow .
for any positive integer @xmath255 , the elementary patterns of @xmath175 can be decomposed by applying ( [ eqn:2.28 ] ) @xmath256 times .
indeed , for @xmath255 and @xmath257 , define @xmath258 .
\end{array}\ ] ] therefore , for any @xmath134 , @xmath175 can be represented as a @xmath140 matrix @xmath259_{p^{q+1}\times p^{q+1}}=\left[h_{m , n+q;\beta_{1};\beta_{2};\cdots ; \beta_{q+1}}\right]\ ] ] where @xmath260 is a @xmath124 matrix .
notably , the relation between @xmath144 and @xmath145 is the same as ( [ eqn:2.13 ] ) and ( [ eqn:2.14 ] ) .
define @xmath261 as in proposition [ proposition:2.2 ] , the elementary patterns of @xmath175 can be expressed as the product of @xmath256-many @xmath249 and the elementary patterns of @xmath176 .
[ proposition:2.3 ] for any @xmath17 and @xmath262 , @xmath263 where @xmath264 , @xmath265 .
moreover , @xmath266 similarly , for @xmath267 , the connecting operators are denoted by @xmath268 $ ] ( corresponding to @xmath195 $ ] for @xmath269 ) and @xmath270 $ ] ( corresponding to @xmath236 $ ] for @xmath269 ) .
the arguments that hold for @xmath49 are also valid for @xmath50 .
this section investigates extendability and mixing of @xmath39 .
it is clear that primitivity of @xmath49 and @xmath50 for all @xmath51 can be interpreted as topological mixing in horizontal and vertical directions respectively . in general , the relative position of lattices @xmath24 and @xmath42 are not located horizontally or vertically . accordingly , mixing in the directions other than horizontal and vertical directions need to be studied .
to treat these situations , rectangle - extendability , corner - extendable conditions and crisscross - extendability are introduced .
first , the rectangle - extendability of @xmath13 is defined as follows .
[ definition:3.1 ] for @xmath11 , @xmath13 is called rectangle - extendable if for every pattern @xmath271 , @xmath17 , there exists @xmath272 such that @xmath273 .
previously , the importance of corners of lattices has been noticed @xcite .
indeed , in studying mixing properties , the concept of corner gluing was introduced by boyle et al .
similarly , for studying rectangle - extendability and mixing , the corners of the rectangular lattice need to be studied closely .
indeed , let the @xmath45-shaped lattices @xmath274 , @xmath275 , @xmath276 and @xmath277 , that is , @xmath278{c1.eps } , & \psfrag{a}{$\mathbb{l}_{2}=$ } \includegraphics[scale=0.6]{c2.eps } , & \psfrag{a}{$\mathbb{l}_{3}=$ } \includegraphics[scale=0.6]{c3.eps } , & \psfrag{a}{$\mathbb{l}_{4}=$ } \includegraphics[scale=0.6]{c4.eps } \end{array}.\ ] ] @xmath279 [ definition:3.2 ] let @xmath11 . for @xmath63 , @xmath13 satisfies corner - extendable condition @xmath62 if for any @xmath280 , there exists @xmath281 such that @xmath282 . whether or not @xmath283 or @xmath284 contains a zero row or a zero column are very much different in studying mixing problem .
we begin with the study when there is no zero row and column .
we need the following notation .
[ definition:3.3 ] a matrix @xmath160_{n\times n}$ ] is called compressible if contains a zero row or a zero column .
a matrix is non - compressible if it is not compressible .
an @xmath269 ( or @xmath267 ) is degenerated if @xmath285 ( or @xmath286 ) is compressible for some @xmath187 .
an @xmath269 ( or @xmath267 ) is non - degenerated if it is not degenerated .
first , consider the case for @xmath61 when @xmath283 and @xmath284 are non - degenerated .
clearly , if both @xmath52 and @xmath287 are non - negative and non - compressible matrices , then @xmath288 is non - compressible . from ( [ eqn:2.9 ] ) , ( [ eqn:2.10 ] ) , ( [ eqn:2.12 ] ) and ( [ eqn:2.20 ] ) , the following result is easily obtained , and the similar result for @xmath267 also holds .
[ proposition:3.4 ] if @xmath269 is non - degenerated , then @xmath289 are non - compressible for @xmath51 and @xmath230 . in particular , @xmath49 is non - compressible for all @xmath51 .
moreover , @xmath193 are also non - compressible for @xmath290 , @xmath230 and @xmath188 .
the following lemma is easily proven and the proof is omitted .
[ lemma:3.5 ] given @xmath11 , if @xmath64 and @xmath65 are non - compressible for all @xmath51 , then @xmath13 is rectangle - extendable .
the non - degeneracy of @xmath283 and @xmath284 implies rectangle - extendability and , moreover , three of the corner - extendable conditions , as follows .
[ theorem:3.6 ] given @xmath61 , if @xmath283 and @xmath284 are non - degenerated , then 1 .
@xmath13 is rectangle - extendable , 2 .
@xmath13 satisfies @xmath291 , @xmath292 and @xmath293 .
\(i ) is obtained directly from proposition [ proposition:3.4 ] and lemma [ lemma:3.5 ] .
\(ii ) since @xmath283 is non - degenerated . from ( [ eqn:2.2 ] ) , for any @xmath294 , there exist @xmath295 such that @xmath296{z22_1.eps } & \text{and } & \psfrag{b}{$b$ } \psfrag{c}{{\footnotesize $ u_{1,1}$ } } \psfrag{d}{{\footnotesize $ u_{0,0}$ } } \psfrag{e}{{\footnotesize $ u_{1,0}$ } } \includegraphics[scale=1.2]{z22_2.eps } \end{array}\ ] ] @xmath297 are in @xmath13 , which implies that conditions @xmath291 and @xmath292 are satisfied . similarly , that @xmath284 is non - degenerated implies that @xmath13 satisfies conditions @xmath291 and @xmath293 .
the proof is complete .
now , @xmath39 is mixing follows from the non - degeneracy of @xmath283 and @xmath284 and primitivity of @xmath49 and @xmath50 , @xmath51 .
[ theorem:3.7 ] given @xmath61 , if @xmath283 and @xmath284 are non - degenerated , then the following statements are equivalent .
@xmath64 and @xmath65 are primitive for all @xmath51 .
@xmath39 is mixing .
( i)@xmath298(ii ) .
let @xmath24 and @xmath25 be finite sublattices of @xmath0 . then , there exist @xmath299 and @xmath300,@xmath301 such that @xmath302 , @xmath303 . from ( i ) , there exists @xmath304 such that @xmath305 and @xmath306
then , take @xmath307 .
let @xmath308 with @xmath28 and any two allowable patterns @xmath309 and @xmath310 .
clearly , @xmath40 and @xmath41 can be extended as @xmath311 on @xmath312 and @xmath313 on @xmath314 by using the local patterns in @xmath13 , respectively .
it is not difficult to prove that @xmath311 and @xmath313 can be connected as the l - shaped pattern @xmath315 by using the local patterns in @xmath13 , as follows .
@xmath316{l1.eps } & \text{or } & \psfrag{a}{{\footnotesize$n$ } } \psfrag{b}{$u_{l}$ } \includegraphics[scale=0.65]{l2.eps } \end{array}\ ] ] @xmath317 notably , the l - shaped lattices may degenerate into rectangular lattices .
since @xmath283 and @xmath284 are non - degenerated , by theorem [ theorem:3.6 ] , @xmath13 satisfies conditions @xmath291 and @xmath292 .
then , @xmath315 can be extended as @xmath318 on the rectangular lattice by using the local patterns in @xmath13 , which is obtained by filling the corner of the l - shaped lattices . from theorem [ theorem:3.6 ]
, @xmath13 is rectangle - extendable .
then , @xmath318 can be extended as @xmath272 with @xmath32 and @xmath33 .
therefore , @xmath39 is mixing .
( ii)@xmath298(i ) . from proposition
[ proposition:3.4 ] , @xmath49 and @xmath50 are non - compressible for all @xmath51 .
then , for @xmath51 , any pattern in @xmath319 or @xmath320 can be extended to @xmath0 by using the local patterns in @xmath13 .
it can be easily verified that ( ii)@xmath298(i ) ; the details are omitted .
the proof is complete .
next , consider the case for @xmath61 when @xmath283 or @xmath284 is degenerated .
theorems [ theorem:3.6 ] and [ theorem:3.7 ] are to be generalized when @xmath13 satisfies corner - extendable conditions and crisscross - extendability , which is introduced as follows .
the crisscross lattice @xmath321 is defined by @xmath322 indeed , @xmath323{crisscross.eps}\ ] ] @xmath324 where @xmath325 is the origin of @xmath0 . for @xmath11 ,
denote by @xmath326 later , the extendability of @xmath327 on @xmath0 will be reduced to study the extendability on @xmath328 .
first , the subset @xmath329 of @xmath13 , which is the collection of all crisscross - extendable patterns in @xmath13 , is introduced .
[ definition:3.8 ] for @xmath11 , @xmath330 is the maximal subset of @xmath13 such that if @xmath331 , there exists @xmath332 with @xmath333 .
clearly , @xmath329 can be obtained through the following finite processes .
[ proposition:3.9 ] for @xmath11 , @xmath334 where @xmath335 and @xmath336 that the shift spaces @xmath39 and @xmath337 are equal is proven as follows .
[ proposition:3.10 ] for @xmath11 , @xmath338 in particular , @xmath337 is mixing if and only if @xmath39 is mixing .
clearly , @xmath339 .
suppose that @xmath340
. then , there exists @xmath341 but @xmath342 , that is , @xmath343 for some @xmath344 . since @xmath341 , from the definition of @xmath329
, we have @xmath345 .
this leads a contradiction .
thus , @xmath346 .
the proof is complete .
the following notation is important in studying rectangle - extendability and mixing for @xmath39 when @xmath283 or @xmath284 is degenerated .
[ definition:3.11 ] a basic set @xmath11 is called crisscross - extendable if @xmath347 , that is , for every @xmath348 , there exists @xmath332 with @xmath333 .
when @xmath11 is not crisscross - extendable , the maximal crisscross - extendable subset @xmath349 of @xmath13 can be obtained as follows .
indeed , let @xmath350 and @xmath351 for all @xmath352 . clearly , @xmath353 for all @xmath352 .
we have that the size of @xmath283 and @xmath284 is @xmath354 .
define @xmath355 from the construction of @xmath349 , by proposition [ proposition:3.10 ] , the following result can be obtained easily and the proof is omitted .
[ proposition:3.12 ] for @xmath11 , @xmath349 is crisscross - extendable , @xmath356 .
therefore , if @xmath11 is not crisscross - extendable , then the discussion concerning rectangle - extendability and mixing of @xmath39 is the same as that of @xmath357 .
hereafter , @xmath11 is always assumed to be crisscross - extendable when @xmath283 or @xmath284 is degenerated .
[ proposition:3.13 ] if @xmath13 satisfies either @xmath291 and @xmath358 or @xmath292 and @xmath293 , then the following statements are equivalent .
@xmath13 is rectangle - extendable .
@xmath13 is crisscross - extendable .
clearly , ( i ) implies ( ii ) .
( ii)@xmath298(i ) .
assume that @xmath13 satisfies @xmath291 and @xmath358 .
the case in which it satisfies @xmath292 and @xmath293 is similar .
let @xmath359 , @xmath17 .
since @xmath13 satisfies @xmath291 and @xmath358 , from ( ii ) , @xmath360 can be extended in both positive and and negative vertical directions by using the local patterns in @xmath13 , as follows .
@xmath361{extend.eps}\ ] ] @xmath362 similarly , the above pattern can be extended in both positive horizontal and negative horizontal directions by using the local patterns in @xmath13 .
therefore , by the above method , @xmath360 can be extended to @xmath0 by using the local patterns in @xmath13 .
the proof is complete .
theorem [ theorem:3.7 ] can now be generalized .
indeed , the following theorem shows that primitivity and mixing are equivalent when @xmath13 is crisscross - extendable and satisfies the corner - extendable conditions .
[ theorem:3.14 ] if 1 .
@xmath61 is crisscross - extendable , 2 .
@xmath13 satisfies three of corner - extendable conditions @xmath62 , @xmath63 , then @xmath64 and @xmath65 are primitive for all @xmath51 if and only if @xmath39 is mixing . @xmath363 . from ( ii ) , without loss of generality , assume that @xmath13 satisfies conditions @xmath291 , @xmath292 and @xmath358 .
let @xmath24 and @xmath25 be finite sublattices of @xmath0 . since @xmath13 satisfies @xmath291 and @xmath292 , as in the proof of theorem [ theorem:3.7 ] , there exists @xmath364 such that for all @xmath308 with @xmath28 and any two allowable patterns @xmath309 and @xmath310 , @xmath40 and @xmath41 can be extended as @xmath318 on the rectangular lattice by using the local patterns in @xmath13 . since @xmath13 is crisscross - extendable and satisfies conditions @xmath291 and @xmath358 , by proposition [ proposition:3.13 ] , @xmath13 is rectangle - extendable .
then , @xmath318 can be extended as @xmath272 with @xmath32 and @xmath33 .
therefore , @xmath39 is mixing .
@xmath365 . from ( i ) and ( ii ) , by proposition [ proposition:3.13 ] , @xmath13 is rectangle - extendable .
then , for @xmath51 , any pattern in @xmath366 can be extended to @xmath0 by using the local patterns in @xmath13 .
therefore , that @xmath39 is mixing implies that @xmath64 is primitive for all @xmath51 .
similarly , @xmath65 is primitive for all @xmath51 .
the proof is complete .
the following example demonstrates that the corner - extendable conditions of theorem [ theorem:3.14 ] are crucial : if corner - extendable conditions fail , then crisscross - extendability ( or rectangle - extendability ) and primitivity may not imply mixing .
[ example:3.15 ] let @xmath367{example_1.eps } \end{array } \hspace{5.5cm}\right\}.\ ] ] clearly , @xmath368.\ ] ] from ( [ eqn:2.9 ] ) and ( [ eqn:2.10 ] ) , it can be verified that @xmath49 is non - compressible for all @xmath51 . since @xmath369 , @xmath50 is also non - compressible for all @xmath51 . by lemma [ lemma:3.5 ] , @xmath370 is rectangle - extendable . in particular , @xmath370 is crisscross - extendable . for @xmath51 ,
any two @xmath371 local patterns @xmath372 and @xmath373 , @xmath374 , @xmath375 , can be connected in the horizontal direction by using the local patterns in @xmath370 , as follows .
@xmath376{example_2.eps}\ ] ] @xmath377 then , @xmath378 for all @xmath51 .
since @xmath369 , @xmath379 for all @xmath51 .
let @xmath380 .
consider @xmath381 and @xmath382 .
clearly , @xmath383 , but @xmath384 can not connect with @xmath385 by using the local patterns in @xmath370 for all @xmath386 . therefore , @xmath387 is not mixing .
this claim does not contradict theorem [ theorem:3.14 ] since @xmath370 does not satisfy conditions @xmath292 and @xmath293 : neither @xmath388{l2_1.eps } & \hspace{2.0 cm } \text{nor } & \hspace{1.0 cm } \psfrag{a}{{\small$0 $ } } \psfrag{b}{{\small$1 $ } } \psfrag{c}{$\in\sigma_{\mathbb{l}_{4}}(\mathcal{b}_{\pi/4})$ } \includegraphics[scale=0.7]{l4_1.eps } \end{array}\ ] ] can be extended to @xmath389 by using the local patterns in @xmath370 .
the following corollary follows directly from the proof of theorem [ theorem:3.14 ] and is useful for checking that @xmath39 is not mixing .
[ corollary:3.16 ] given @xmath61 , assume that @xmath13 is rectangle - extendable .
if there exists @xmath299 such that @xmath390 or @xmath391 is not primitive , then @xmath39 is not mixing . at the end of this section , we recall and compare various mixing properties which are described in the introduction . [ definition :
a.3 ] suppose @xmath20 is a @xmath0 shift .
@xmath20 has the uniform filling property ( ufp ) if a number @xmath34 exists such that for any two allowable patterns @xmath29 and @xmath35 with @xmath36 , where @xmath392 , @xmath1 and @xmath2 , and @xmath393 , there exists a global pattern @xmath31 with @xmath32 and @xmath38 .
@xmath20 is strongly irreducible if a number @xmath34 exists such that for any two allowable patterns @xmath29 and @xmath35 with @xmath36 , where @xmath394 is finite and @xmath393 , there exists a global pattern @xmath31 with @xmath32 and @xmath38 .
@xmath20 is corner gluing if a number @xmath34 exists such that for any two allowable patterns @xmath29 and @xmath35 with @xmath36 , where @xmath392 , @xmath1 and @xmath2 , and @xmath395 , @xmath396 and @xmath397 , there exists a global pattern @xmath31 with @xmath32 and @xmath38 .
@xmath20 is block gluing if a number @xmath34 exists such that for any two allowable patterns @xmath29 and @xmath35 with @xmath36 , where @xmath398 and @xmath399 , @xmath400 and @xmath401 , @xmath402 , there exists a global pattern @xmath31 with @xmath32 and @xmath38 .
@xmath20 has periodic specification if a number @xmath34 exists such that for any two allowable patterns @xmath29 and @xmath35 with @xmath36 , where @xmath403 are finite , there exists a periodic pattern @xmath404 , @xmath1 , with @xmath32 and @xmath38 , where @xmath405 is the set of all periodic patterns in @xmath20 with period @xmath406 in horizontal direction and period @xmath101 in vertical direction .
notably , ( i)@xmath407(iv ) are introduced in boyle et al .
@xcite and ( iv ) is newly introduced .
[ remark:3.14 - 0 ] boyle et al .
@xcite showed that the strong irreducibility for both @xmath24 and @xmath25 are finite is equivalent to @xmath24 is finite and @xmath25 is arbitrary .
hence , it is clear that periodic specification implies strong irreducibility . by a similar argument as in ward @xcite in which @xmath39 with strong specification has dense periodic patterns
, it can be shown that strong specification for @xmath39 implies periodic specification for @xmath39 .
recently , ceccherini - silberstein and coornaert @xcite proved that strong specification is equivalent to strong irreducibility .
therefore , strong specification , periodic specification and strong irreducibility for @xmath39 are all equivalent . as theorem [ theorem:3.14 ]
, it can be proven that @xmath408-primitivity and block gluing are equivalent when @xmath13 is rectangle - extendable . for brevity
, the proof is omitted [ theorem:3.14 - 1 ] if @xmath61 is rectangle - extendable , then the following statements are equivalent .
there exists @xmath409 such that @xmath49 and @xmath50 are @xmath408-primitive for all @xmath51 @xmath39 is block gluing .
this section introduces invariant diagonal cycles and primitive commutative cycles to provide finitely checkable conditions for the primitivity of @xmath49 or @xmath50 for @xmath51 . for brevity ,
the discussion for @xmath49 is more addressed .
the discussion for @xmath50 is similar to that for @xmath49 .
this subsection introduces invariant diagonal cycles to provide finitely checkable conditions for the primitivity of @xmath49 or @xmath50 .
first , the diagonal index set is defined by @xmath410 clearly , if @xmath411 , then @xmath412 lies on the diagonal of @xmath175 in ( [ eqn:2.34 ] ) .
[ definition:4.9 ] for @xmath134 , a finite sequence @xmath413 is called a diagonal cycle with length @xmath256 if @xmath414 for @xmath415 . a diagonal cycle @xmath413
is called an @xmath416-invariant diagonal cycle of order @xmath417 if there exist @xmath177 and an invariant index set @xmath418 such that @xmath419 for all @xmath420 . a diagonal cycle @xmath413
is called a @xmath421-invariant diagonal cycle of order @xmath417 if there exist @xmath177 and an invariant index set @xmath418 such that @xmath422 for all @xmath420 .
notably , it is easy to show that for any @xmath352 , @xmath423 for all @xmath420 if ( [ eqn:4.9 ] ) holds .
the case for @xmath421-invariant diagonal cycles is similar . in the case of non - degeneracy
, the following simpler sufficient condition for primitivity holds . the proof is postponed until after theorem [ theorem:4.10 ] which treats more general situation . [ theorem:4.11 - 1 ] given @xmath424 , if 1 .
@xmath283 is non - degenerated , 2 .
there exists an @xmath416-invariant diagonal cycle @xmath425 of order @xmath426 with its invariant index set @xmath427 , 3 .
@xmath428 is primitive for @xmath429 , then @xmath49 is primitive for all @xmath51 .
similarly , if 1 .
@xmath284 is non - degenerated , 2 .
there exists a @xmath421-invariant diagonal cycle @xmath425 of order @xmath426 with its invariant index set @xmath427 , 3 .
@xmath430 is primitive for @xmath429 , then @xmath50 is primitive for all @xmath51 .
furthermore , if ( i)@xmath407(iii ) and ( i)@xmath407(iii) hold , then @xmath39 is mixing . the following example illustrates the application of theorem [ theorem:4.11 - 1 ] .
[ example:4.12 ] consider @xmath431.\ ] ] clearly , @xmath269 is non - degenerated . from ( [ eqn:2.27 ] ) , @xmath432 .\ ] ]
let @xmath433 and @xmath434 .
since @xmath435 for @xmath420 , @xmath436 is an s - invariant diagonal cycle of order @xmath437 with index set @xmath427 .
clearly , @xmath438\ ] ] is primitive . by using theorem [ theorem:4.11 - 1 ] , @xmath49 is primitive for all @xmath51 .
mixing of @xmath39 will be proven in example [ example:5.8 ] . to study the general cases , which includes degenerated case , more notation and lemmas are required .
first , the crisscross lattice @xmath328 is decomposed into four rectangular lattices : @xmath439 where @xmath440 [ definition:4.1 ] for @xmath441 , a basic set @xmath11 is called @xmath442-extendable if for any @xmath327 , there exists @xmath443 such that @xmath444 . clearly , a basic set @xmath11 is crisscross - extendable if and only if @xmath13 is @xmath442-extendable for all @xmath441 .
let @xmath160_{n\times n}$ ] be a non - negative matrix ; the index set of non - zero rows of @xmath52 and the index set of non - zero columns of @xmath52 are denoted by @xmath445 respectively . from ( [ eqn:2.8 ] ) , it is easy to show that @xmath446 now , the following lemma is obtained .
[ lemma:4.2 ] assume that @xmath11 is @xmath447-extendable .
for @xmath51 and @xmath134 , denote @xmath448_{p^{q+1}\times p^{q+1}}$ ] and @xmath152_{p^{q+1}\times p^{q+1}}$ ] .
if @xmath449 , then @xmath450 is not a zero matrix .
since @xmath13 is @xmath447-extendable , any pattern in @xmath451 can be extended to @xmath452 by using the local patterns in @xmath13 . from ( [ eqn:2.16 ] ) ,
the result follows . in degenerated case ,
the weak non - degeneracy of @xmath283 ( or @xmath284 ) is introduced and is useful in establishing the primitivity of @xmath64 ( or @xmath65 ) .
[ definition:4.3 ] given @xmath11 , @xmath453_{p\times p}$ ] is weakly non - degenerated if 1 . when both @xmath454 and @xmath455 are not zero matrices , @xmath456 , @xmath457 2 . when both @xmath458 and @xmath459 are not zero matrices , @xmath460 , @xmath461 weak non - degeneracy of @xmath284 is defined analogously .
clearly , if @xmath283 ( or @xmath284 ) is non - degenerated , then @xmath283 ( or @xmath284 ) is weakly non - degenerated .
the following lemma can be proven straightforwardly from lemma [ lemma:4.2 ] and the proof is omitted .
[ lemma:4.4 ] given @xmath11 , if @xmath13 is @xmath447-extendable and @xmath283 is weakly non - degenerated , then 1 . when both @xmath454 and @xmath455 are not zero matrices , @xmath456 , @xmath462 2 .
when both @xmath458 and @xmath459 are not zero matrices , @xmath460 , @xmath463 in the following lemma , the corner - extendable conditions @xmath62 can be obtained from the @xmath442-extendability and weak non - degeneracy . [ lemma:4.5 ] given @xmath61 , assume @xmath283 is weakly non - degenerated , then 1 .
if @xmath13 is @xmath464- and @xmath447-extendable , then @xmath13 satisfies @xmath291 ; 2 .
if @xmath13 is @xmath447- and @xmath465-extendable , then @xmath13 satisfies @xmath292 .
similarly , assume @xmath284 is weakly non - degenerated , then 1 .
if @xmath13 is @xmath464- and @xmath447-extendable , then @xmath13 satisfies @xmath291 ; 2 .
if @xmath13 is @xmath464- and @xmath466-extendable , then @xmath13 satisfies @xmath293 . for simplicity ,
only ( i ) is proven .
the other are proven similarly . for any @xmath467 ,
let @xmath468 , @xmath469 and @xmath470 .
the @xmath447-extendability of @xmath13 implies that @xmath471 is not a zero matrix . from the @xmath464-extendability of @xmath13
, there exist @xmath472 such that @xmath473 .
since @xmath283 is weakly non - degenerated , there exists @xmath474 such that @xmath475 , that is , @xmath476{x_ij.eps } \end{array}\ ] ] where @xmath477 .
therefore , @xmath478 can be extended to @xmath389 by using the local patterns in @xmath13 , which implies @xmath13 satisfies @xmath291 .
the proof is complete . for easily expressing the primitivity of compressible matrices ,
the following definition is introduced .
[ definition:4.7 ] if @xmath160_{n\times n}$ ] is a matrix with @xmath479 , the associated saturated matrix @xmath480_{n\times n}$ ] of @xmath52 is defined by @xmath481 clearly , given @xmath160_{n\times n}$ ] with @xmath479 , if there exists @xmath54 such that @xmath482 , then @xmath52 is primitive ( @xmath53-primitive ) ; here , if @xmath483_{n\times n}$ ] and @xmath165_{n\times n}$ ] are two matrices , @xmath484 means @xmath485 for all @xmath486 . for @xmath51 and @xmath139 , as ( [ eqn:2.11 ] ) , let @xmath487_{p^{q+1}\times p^{q+1}}=\left[h_{n+q;\beta_{1};\beta_{2};\cdots;\beta_{q+1}}\right].\ ] ] for convenience , @xmath488 is denoted by @xmath489_{p^{q+1}\times p^{q+1}}=\left[e_{n+q;\beta_{1};\beta_{2};\cdots;\beta_{q+1}}\right].\ ] ] from ( [ eqn:2.19 ] ) , it is easy to verify that @xmath490_{p\times p } \right].\ ] ] lemmas [ lemma:4.4 ] and [ lemma:4.5 ] yield the following lemma .
[ lemma:4.6 ] given @xmath11 , for @xmath51 , let @xmath491_{p \times p}$ ] and @xmath492_{p \times p}$ ] .
assume that @xmath13 is @xmath493-extendable , @xmath494 , and @xmath283 is weakly non - degenerated .
then , for @xmath51 and @xmath495 , @xmath496 when @xmath471 is not a zero matrix ; @xmath497 when @xmath498 is not a zero matrix . by lemma [ lemma:4.5 ]
, we have that @xmath13 satisfies @xmath291 and @xmath292 . from the @xmath464-extendability of @xmath13 and @xmath291 , for @xmath499
, every pattern @xmath500 can be extended to @xmath501 by using the local patterns in @xmath13 .
therefore , by lemma [ lemma:4.4 ] , it can be verified that ( [ eqn:4.4 ] ) holds ; the details of the proof are omitted .
similarly , ( [ eqn:4.5 ] ) can be shown to hold .
we also need following notation in proving the theorem for the primitivity of @xmath49 .
[ definition:4.8 ] let @xmath502_{n\times n}$ ] , where @xmath503 is an @xmath504 non - negative matrix for @xmath505 .
the indicator matrix @xmath506_{n\times n}$ ] of @xmath507 is defined by @xmath508 where @xmath509 is the sum of all entries in @xmath503 .
the following theorem provides a sufficient condition for the primitivity of @xmath49 when @xmath269 is weakly non - degenerated with some @xmath442-extendability .
[ theorem:4.10 ] given @xmath424 , if 1 .
@xmath283 is weakly non - degenerated , 2 .
@xmath13 is @xmath442-extendable , @xmath510 , 3 .
there exists an @xmath416-invariant diagonal cycle @xmath425 of order @xmath426 with its invariant index set @xmath427 , 4 . for @xmath429
, there exists @xmath511 such that @xmath512 5 .
@xmath49 is primitive for @xmath429 , then @xmath49 is primitive for all @xmath51 .
the result that @xmath49 is primitive for @xmath513 is proven by induction , as follows . for @xmath514 , the statement @xmath515 means that @xmath49 is primitive for @xmath516 . from ( v ) , @xmath517 is true .
assume that @xmath518 follows for some @xmath519 , that is , @xmath49 is primitive for @xmath520 . for @xmath521
, let @xmath522 , where @xmath523 .
let @xmath524 , define @xmath525 . from ( [ eqn:4.10 ] )
, @xmath526 is an @xmath416-invariant diagonal cycle of order @xmath527 with invariant index set @xmath427
. from ( [ eqn:2.34 ] ) , let @xmath528_{p^{n}\times p^{n}}$ ] and @xmath529_{p^{n}\times p^{n}}$ ] for @xmath530 .
then , from ( [ eqn:2.19 ] ) , @xmath531=\left ( \mathbb{h}_{n } \right)_{p^{n}\times p^{n}}\circ\left [ e_{p^{n-1}\times p^{n-1}}\otimes \left[h_{r;i , j}\right]_{p\times p } \right].\ ] ] by lemma [ lemma:4.2 ] , @xmath532 is the indicative matrix of @xmath528 $ ] . by the assumption for @xmath518
, @xmath532 is primitive ; then suppose @xmath533 for some @xmath534 .
therefore , for any @xmath535 with @xmath536 , there exist @xmath537 such that @xmath538 where @xmath539 is not a zero matrix for all @xmath540 . from ( [ eqn:2.36 ] ) ,
@xmath541 notably , @xmath542 is on the diagonal of @xmath543 ; then let @xmath544 for some @xmath545 .
hence , form ( iv ) , @xmath546 . since @xmath547 , @xmath548 for some @xmath549 .
let @xmath550 and @xmath551_{p^{n}\times p^{n } } $ ] .
for @xmath535 with @xmath536 , from ( [ eqn:4.4 ] ) and ( [ eqn:4.5 ] ) , @xmath552 hence , from ( [ eqn:4.8 ] ) , @xmath553_{p\times p}\right ] \\
\geq \mathbb{e}(\mathbb{h}_{n}).\ ] ] then , @xmath49 is primitive for @xmath521 , that is , @xmath554 holds . therefore , @xmath515 is true for all @xmath514 , implying that @xmath49 is primitive for all @xmath51 .
the proof is complete .
now , the proof of theorem [ theorem:4.11 - 1 ] is given as follows . from ( [ eqn:2.3 ] ) , ( [ eqn:2.4 ] ) and ( [ eqn:4.2 ] ) ,
clearly , @xmath13 is @xmath442-extendable for @xmath510 . from ( iii ) , by proposition [ proposition:3.4 ]
, it can be verified that @xmath49 is primitive for @xmath429 and the detail is omitted for brevity
. therefore , by theorem [ theorem:4.10 ] , @xmath49 is primitive for all @xmath51 . for @xmath51 ,
let @xmath555_{1\leq\alpha\leq p^{2}}$ ] and @xmath556_{1\leq\alpha\leq p^{2}}$ ] .
like theorem [ theorem:4.10 ] , the following theorem provides a sufficient condition for the primitivity of @xmath50 .
[ theorem:4.11 ] given @xmath424 , if 1 .
@xmath284 is weakly non - degenerated , 2 .
@xmath13 is @xmath442-extendable , @xmath557 , 3 .
there exists a @xmath421-invariant diagonal cycle @xmath425 of order @xmath426 with its invariant index set @xmath427 , 4 . for @xmath429
, there exists @xmath558 such that @xmath559 5 .
@xmath50 is primitive for @xmath429 , then @xmath50 is primitive for all @xmath51 . in viewing the primitive properties in theorem [ theorem:4.10 ] and theorem [ theorem:4.11 ] , we introduce the @xmath560- and @xmath561-primitive conditions for @xmath13 has invariant diagonal cycles as follows .
later , @xmath562- and @xmath563-primitive conditions are introduced for @xmath13 has primitive commutative cycles ; see definition [ definition:5.6 ] .
[ definition:4.13 ] let @xmath11 .
@xmath13 satisfies @xmath560-primitive condition if the conditions ( iii)@xmath407(v ) of theorem [ theorem:4.10 ] are satisfied .
@xmath13 satisfies @xmath561-primitive condition if the condition ( iii)@xmath407(v ) of theorem [ theorem:4.11 ] are satisfied .
[ remark:4.13 - 1 ] from the proof of theorem [ theorem:4.11 - 1 ] , if @xmath283 is non - degenerated and the conditions ( ii)@xmath407(iii ) of theorem [ theorem:4.11 - 1 ] are satisfied , then @xmath13 satisfies the @xmath560-primitive condition .
a similar result holds when @xmath284 is non - degenerated . from lemma [ lemma:4.5 ] and theorems
[ theorem:3.14 ] , [ theorem:4.10 ] and [ theorem:4.11 ] , mixing of @xmath39 follows .
[ theorem:4.14 ] given @xmath424 , if 1 .
@xmath283 and @xmath284 are weakly non - degenerated , 2 .
@xmath13 is crisscross - extendable , 3 .
@xmath13 satisfies @xmath560- and @xmath561-primitive conditions , then @xmath39 is mixing .
the following well - known examples illustrate theorem [ theorem:4.14 ] .
the first example is the golden - mean shift ( or the hard square model ) .
[ example:4.15 ] the rule of the golden - mean shift @xmath564 , @xmath565 , is that there is no two 1 s next to each other in horizontal or vertical direction . from ( [ eqn:2.2 - 1 ] ) , @xmath566= \left [ \begin{array}{cc } h_{2;1,1 } & h_{2;1,2 } \\ h_{2;2,1 } & h_{2;2,2 } \end{array } \right]\ ] ] and @xmath567 = \left [ \begin{array}{cc } e_{2;1,1 } & e_{2;1,2 } \\ e_{2;2,1 } & e_{2;2,2 } \end{array } \right].\ ] ] clearly , @xmath568 . from ( [ eqn:4.2 ] ) ,
@xmath569 is crisscross - extendable . that @xmath570 and @xmath571 are weakly non - degenerated are easily seen .
now , @xmath572.\ ] ] then , @xmath433 is an s - invariant diagonal cycle of order @xmath573 with index set @xmath574 . clearly , @xmath575\geq e_{2;1,1}=e_{2;1}.\ ] ] since @xmath576 , @xmath269 is primitive
. therefore , @xmath569 satisfies the @xmath560-primitive condition . since @xmath369 , @xmath569 also satisfies @xmath561-primitive condition .
hence , by theorem [ theorem:4.14 ] , the golden - mean shift @xmath564 is mixing .
in fact , that the golden - mean shift has strong specification will be shown in example [ example:6.5 ] .
the next example concerns the three - coloring of the square lattice , which is closely related to the six - vertex ice model in statistical physics @xcite , see also example [ example:7.7 ] of this paper .
[ example:4.16 ] the three - coloring of the square lattice is the coloring of the square lattice @xmath0 with three colors such that no two adjacent vertices have the same color .
let @xmath577 be the basic set of the three - colouring of the square lattice .
we have that @xmath578=\left [ \begin{array}{ccc } h_{2;1,1 } & h_{2;1,2 } & h_{2;1,3 } \\
h_{2;2,1 } & h_{2;2,2 } & h_{2;2,3 } \\
h_{2;3,1 } & h_{2;3,2 } & h_{2;3,3 } \\
\end{array } \right].\ ] ] clearly , @xmath369 .
it is easy to verify that @xmath579 is crisscross - extendable and @xmath580 is weakly non - degenerated .
we have that @xmath581 & \text{and } & s_{2;5,1}=c_{2;4,4}= \left [ \begin{array}{ccc } 0 & 1 & 1 \\ 0 & 0 & 0 \\ 0 & 1 & 0 \end{array } \right ] .
\end{array}\ ] ] since @xmath582 $ ] , @xmath583 is an @xmath416-invariant diagonal cycle of order @xmath584 with its invariant index set @xmath585 . that the conditions ( iv ) and ( v ) of theorem [ theorem:4.10 ] are satisfied can be checked straightforwardly ; the details are omitted for brevity
hence , @xmath579 satisfies the @xmath560-primitive condition .
since @xmath369 , @xmath579 satisfies the @xmath561-primitive condition .
therefore , by theorem [ theorem:4.14 ] , @xmath586 is mixing .
it can be proved that @xmath586 does not have strong specification ; the detail is omitted .
this subsection introduces primitive commutative cycles in order to obtain another finitely checkable sufficient condition for the primitivity of @xmath49 or @xmath50 when invariant diagonal cycles are unavailable . for @xmath587 ,
let @xmath588 and @xmath589 are two cycles , where @xmath590 for @xmath591 and @xmath592 .
[ definition:5.1 ] if @xmath593 , let @xmath594 and @xmath595 .
the pair @xmath596 and @xmath597 is called a commutative cycle pair .
given a commutative cycle pair @xmath596 and @xmath597 , denote the index of @xmath596 and @xmath597 by @xmath598 , where @xmath599 from ( [ eqn:2.20 ] ) , it is easy to check that @xmath600 moreover , the number @xmath601 is a member of the diagonal index set @xmath602 , and then @xmath603 lies on the diagonal of @xmath176 .
[ definition:5.2 ] a commutative cycle pair @xmath596 and @xmath597 with index @xmath598 is called an @xmath604-primitive commutative cycle pair if there exists @xmath605 such that @xmath606 a @xmath607-primitive commutative cycle pair is similarly specified , and the details are omitted .
the following theorem provides a sufficient condition for the primitivity of @xmath49 when @xmath269 is weakly non - degenerated .
[ theorem:5.3 ] given @xmath11 , if 1 .
@xmath269 is weakly non - degenerated , 2 .
@xmath13 is @xmath442-extendable , @xmath510 , 3 .
there exists an @xmath604-primitive commutative cycle pair @xmath596 and @xmath597 with index @xmath598 such that @xmath608 or @xmath609 , 4 .
@xmath269 is primitive , then @xmath49 is primitive for all @xmath51 . by lemma [ lemma:4.5 ] , @xmath13 satisfies @xmath291 and @xmath292 .
suppose @xmath608 .
the case for @xmath609 is similar . from ( i ) and
( ii ) , by lemma [ lemma:4.6 ] , ( [ eqn:4.4 ] ) and ( [ eqn:4.5 ] ) follow . first , we show that for @xmath51 , there exists @xmath610 such that @xmath611 by induction . from ( [ eqn:4.4 ] ) , ( [ eqn:4.5 ] ) and ( [ eqn:5.3 ] ) , it is clear that ( [ eqn:5.4 ] ) holds for @xmath612 .
assume that the case for @xmath613 is true , @xmath614 .
let @xmath615_{p\times p}=\left[h_{m,2;\bar{\alpha};\alpha}^{(k)}\right]_{1\leq \alpha\leq p^{2}}$ ] , from ( [ eqn:2.33 ] ) , it is clear that @xmath616 for all @xmath617 .
let @xmath618 be the indicator matrix of @xmath615_{p\times p}$ ] . since @xmath619 , we have @xmath620 .
let @xmath621_{p\times p}=\left[h_{m , t+1;\bar{\alpha};\alpha}^{(k)}\right]_{1\leq \alpha\leq p^{2}}$ ] , by proposition [ proposition:2.2 ] , @xmath622 for all @xmath617 .
since @xmath13 is @xmath447-extendable and satisfies condition @xmath291 , for @xmath177 , every pattern @xmath623 can be extended to @xmath624 by using the local patterns in @xmath13 .
thus , if @xmath625 , then @xmath626 is not a zero matrix for @xmath230 and @xmath627 .
hence , @xmath618 is also the indicator matrix of @xmath621_{p\times p}$ ] . since @xmath608 and @xmath628 , @xmath629 .
notably , @xmath630 lies on the diagonal of @xmath631 .
let @xmath632 . since @xmath620 and @xmath629 , from ( [ eqn:4.8])@xmath407([eqn:4.5 ] ) , @xmath633_{1\leq\alpha\leq p^{2 } } \right ] \geq e_{t+1;\bar{\alpha}}.\ ] ] from ( [ eqn:4.4 ] ) and ( [ eqn:4.5 ] ) , @xmath634 .
hence , the case for @xmath635 holds . indeed , @xmath636 .
therefore , ( [ eqn:5.4 ] ) follows .
now , we want to show that @xmath49 is primitive for all @xmath51 .
first , the case for @xmath612 directly follows from ( iv ) . for @xmath637 , from ( [ eqn:2.19 ] ) , @xmath638_{p^{2}\times p^{2 } } = \left ( \mathbb{h}_{2 } \right)_{p^{2}\times p^{2}}\circ\left [ e_{p\times p}\otimes \left[h_{n-1;\alpha}\right]_{1\leq\alpha\leq p^{2 } } \right].\ ] ] from ( ii ) , by lemma [ lemma:4.2 ] , if @xmath639
, @xmath640 is not a zero matrix .
hence , @xmath269 is the indicator matrix of @xmath641_{p^{2}\times p^{2 } } $ ] .
let @xmath642_{p^{2}\times p^{2 } } = \left[h_{m , n;\beta_{1};\beta_{2}}\right]$ ] .
since @xmath608 , @xmath643 from ( [ eqn:5.4 ] ) , clearly , @xmath644 .
notably , @xmath645 is on the diagonal of @xmath176 .
therefore , from ( iv ) and ( [ eqn:4.8])@xmath407([eqn:4.5 ] ) , it can be verified that there exists @xmath646 such that @xmath647_{p\times p}\right ] \geq \mathbb{e}(\mathbb{h}_{n}).\ ] ] therefore , @xmath49 is primitive for all @xmath51 .
the proof is complete .
similarly , the following theorem provides a sufficient condition for the primitivity of @xmath50 , and the proof is omitted .
[ theorem:5.4 ] given @xmath11 , if 1 .
@xmath267 is weakly non - degenerated , 2 .
@xmath13 is @xmath442-extendable , @xmath557 , 3 .
there exists a @xmath607-primitive commutative cycle pair @xmath596 and @xmath597 with index @xmath598 such that @xmath648 or @xmath649 , 4 .
@xmath267 is primitive , then @xmath50 is primitive for all @xmath51 .
[ remark:5.4 - 1 ] in practice , to find an invariant diagonal cycle is easier than to find a primitive commutative cycle pair . however , it may be more convenient to verify primitivity of @xmath49 or @xmath50 by using the method of primitive commutative cycle pair ; see example [ example:5.5 ] and [ example:5.8 ] .
the following example illustrates theorem [ theorem:5.3 ] .
[ example:5.5 ] consider @xmath650.\ ] ] clearly , @xmath269 is weakly non - degenerated .
it can be verified that @xmath13 is @xmath442-extendable , @xmath510 .
we also have that @xmath269 is primitive .
let @xmath651 and @xmath652 ; the following is easily computed .
@xmath653\ ] ] and @xmath654.\ ] ] we have that @xmath655.\ ] ] then , @xmath656 hence , @xmath657 and @xmath658 is an @xmath604-primitive commutative cycle pair with index @xmath659 .
moreover , @xmath660 therefore , by theorem [ theorem:5.3 ] , @xmath49 is primitive for all @xmath51 . as theorem
[ theorem:4.11 - 1 ] , theorem [ theorem:5.3 ] can be made simpler when @xmath283 is non - degenerated .
the proof is similar to that of theorem [ theorem:4.11 - 1 ] and is omitted here .
the result for @xmath267 is also valid .
[ theorem:5.4 - 1 ] given @xmath11 , if 1 .
@xmath269 is non - degenerated , 2 .
there exists an @xmath604-primitive commutative cycle pair @xmath596 and @xmath597 with index @xmath598 such that @xmath608 or @xmath609 , then @xmath49 is primitive for all @xmath51 . as the primitive conditions in definition 4.14 for invariant diagonal cycles ,
the primitive conditions for primitive commutative cycles in theorem [ theorem:5.3 ] and [ theorem:5.4 ] are introduced as follows .
[ definition:5.6 ] let @xmath11 .
@xmath13 satisfies @xmath562-primitive condition if the conditions ( iii ) and ( iv ) of theorem [ theorem:5.3 ] are satisfied .
@xmath13 satisfies @xmath563-primitive condition if the conditions ( iii ) and ( iv ) of theorem [ theorem:5.4 ] are satisfied .
[ remark:5.6 - 1 ] if @xmath283 is non - degenerated and condition ( ii ) of theorem [ theorem:5.4 - 1 ] is satisfied , then @xmath13 satisfies the @xmath562-primitive condition .
a similar result holds when @xmath284 is non - degenerated . since it may happen that @xmath661 ( or @xmath662 ) has invariant diagonal cycle and @xmath662 ( or @xmath661 ) has primitive commutative cycles
therefore , combining the primitive conditions for invariant diagonal cycles and primitive commutative cycles , theorem [ theorem:4.14 ] is generalized , yielding a finitely checkable condition for mixing of @xmath39 .
[ theorem:5.7 ] given @xmath424 , if 1 .
@xmath283 and @xmath284 are weakly non - degenerated , 2 .
@xmath13 is crisscross - extendable , 3 .
@xmath13 satisfies @xmath663- and @xmath664-primitive conditions for some @xmath665 , then @xmath39 is mixing .
notably , when both @xmath283 and @xmath284 are non - degenerated , the condition ( ii ) of theorem [ theorem:5.7 ] automatically holds .
the following example illustrates the application of theorem [ theorem:5.7 ] .
[ example:5.8 ] ( continued ) in example [ example:4.12 ] , @xmath666.\ ] ] clearly , @xmath284 is non - degenerated .
let @xmath667 and @xmath668 .
that @xmath669\ ] ] and @xmath670\ ] ] are primitive can be easily verified .
hence , @xmath671 and @xmath672 form a @xmath607-primitive commutative cycle pair with index @xmath673 .
moreover , @xmath674 therefore , @xmath13 satisfies the @xmath563-primitive condition . from the result in example [ example:4.12 ]
, @xmath13 satisfies the @xmath560-primitive condition .
therefore , by theorem [ theorem:5.7 ] , @xmath39 is mixing .
the following example demonstrates that the weakly non - degenerated condition is crucial in order to have topological mixing .
[ example:5.7 - 1 ] ( continued ) in example [ example:3.15 ] , clearly , @xmath675 is not weakly non - degenerated . from ( [ eqn:2.26 ] ) , @xmath676.\ ] ] hence , @xmath433 is an @xmath416-invariant diagonal cycle of order @xmath573 with index set @xmath677 , and @xmath678\ ] ] is primitive .
therefore , @xmath370 satisfies the @xmath560-primitive condition .
since @xmath675 , @xmath370 also satisfies the @xmath561-primitive condition . from example
[ example:3.15 ] , @xmath370 is crisscross - extendable .
thus , @xmath370 satisfies conditions ( ii ) and ( iii ) of theorem [ theorem:5.7 ] .
however , @xmath387 is not mixing .
therefore , the importance of the weakly non - degenerated condition is established .
[ remark:5.9 ] when invariant diagonal cycles and primitive commutative cycles are unavailable , some examples can also be shown to be mixing .
for example , consider @xmath679{example_1.eps } \end{array } \hspace{8.5cm}\right\}.\ ] ] this shift space @xmath680 is related to the shift space @xmath681 that is given by @xmath682{example_1.eps } \end{array } \hspace{8.5cm}\right\},\ ] ] which was first investigated by ledrappier @xcite who showed that @xmath681 was mixing with zero entropy .
clearly , @xmath683\ ] ] is non - degenerated .
@xmath684 can be shown to be primitive for all @xmath51 .
then , @xmath680 is mixing .
a systematic method for solving this type of problem is being developed .
[ remark:5.10 ] for studying mixing problem , the ideas in this work can be applied to higher - dimensional shifts of finite type ; see @xcite .
this section introduces the @xmath73 hole - filling condition as in step ( 5 ) in the introduction , and provides finitely checkable conditions for the strong specification of @xmath39 .
first , for @xmath685 and @xmath686 , the rectangularly annular lattice @xmath687 with hole @xmath688 and width @xmath23 ( called the annular lattice for short ) is defined by @xmath689 in particular , @xmath690 hole - filling condition is defined as follows .
[ definition:6.1 ] for @xmath11 and @xmath75 , @xmath13 satisfies @xmath73 hole - filling ( @xmath77hfc@xmath74 ) with size @xmath70 , @xmath78 , if for any @xmath691 , there exists @xmath692 such that @xmath693 . in particular , @xmath77hfc@xmath694 is also called hole - filling condition ( hfc ) . @xmath695{squarefilling.eps}\ ] ] @xmath696 @xmath77hfc@xmath74 can be expressed in terms of the horizontal transition matrices @xmath49 and the connecting operators @xmath249 and @xmath697 .
therefore , the condition can be easily checked , especially using computer programs .
the following theorem presents only the case in which @xmath13 satisfies hole - filling ; for brevity , the general case in which @xmath13 satisfies @xmath77hfc@xmath74 , @xmath698 , is presented in theorem [ theorem : a.1 ] [ theorem:6.2 ] given @xmath11 , for @xmath685 , @xmath13 satisfies hfc with size @xmath70 if and only if for @xmath699 and @xmath700 , @xmath701 , if 1 .
@xmath702 , @xmath703 , 2 .
@xmath704 , @xmath705 , where @xmath706 then @xmath707 for @xmath699 , choose @xmath708 , @xmath709 , such that @xmath710 and @xmath711 .
let @xmath712 and @xmath713 .
clearly , conditions ( i ) , ( ii ) , ( iii ) and ( iv ) imply that the empty places in the patterns @xmath714 , @xmath715 , @xmath716 and @xmath318 can can be filled with some colors in @xmath7 , such that the patterns @xmath714 , @xmath715 , @xmath716 and @xmath318 are @xmath717-admissible , respectively .
furthermore , the following annular pattern is @xmath717-admissible .
@xmath718{annular.eps}\ ] ] @xmath719 therefore , by the construction of connecting operators , ( [ eqn:6.5 ] ) is equivalent to the hole - filling condition with size @xmath70 .
the proof is complete . before showing the main theorem ,
the following notation is needed .. [ definition:6.3 - 2 ] for @xmath75 , @xmath11 is called @xmath73 crisscross - extendable if @xmath720 and @xmath721 . in particular , @xmath722 crisscross - extendability is the crisscross - extendability . in the following ,
the @xmath73 crisscross - extendability , @xmath77hfc@xmath74 and primitivity of @xmath79 and @xmath81 are shown to provide sufficient conditions for strong specification of @xmath39 .
since all conditions are finitely checkable , the theorem provides a finitely checkable sufficient condition for the strong specification of @xmath39 .
[ theorem:6.4 ] given @xmath11 , if there exists @xmath75 such that 1 .
@xmath13 is @xmath73 crisscross - extendable , 2 .
@xmath13 satisfies @xmath77hfc@xmath74 with size @xmath70 for some @xmath78 , 3 .
@xmath79 is @xmath80-primitive and @xmath81 is @xmath82-primitive , then @xmath39 has strong specification .
let @xmath723 and @xmath724 .
first , define the lattice @xmath725 , which is like the grid on a checkerboard with line width @xmath73 and @xmath726 blank spaces , as @xmath727 denote the lattice of blank spaces on the checkerboard by @xmath728 @xmath729{chess_line_3.eps } \\ \\ \text{(a ) the shadowed lattice is } \mathbb{l}_{g;3}. \hspace{1.2 cm } \text{(b ) } \text{the white lattice is } \mathbb{l}_{b;3}.\\ \ \hspace{0.5cm}\text{for } m = n=4 .
\hspace{3.3cm}\text { for } m = n=4 . \end{array}\ ] ] @xmath730 for @xmath686 , define @xmath731 @xmath732{l_bar1.eps } \end{array}\ ] ] @xmath733 clearly , @xmath734 and @xmath735 if @xmath736 .
then , @xmath0 lattice can be decomposed into disjoint sublattices : @xmath737 take @xmath738 let @xmath739 with @xmath740 .
for any @xmath741 with @xmath742 , @xmath303 , let @xmath743 for @xmath303 .
hence , @xmath716 can be extended as @xmath744 , @xmath303 .
clearly , for @xmath303 , @xmath745 from ( [ eqn:6.7 ] ) , it can be verified that it never occurs both @xmath746 and @xmath747 for all @xmath2 .
now , from conditions ( i ) and ( iii ) , there exists a @xmath13-admissible pattern @xmath748 on @xmath749 such that @xmath750 , @xmath751 . clearly , @xmath752 is the union of the discrete @xmath726 rectangular lattices .
hence , from ( [ eqn:6.7 - 1 ] ) and condition ( ii ) , there exists @xmath272 such that @xmath753 for @xmath751 .
notably , in general , @xmath754 is not equal to @xmath748 since condition ( ii ) may change the colors on the boundary of @xmath755 with width @xmath756 .
therefore , @xmath39 has strong specification .
the proof is complete .
[ remake:6.4 - 1 ] in the proof of theorem [ theorem:6.4 ] , the constant @xmath757 of strong specification is less than or equal to @xmath758 , which is given by ( [ eqn:6.7 ] ) .
it is of interest to know the optimal ( least ) value of @xmath757 .
lightwood @xcite showed that if a @xmath0 shift of finite type @xmath20 is square filling and topologically mixing , then @xmath20 has the ufp .
the following two well - known examples illustrate theorem [ theorem:6.4 ] .
the first is the golden - mean shift , which is considered in example [ example:4.15 ] .
[ example:6.5 ] from example [ example:4.15 ] , @xmath759.\ ] ] clearly , @xmath569 is crisscross - extendable and @xmath580 is @xmath722-primitive .
it can be computed that @xmath760 , & s_{2;1,2}= \left [ \begin{array}{cc } 1 & 0 \\ 1 & 0\\ \end{array } \right ] , & s_{2;1,3}= \left [ \begin{array}{cc } 1 & 0 \\ 1 & 0\\ \end{array } \right ] , \end{array}\ ] ] @xmath761 , & s_{2;2,1}= \left [ \begin{array}{cc } 1 & 1 \\ 0 & 0\\ \end{array } \right ] , & s_{2;2,3}= \left [ \begin{array}{cc } 1 & 0 \\ 0 & 0\\ \end{array } \right ] , \end{array}\ ] ] @xmath762 , & s_{2;3,2}= \left [ \begin{array}{cc } 1 & 0 \\ 0 & 0\\ \end{array } \right ] , & s_{2;4,1}= \left [ \begin{array}{cc } 1 & 1 \\ 0 & 0\\ \end{array } \right ] .
\end{array}\ ] ] the other @xmath763 are zero matrices .
then , by theorem [ theorem:6.2 ] , it is easy to verify that @xmath569 satisfies hfc with size @xmath764 .
therefore , by theorem [ theorem:6.4 ] , @xmath564 has strong specification .
burton and steif @xcite introduced the following example , which is closely related to the ferromagnetic ising model in statistical physics .
[ example:6.6 ] consider the color set @xmath765 .
the rule of @xmath766 is that a negative is disallowed to sit to a positive unless they are both @xmath767 . to fit @xmath768 to the color set @xmath769 used in this work , @xmath770 , @xmath771 , @xmath772 and @xmath722 are replaced with @xmath773 , @xmath772 , @xmath722 and @xmath774 , respectively .
then , the following can be shown ; @xmath775.\ ] ] clearly , @xmath580 is @xmath774-primitive and @xmath776 is crisscross - extendable . that @xmath776 satisfies hfc with size @xmath584 can be proven and the details are omitted .
therefore , by theorem [ theorem:6.4 ] , @xmath777 has strong specification .
the size @xmath70 of hole - filling condition may be even larger than @xmath584 , as in the following example .
[ example:6.13 ] it can be verified that @xmath778\ ] ] satisfies hfc with size @xmath779 and @xmath780\ ] ] satisfies hfc with size @xmath781 .
both have strong specification .
the details are omitted .
the following example concerns the simplified golden mean ( sgm ) , which does not satisfy hfc but satisfies @xmath77hfc@xmath782 .
[ example:6.15](simplified golden - mean ) consider @xmath783 and @xmath784.\ ] ] that @xmath785 has strong specification can be easily shown . indeed , for any two patterns @xmath786 and @xmath787 , @xmath788 , with @xmath789 , coloring the vertices in @xmath790 by @xmath773 yields a @xmath791-admissible global pattern .
therefore , sgm has strong specification with size @xmath792
. however , consider the @xmath791-admissible pattern @xmath43 in fig .
@xmath793{annular3.eps}\ ] ] @xmath794 clearly , @xmath43 can not be extended any further on the corner since the local patterns and are forbidden .
therefore , sgm does not satisfy hfc .
now , by theorem [ theorem : a.1 ] , it can be verified that @xmath791 satisfies @xmath77hfc@xmath782 with size @xmath779 .
clearly , @xmath791 is @xmath774 crisscross - extendable , and @xmath795 is @xmath722-primitive .
therefore , by theorem [ theorem:6.4 ] , sgm has strong specification .
the following example demonstrates a genuine failure of the hole - filling condition , which causes the failure of strong specification .
indeed , in @xcite , boyle at el .
consider this example to prove that block gluing is strictly weaker than corner gluing .
hence , this example does not have strong specification .
[ example:6.16 ] consider @xmath796{z22_3.eps } \end{array } \right\}$ ] , meaning that @xmath797.\ ] ] now , @xmath798 lacks the corner gluing property , as follows . consider @xmath799 and @xmath800 .
let @xmath801 , @xmath751 , where @xmath24 is the l - shaped lattice and @xmath25 is the rectangular lattice ; see fig .
@xmath802{cornergluing.eps}\ ] ] @xmath803 therefore , the following pattern on @xmath804 , in fig .
5.7 for all @xmath1 where @xmath805 , can not be filled without , which is forbidden .
@xmath806{annular4.eps}\ ] ] @xmath807 similarly , the corner gluing property fails in one of the four corners when @xmath808{z22_31.eps},\psfrag{c}{{\scriptsize $ 1 $ } } \psfrag{b}{{\scriptsize $ 1 $ } } \psfrag{d}{{\scriptsize $ 1 $ } } \psfrag{e}{{\scriptsize $ 1 $ } } \includegraphics[scale=0.5]{z22_31.eps}\in\mathcal{b}$ ] and @xmath809 for some @xmath810{z22_31.eps } , & \psfrag{c}{{\scriptsize $ 0 $ } } \psfrag{b}{{\scriptsize $ 0 $ } } \psfrag{d}{{\scriptsize $ 1 $ } } \psfrag{e}{{\scriptsize $ 0 $ } } \includegraphics[scale=0.5]{z22_31.eps } , & \psfrag{c}{{\scriptsize $ 0 $ } } \psfrag{b}{{\scriptsize $ 0 $ } } \psfrag{d}{{\scriptsize $ 0 $ } } \psfrag{e}{{\scriptsize $ 1 $ } } \includegraphics[scale=0.5]{z22_31.eps } , & \psfrag{c}{{\scriptsize $ 0 $ } } \psfrag{b}{{\scriptsize $ 1 $ } } \psfrag{d}{{\scriptsize $ 1 $ } } \psfrag{e}{{\scriptsize $ 1 $ } } \includegraphics[scale=0.5]{z22_31.eps } , & \psfrag{c}{{\scriptsize $ 1 $ } } \psfrag{b}{{\scriptsize $ 0 $ } } \psfrag{d}{{\scriptsize $ 1 $ } } \psfrag{e}{{\scriptsize $ 1 $ } } \includegraphics[scale=0.5]{z22_31.eps } , & \psfrag{c}{{\scriptsize $ 1 $ } } \psfrag{b}{{\scriptsize $ 1 $ } } \psfrag{d}{{\scriptsize $ 0 $ } } \psfrag{e}{{\scriptsize $ 1 $ } } \includegraphics[scale=0.5]{z22_31.eps } , & \psfrag{c}{{\scriptsize $ 1 $ } } \psfrag{b}{{\scriptsize $ 1 $ } } \psfrag{d}{{\scriptsize $ 1 $ } } \psfrag{e}{{\scriptsize $ 0 $ } } \includegraphics[scale=0.5]{z22_31.eps } \end{array } \right\},\ ] ] and then @xmath39 does not have strong specification .
[ remark:6.17 ] the ideas for strong specification in this section can be applied to higher - dimensional shifts of finite type ; see @xcite .
edge coloring models are very common in statistical physics and other fields @xcite . for completeness , this section briefly discusses edge coloring .
the ideas of corner coloring in the previous sections apply to edge coloring with some modifications . for simplicity ,
only the case of two colors is considered : @xmath783 .
the results hold for all @xmath7 , @xmath19 .
the unit square lattice is still denoted by @xmath811 . for @xmath17 ,
denote the set of all local patterns with colored edges on @xmath812 over @xmath813 by @xmath814 . where @xmath822 with @xmath823 .
for @xmath637 , the higher order ordering matrix @xmath824 can be recursively defined recursively by @xmath825 similarly , @xmath826 , @xmath637 , can be defined recursively as above .
notably , comparing ( [ eqn:7.1 ] ) with ( [ eqn:2.5 ] ) , the discussion of edge coloring is very different from that of corner coloring .
accordingly , the formulae for the transition matrices and connecting operators must be changed .
given @xmath837 , for @xmath177 and @xmath827 , the connecting operator @xmath843_{2^{m}\times 2^{m}}$ ] can be defined , where @xmath844 is the cardinal number of all @xmath845-admissible local patterns in @xmath829 . furthermore , from ( [ eqn:7.2 ] ) , @xmath846 for @xmath177 and @xmath827 .
similarly , for @xmath847 , the connecting operator is denoted by @xmath848 , @xmath177 and @xmath836 . then , for @xmath859 , apply ( [ eqn:7.7 ] ) @xmath256 times to decompose @xmath860_{1\leq
\beta_{j}\leq 4,1\leq j\leq q}$ ] into @xmath861-many @xmath862 submatrices @xmath863 . the results that hold for @xmath864 are also valid for @xmath865 . notably , in edge coloring , to color a lattice @xmath870 is to color the horizonal and vertical edges that connect the vertices in @xmath14 .
the definitions of rectangle - extendability , crisscross - extendability and corner - extendable conditions @xmath60 for edge coloring are similar to those for corner coloring .
furthermore , the following theorem is obtained as theorem [ theorem:3.14 ] .
the details are omitted for brevity . 1 .
@xmath890 is non - degenerated , 2 .
there exists an @xmath884-invariant diagonal sequence @xmath891 of order @xmath426 with its invariant index set @xmath427 , 3 .
@xmath892 is primitive for @xmath429 , as above , the method of primitive commutative cycles for corner coloring in section 4 is also valid for edge coloring . for brevity ,
the detailed statements of primitive commutative cycles for edge coloring are omitted .
the results concerning strong specification in section 5 apply to the edge coloring problem and detailed statements are also omitted here . it is known edge - coloring problem can convert into corner - coloring problem , but the number of symbols will become large in general . in practice , many problems are more convenient to be studied by transforming into edge - coloring problems ; see examples [ example:7.7 ] and [ example:7.8 ] .
hence , the methods for edge coloring are discussed in this section .
[ example:7.7 ] the rule of the six - vertex model is that the number of arrows that point inwards at each vertex is two , such that @xmath895{6_1.eps } , & \includegraphics[scale=0.45]{6_2.eps } , & \includegraphics[scale=0.45]{6_3.eps } , & \includegraphics[scale=0.45]{6_4.eps } , & \includegraphics[scale=0.45]{6_5.eps } , & \includegraphics[scale=0.45]{6_6.eps } \end{array } \right\}.\ ] ] for ease of computing , the rightward and upward arrows in each pattern in @xmath13 are replaced by the digit ( color ) @xmath772 and the leftward and downward arrows in each pattern are replaced by the digit ( color ) @xmath773 .
then , the six - vertex model can be transformed into an edge - coloring problem with the colors in @xmath783 .
indeed , @xmath896{t6_1.eps } , & \includegraphics[scale=0.5]{t6_2.eps } , & \includegraphics[scale=0.5]{t6_3.eps } , & \includegraphics[scale=0.5]{t6_4.eps } , & \includegraphics[scale=0.5]{t6_5.eps } , & \includegraphics[scale=0.5]{t6_6.eps } \end{array } \right\}.\ ] ] @xmath897 , & \mathbb{h}^{e}_{2;2}= \left [ \begin{array}{cc } 0 & 0 \\ 1 & 0 \end{array } \right ] & \text{and } & \mathbb{h}^{e}_{2;3}= \left [ \begin{array}{cc } 0 & 1 \\ 0 & 0 \end{array } \right ] .
\end{array}\ ] ] is easily verified .
then , @xmath901 is an @xmath884-invariant diagonal sequence of order @xmath573 with index set @xmath902 .
clearly , @xmath903 is primitive .
hence , by theorem [ theorem:7.5 ] , @xmath864 is primitive for all @xmath51 .
since @xmath898 , @xmath899 , @xmath865 is also primitive for all @xmath51 .
therefore , the six - vertex model is mixing .
however , it does not have strong specification .
[ example:7.8 ] the rule of the eight - vertex model is that the number of arrows that point inwards at each vertex is even , such that @xmath904{6_1.eps } , & \includegraphics[scale=0.45]{6_2.eps } , & \includegraphics[scale=0.45]{6_3.eps } , & \includegraphics[scale=0.45]{6_4.eps } , & \includegraphics[scale=0.45]{6_5.eps } , & \includegraphics[scale=0.45]{6_6.eps } , & \includegraphics[scale=0.45]{6_7.eps } , & \includegraphics[scale=0.45]{6_8.eps } \end{array } \right\}.\ ] ] as in example [ example:7.7 ] , the basic set @xmath905 of the eight - vertex model can be transformed as follows ; @xmath906{t6_1.eps } , & \includegraphics[scale=0.5]{t6_2.eps } , & \includegraphics[scale=0.5]{t6_3.eps } , & \includegraphics[scale=0.5]{t6_4.eps } , & \includegraphics[scale=0.5]{t6_5.eps } , & \includegraphics[scale=0.5]{t6_6.eps } , & \includegraphics[scale=0.5]{t6_7.eps } , & \includegraphics[scale=0.5]{t6_8.eps } \end{array } \right\}.\ ] ] it can be verified that for any @xmath75 , the following admissible pattern @xmath43 on @xmath909 , @xmath78 , can be extended to @xmath910 but @xmath43 can not extend to @xmath911 . therefore , @xmath907 does not satisfy @xmath77hfc@xmath74 for all @xmath75 . before proving strong specification of @xmath914 , some notation must be defined .
a lattice @xmath915 , @xmath916 , is called a simple - curve lattice if each of its unit square lattices has exactly two shared edges that connect to two other unit square lattices of @xmath917 .
clearly , two kinds of simple - curve lattices exist .
they are ( i ) bounded simple - curve lattices and ( ii ) unbounded simple - curve lattices .
@xmath918{simple_curve1.eps } & \includegraphics[scale=0.35]{simple_curve2.eps } \\
\text{\footnotesize(i ) a bounded simple - curve lattice . } & \text{\footnotesize(ii ) an unbounded simple - curve lattice . } \end{array}\ ] ] 1 .
suppose @xmath917 is unbounded .
if the non - shared edges on @xmath917 are colored with @xmath773 and @xmath772 , then there exists exactly one pattern @xmath920 in which colors on the non - shared edges are as assumed . 2 .
suppose @xmath917 is bounded and its non - shared edges are colored with @xmath773 and @xmath772 .
let @xmath921 be the set of all interior non - shared edges and @xmath922 be the set of all exterior non - shared edges .
+ given a global pattern @xmath923 , for any bounded simple - curve lattice @xmath917 , both sums of the digits ( colors ) of @xmath924 on @xmath921 and @xmath922 are even .
+ conversely , if both sums of digits on @xmath921 and on @xmath922 are even , then there exists a pattern @xmath920 in which the colors of its non - shared edges are as assumed .
let two allowable patterns @xmath925 and @xmath926 with @xmath927 where @xmath928 and @xmath403 .
the jordan curve theorem can be used to verify the existence of @xmath929 , @xmath930 , satisfting the following ; notably , conditions ( i ) ( a)@xmath407(d ) present that the patterns @xmath43 on @xmath909 are @xmath13-admissible as in theorem [ theorem:6.2 ] . as in fig .
a.1 , conditions ( ii ) ( a)@xmath407(d ) present that @xmath43 can be extended to @xmath957 .
[ theorem : a.2 ] suppose @xmath11 is rectangle - extendable . @xmath39 has the ufp if and only if there exist a positive integer @xmath960 such that for @xmath17 and @xmath961 , @xmath962 , @xmath963 and @xmath964 with [ theorem : a.3 ] suppose @xmath11 is rectangle - extendable .
@xmath39 is corner gluing if and only if there exist a positive integer @xmath960 such that for @xmath17 and @xmath985 , @xmath986 , @xmath987 and @xmath988 with s. j. lightwood , _ morphisms from non - periodic @xmath1007 subshifts ii : constructing homomorphisms to square - filling mixing shifts of finite type _ , ergodic theory and dynamical systems * 24 * ( 2004 ) , 12271260 . n. g. markley and m. e. paul , _
maximal measures and entropy for @xmath1008 subshift of finite type _ , classical mechanics and dynamical systems , lecture notes in pure and appl . math .
* 70 * ( medford , mass . ) ( 1979 ) , 135157 . | this investigation studies topological mixing and strong specification of two - dimensional shifts of finite type . connecting operators
are introduced to reduce the order of high - order transition matrices to yield lower - order transition matrices that are useful in establishing finitely checkable conditions for the primitivity of all transition matrices .
two kinds of sufficient condition for the primitivity of transition matrices are provided ; ( i ) invariant diagonal cycles and ( ii ) primitive commutative cycles .
after primitivity is established , the corner - extendability and crisscross - extendability are introduced to demonstrate topological mixing .
in addition to these sufficient conditions for topological mixing , the hole - filling condition implies the strong specification .
all mentioned conditions are finitely checkable .
the author would like to thank the national science council , r.o.c .
( contract no .
nsc 99 - 2115-m-259 - 003 ) and the national center for theoretical sciences for partially supporting this research . ]
the author would like to thank the national science council , r.o.c .
( contract no .
nsc 98 - 2115-m-009 ) for partially supporting this research . ] |
we consider the elliott mechanism of spin relaxation@xcite in which the conduction electron spin interacts with its motion in the electric field of the host lattice described by the periodic potential @xmath26 .
this lattice induced spin - orbit coupling , given by the hamiltonian @xmath27 leads to the mixing of the originally pure spin states in the electron wave function .
thus , the conduction electron spins can now relax through ordinary momentum scattering caused by impurity atoms , for example .
the spin - relaxation rate is estimated from the spin - flip transition probability @xmath28 of an electron scattering on the impurity potential @xmath29 , and similarly the momentum - relaxation rate from the non spin - flip transition probability @xmath30 . in this picture
we neglect the soc caused by the electric field of the impurity atoms , and also assume that they give rise to much larger momentum scattering than the host ions . to formulate the above described elliott mechanism , we start from the wave functions of electrons in the periodic potential @xmath26 , which are bloch - type as @xmath31 where @xmath32 is the band index , and the bloch functions @xmath33 are lattice - periodic .
each band is at least twofold degenerate due to the presence of time - reversal symmetry .
the spin - flip transition probability @xmath28 is determined by the matrix element @xmath34 within a first - order perturbation theory , and in parallel , the non spin - flip transition element @xmath35 is related to the matrix element @xmath36 . if we assume that the impurity scattering potential @xmath29 is slowly varying on the scale of the unit cell , the matrix elements can be approximated by the factorization @xmath37 with @xmath38 , which gives the spin - flip and non spin - flip transition elements as @xmath39 within our approximation obtained by applying the fermi s golden rule .
finally , the spin- and momentum - relaxation rates are calculated as @xmath40 after a double integration of the transition probabilities on the fermi surface given with the wave vector @xmath41 .
we consider a four - band model which consists of two nearby bands each with a two - fold degeneracy with respect to the spin - up and spin - down states .
the bands are separated by an energy gap @xmath7 , and we neglect the dispersion of the bands . we found that this simplified model allows to derive analytic expression for the spin - relaxation rate for arbitrary value of the spin - orbit interaction .
the most general form of the spin - orbit interaction with inversion symmetry is discussed in ref .
: @xmath42 within the basis @xmath43 where @xmath44 and @xmath45 label the bands , and @xmath46 are soc matrix elements . for inversion and time - reversal symmetries ,
the soc matrix elements satisfy : @xmath47 and @xmath48 .
the detailed group theoretical considerations are given in the supplementary material .
for simplicity , we keep only spin - flip soc matrix elements which connect the states @xmath49 and @xmath50 and neglect the ones which mix the same spin directions .
we also consider a small zeeman energy , @xmath51 , to handle computational difficulties due to the doubly degenerate bands but we reinsert @xmath52 at the end of the calculations .
this hamiltonian of this simplified model is then given by : @xmath53 which includes kinetic , soc , and the zeeman terms .
the level structure and the relevant terms of this model are depicted in fig .
[ four_band_model ] . .
] we retain the @xmath54 index for the soc matrix elements in eq .
( [ ham-1 ] ) to indicate that it depends on both the direction and magnitude of the wave vector , even though we consider dispersionless bands .
this is required as otherwise @xmath55 and @xmath56 are orthogonal , which would give zero spin transition rate : @xmath57 .
the original energy splitting , @xmath7 , is modified by the soc as @xmath58 .
the bloch functions in eq .
( [ eq - wavef1 ] ) are obtained as the eigenstates of hamiltonian ( [ ham-1 ] ) : @xmath59 denotes mixed spin states due to the soc . for example , the lowest two states are obtained as @xmath60 which have the same form as the ones given in eqs .
( [ eq - st1 ] ) , ( [ eq - st2 ] ) derived by elliott .
the explicit expressions for the coefficients @xmath61 and @xmath62 in eqs .
( [ eq - nst1 ] ) , ( [ eq - nst2 ] ) for arbitrary value of the soc are given in the supplementary material .
in the limit of small soc , i.e. @xmath63 , we recover the perturbation result of elliott@xcite as @xmath64 and @xmath65 .
the spin - flip and non spin - flip transition probabilities can be calculated at each band based on the formulae ( [ wuddef1 ] ) and ( [ wuudef1 ] ) and the general result is given in the supplementary material .
analytic results are obtained when i ) the fermi surface is approximated with a sphere with radius @xmath66 which is a good approximation for ( @xmath67 ) , ii ) the unknown overlap integrals of the orbital part of the states and the impurity potential is estimated by constants as @xmath68 , @xmath69 , @xmath70 , @xmath71 , and iii ) the soc matrix elements are approximated by their average values at the fermi surface : @xmath72 .
the latter two approximations are justified as these quantities are band structure dependent constants after the integration over the fermi surface such as @xmath73 is introduced in eq .
( [ elliottrel1 ] ) .
with these simplifications the transition probabilities read : @xmath74 ^ 2}{4\delta(k_{\rm f})^2 \left ( \delta(k_{\rm f } ) + \delta \right)^2 } \label{eq - wupup3}.\end{aligned}\ ] ] there is no spin relaxation ( @xmath75 ) in the absence of soc , i.e. @xmath76 . since the @xmath77-integrations in eqs .
( [ gammas ] ) , ( [ gamma ] ) can not be performed in the simplified four - band model , we use the reasonable assumption that @xmath78 and @xmath79 which will be justified by the subsequent numerical calculations . as a result ,
the ratio @xmath80 reads : @xmath81 ^ 2 } .\nonumber\\ \label{eq - ratio}\end{aligned}\ ] ] in the limit of small soc , i.e @xmath82 , the ratio becomes : @xmath83 when neglecting the constants near unity , it leads to : @xmath84 reproducing the perturbation result of elliott given in eq .
( [ elliottrel1 ] ) . in this limit ,
the spin - relaxation rate is much smaller than the momentum - relaxation rate . in the opposite limit
, i.e. @xmath85 , the ratio tends to a constant as a function of @xmath6 : @xmath86 which means that the spin - relaxation rate approaches to the momentum - relaxation rate ( apart from the constants @xmath87 , @xmath88 , and @xmath89 , which are near unity ) . as a function of the soc matrix element @xmath6 with the parameter choice @xmath90 , @xmath91 , @xmath92 , @xmath93 , @xmath94 ( _ black _ ) , @xmath95 ( _ red _ ) , and @xmath96 ( _ blue _ ) .
_ horizontal dashed lines _ show the large-@xmath6 limits given by @xmath97 .
the perfect fit of the upper curve with the function in eq .
( [ eq - gs3 ] ) is shown by _
dashed green line_. ] fig .
[ fig - ratio ] shows the ratio @xmath98 as a function of the soc matrix element in the entire @xmath6-range with different values of the band - structure dependent constants .
the characteristic energy where the behavior of the ratio changes from @xmath99-dependence to saturation is approximately @xmath100 , i.e. where @xmath101 .
[ fig - ratio ] . also shows a fit with the following phenomenological formula , which was found to well approximate the result in the entire parameter range : @xmath102 with @xmath103 and @xmath104 being constants of order of unity .
we note that the simple phenomenological formula ( [ eq - gs3 ] ) is already implied by the analytic result given in eq .
( [ eq - ratio ] ) after some manipulations ( see the supplementary material ) .
we also discuss the @xmath0-factor shift which is given by the expectation value of the orbital momentum in the bloch state through the zeeman term . a straigthforward calculation ( detailed in the supplementary material )
yields for @xmath105 and @xmath106 , i.e. when the magnetic field is perpendicular or parallel to the @xmath107 axis defined by the soc hamiltonian ( eq . ) , respectively : @xmath108 the ey theory predicts an isotropic @xmath0-factor and the anisotropy in our results is due to the omission of the @xmath109 and @xmath110 terms in eq . .
to test the elliott - relation we proceed with @xmath105 as : @xmath111 which becomes @xmath112 for @xmath82 reproducing the elliott result . combining expression ( [ eq - g_shift ] ) with ( [ eq - gs3 ] )
gives : @xmath113 which recovers the elliott relation given in eq .
( [ elliottrel3 ] ) for arbitrary values of @xmath6 . to complement the results of the model hamiltonian calculations
, we consider a nearly - free electron model where spin and momentum scattering can be calculated .
the hamiltonian of the problem is : @xmath114 we solve it with the pseudopotential method which approximates the lattice potential @xmath115 with its first few fourier components @xmath116 as @xmath117 in the calculations we take @xmath118 with @xmath119 $ ] , which corresponds to a zincblende structure with identical atoms ( example si or ge ) , i.e. with inversion symmetry retained ( for further details , see supplementary material ) .
this model has some advantages and shortcomings when calculating spin relaxation .
an advantage is that the soc is obtained directly from the lattice potential thus it allows to perform a numerical analysis of spin relaxation .
in addition , the pseuodpotential - based soc automatically accounts for the crystal symmetry which is found in the soc , too .
however , the pseudopotential - based soc is known to underestimate the experimental soc as it considers the smooth potential of the valence electrons only and neglects the strongly oscillating potential near the atomic core@xcite . to solve this problem ,
the pseudopotential - based soc is magnified with a scaling parameter with a typical value of @xmath120 .
we chose @xmath121 as it account quantitatively well for the @xmath122 soc gap in ge@xcite .
in addition , we also introduce a _ tuning parameter _
, @xmath123 , for the soc interaction in order to study the soc dependent spin relaxation . as a result
, the soc hamiltonian reads as @xmath124 in eq .
( [ h - so ] ) . along the @xmath125 $ ] direction .
@xmath126 and @xmath127 denote the valence and conduction bands ( each with additional spin degeneracy @xmath128 ) , respectively . ]
we use a pseudopotential parameter set which reproduces well the conduction and valence bands of ge .
the band structure is shown in fig .
[ fig - bandstructure ] for @xmath129 .
considering the nearby valence and conduction bands denoted as @xmath130 and @xmath127 in fig .
[ fig - bandstructure ] , respectively , the characteristic features of metals with hybridized @xmath131-@xmath132 orbitals can be discussed .
namely , the orbital parts of the electron wave functions at @xmath133 ( @xmath20 point ) without soc have the symmetries @xmath134 and @xmath135 which mean @xmath131- and @xmath132-like orbitals , respectively .
symmetry operations acting on the spin wave functions have to be considered in the presence of soc , which can be handled by double group irreducible representations . at @xmath133 , the @xmath132-symmetric ( @xmath136 ) valence band with spin splits into a four- and a two - fold degenerate state , separated by the soc gap .
these split states can be labeled according to the total angular momentum operator as @xmath137 and @xmath138 , respectively . as we move away from the @xmath133 point , the four - fold degenerate valence band splits further but a two - fold degeneracy is kept due to the presence of time - reversal symmetry . in the vicinity of the @xmath133 point , all of the valence and conduction bands have mixed @xmath131-@xmath132 character due to the hybridization .
using eqs .
( [ wuddef1 ] ) , ( [ wuudef1 ] ) , we derive the non spin - flip and spin - flip transition matrix elements in the conduction band @xmath127 as @xmath139 and in the upper valence band @xmath140 as @xmath141 with @xmath142 that is not specified , but the @xmath54-dependence of the wave functions is obtained accurately in the pseudopotential approximation . the momentum- and spin - relaxation rates are obtained according to eqs .
( [ gammas ] ) and ( [ gamma ] ) by averaging over the transition matrix elements numerically with respect to @xmath54 and @xmath143 on the fermi surface with energy @xmath144 as : @xmath145 at @xmath146 in the valence band ( _ upper panel _ ) and in the conduction band ( _ lower panel_).,title="fig : " ] at @xmath146 in the valence band ( _ upper panel _ ) and in the conduction band ( _ lower panel_).,title="fig : " ] figure [ fig - pseudogammagammas ] shows the spin- and momentum - relaxation rates calculated in the pseudopotential approximation both for the upper valence band @xmath140 and conduction band @xmath127 as a function of the spin - orbit coupling , @xmath123 .
we took @xmath146 in the calculation and averaged the transition probabilities given in eqs .
( [ eq-1vuu ] ) , ( [ eq-1vud ] ) over 30 points at the fermi surface to obtain the relaxation rates , @xmath20 and @xmath19 . the seemingly different behavior of the relaxation rates in the valence and conduction bands can be consistently explained . as it is shown in fig .
[ fig - bandstructure ] , the relevant band gap , i.e. the characteristic energy defined as @xmath147 for the valence state and @xmath148 for the conduction band , is quite different for the two cases .
namely , in the case of the upper valence state , @xmath140 , there is another valence state @xmath149 in its close vicinity , while the closest state to the conduction band @xmath127 is @xmath140 , which is much further away , which leads to @xmath150 .
thus , by increasing the soc in the calculations , the range @xmath151 can be reached for the valence band and we indeed find the saturating behavior for both @xmath20 and @xmath19 in this range in well agreement with the result of the four - band model . on the other hand , in the case of the conduction band , the relation @xmath152 holds in the entire range of @xmath123 which we study .
a much larger @xmath123 would lead to a rearranged band structure .
therefore , for the conduction band in our model , we observe the perturbative ( or elliott - yafet ) behavior @xmath153 and @xmath154 .
we find that the numerical data for the ratio @xmath98 follows the approximate formula obtained in the model hamiltonian calculation for both the valence and conductions bands .
namely , the approximate formula given in eq .
( [ eq - gs3 ] ) fits well the ratios by taking the numerical data @xmath155 and @xmath156 for the band gap @xmath157 for the valence and conduction band , respectively .
figure [ fig - pseudoratio ] shows the fit for the ratio @xmath98 as a function of @xmath158 with this model .
calculated in the pseudopotential approximation for the valence and conduction bands as a function of soc strength ( @xmath123 ) at @xmath146 with the formula ( [ eq - gs3 ] ) . ]
the good agreement between the numerically obtained data and the phenomenological formula ( [ eq - gs3 ] ) obtained in the four - band model calculation supports the previous assumption that @xmath78 and @xmath79 .
in addition , we find it compelling that the two different approaches result in @xmath98 ratios as a function of the soc which are both well approximated by the formula in eq . .
this as we believe , summarizes the elliott - yafet theory generalized for the case of strong spin - orbit coupling . among all the metals where
experimental data on spin relaxation exists , au is particularly suited to test the validity of the above discussion as it has the strongest soc .
in addition , gold has a single conduction electron per unit cell , i.e. its fermi surface does not extend beyond the first brillouin zone .
it is known that description of spin relaxation is more complicated for metals with two or more conduction electrons per unit cell@xcite ( e.g. mg or al ) , where the fermi surface crosses the brillouin zone boundaries giving rise to the so - called spin relaxation `` hot - spots '' . when hot - spots are present , the elliott relation is known to break down even though the elliott - yafet theory remains valid@xcite .
monod and beuneu found @xcite the soc strength in au as @xmath159 .
it is therefore an appropriate candidate for a case where the perturbative treatment of the soc breaks down .
monod and jnossy reported in ref .
the electron spin resonance linewidth , @xmath23 , which is used to obtain @xmath19 .
@xmath20 is calculated from the resistivity , @xmath160 , data in ref . through : @xmath161 , where @xmath162 is the vacuum permittivity and @xmath163 is the plasma frequency of gold @xcite . for both sets of data ,
the residual linewidth and resistivity was subtracted as both quantities are known to obey the matthiessen s rule @xcite , i.e. that the respective residual and temperature dependent scattering rates are additive .
, in au from ref . as compared to the momentum scattering rate , @xmath20 .
note the factor 10 difference in the energy scales for the two quantities . ]
the comparison , shown in fig .
[ gold_relaxation ] , shows that the measured spin - relaxation rate approaches the momentum - relaxation rate within an order of magnitude such as our theoretical result suggests .
the difference between the predicted @xmath164 and the observed @xmath165 could be due to the band structure dependent constants near unity in eq . .
however the observation of a nearly equal @xmath19 and @xmath20 is itself surprising as to our knowledge it has not been noted elsewhere that the mergence of spin and momentum relaxation times could be realized for a real material .
we finally test empirically the validity of the elliott relation in eq . , i.e. @xmath166 in gold .
monod and jnossy found @xmath167 which gives @xmath168 , which is again in agreement with the theoretical prediction given the presence of the band structure dependent constant factors .
this means that the elliott relation remains a valid empirical tool to test whether the elliott - yafet theory applies for a given system .
we generalized the elliott - yafet theory of spin relaxation in metals with inversion symmetry for the case of arbitrary value of the spin - orbit coupling .
we applied two different approaches , exact diagonalization of a model four - band hamiltonian ( without dispersion ) and numerical calculation of the spin relaxation time for a pseudopotential approach .
the two methods give a qualitatively similar result , which is summarized in eq . .
a calculation of the @xmath0-factor in the four - band model shows that the empirical elliott - relation , which links the @xmath0-factor and spin - relaxation rate , is retained even for strong soc .
our result predicts that spin and momentum relaxation times can have similar orders of magnitude .
we show that this situation has been already observed in au .
our result is an important step toward the unified theory of spin relaxation including the strength of the spin - orbit coupling .
work supported by the erc grant nr .
erc-259374-sylo and by the hungarian scientific research funds no .
a. k. acknowledges the bolyai program of the hungarian academy of sciences .
ak performed the calculations , ls presented the group theoretical considerations of the soc , and fs discussed the comparison between the theory and experiment .
all authors contributed to writing the manuscript .
* competing financial interests : * the authors declare no competing financial interests .
10 url # 1`#1`urlprefix[2]#2 [ 2][]#2 & .
_ _ * * , ( ) . .
_ _ * * , ( ) . .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . & _ _ ( , ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . .
_ _ * * , ( ) . , & .
_ _ * * , ( ) .
the hamiltonian of the spin - orbit interaction given in eq .
( [ ham - so ] ) in the main paper is restricted by the following three properties : i ) it is self adjoint , ii ) it can be written in the form of @xmath169 ( for every @xmath170 m @xmath171 is a kinetic operator ( @xmath172 ) and @xmath173 denotes tensorial product ) , and iii ) it is invariant to the product of inversion and time - reversal operations .
the effect of inversion ( @xmath174 ) and time - reversal ( @xmath175 ) operations on our basis are given by : @xmath176 where @xmath177 denotes the product of inversion and time reversal .
note that time reversal allows arbitrary @xmath178 factors for the two kinetic basis states , @xmath179 and @xmath180 , depending how the basis were chosen .
using the notation used in main text , the second property yields @xmath181 . the third property implies : @xmath182 which in our notation for the matrix elements reads : @xmath183 this , together with the second restriction yields the result quoted in the main text : @xmath184
the coefficients of the mixed spin states in eqs .
( [ eq - nst1 ] ) and ( [ eq - nst2 ] ) of the main paper are obtained as @xmath185 the spin - flip and non spin - flip transition elements in the lowest band calculated from the eigenstates read : @xmath186 where we introduced the notations @xmath187 , @xmath188 , and @xmath189 for the unknown overlap of states with different wave vectors @xmath190 . with the assumptions detailed in the main text
, we obtain for the ratio of spin- and momentum - relaxation rates as : @xmath81 ^ 2 } .\nonumber\\ \label{eq - ratio_suppl}\end{aligned}\ ] ] this can be rewritten as : @xmath191}. \nonumber\\ \label{eq - gs2}\end{aligned}\ ] ] the last term in the denominator of eq .
( [ eq - gs2 ] ) is proportional to @xmath192 for @xmath82 in which limit it can be neglected because the leading soc term is @xmath99 , and proportional to @xmath99 for @xmath85 where it can be summed up together with the original @xmath99 term .
this justifies the form of the empirical fitting function for @xmath98 in the main text .
the @xmath0-factor is obtained from the zeeman hamiltonian containing both the spin and orbital term with the respective operators ( @xmath193 and @xmath6 ) : @xmath194 \right .
\nonumber\\ & + & \left . b_{z }
( l_{z}+g_0 s_{z } ) \right ) .\label{eq - zeeman}\end{aligned}\ ] ] the spin - flip soc matrix element in our four - band model connects the states @xmath49 and @xmath50 , which means that only the orbital momentum operators @xmath195 and @xmath196 are to be considered in eq .
( [ eq - zeeman ] ) .
let us change the direction of the magnetic field in the @xmath197 plane from the @xmath107 axis , which gives @xmath198 , and @xmath199 .
the expectation values of the zeeman term between the states @xmath200 and @xmath201 given in eqs .
( [ eq - nst1 ] ) , ( [ eq - nst2 ] ) in the main paper are calculated as @xmath202.\end{aligned}\ ] ] diagonalizing this @xmath203 problem , we obtain the modified zeeman energies as @xmath204 with @xmath205
we take the first five reciprocal vectors @xmath206 : @xmath207 $ ] , where the degeneracy of each reciprocal vector is indicated , in the fourier expansion of the lattice potential , which leads to 102 lowest - lying states ( @xmath208 , where 2 corresponds to the spin degrees of freedom)@xcite . taking the electron wave functions given in eq .
( [ eq - wavef1 ] ) of the main paper , the schrdinger equation simplifies for each @xmath54 to the eigenvalue problem of the hamiltonian matrix @xmath209 obtained from eq .
( [ ham ] ) of the main paper , where @xmath210 \nonumber\\ & \cdot & { \boldsymbol \sigma}_{\sigma , \sigma^{\prime } } \left [ - i v({\boldsymbol g}- { \boldsymbol g}^{\prime } ) \right ] .
\label{secular - e2}\end{aligned}\ ] ] | we generalize the elliott - yafet ( ey ) theory of spin relaxation in metals with inversion symmetry for the case of large spin - orbit coupling ( soc ) .
the ey theory treats the soc to the lowest order but this approach breaks down for metals of heavy elements ( such as e.g. caesium or gold ) , where the soc energy is comparable to the relevant band - band separation energies .
the generalized theory is presented for a four - band model system without band dispersion , where analytic formulae are attainable for arbitrary soc for the relation between the momentum- and spin - relaxation rates . as an extended description ,
we also consider an empirical pseudopotential approximation where soc is deduced from the band potential ( apart from an empirical scaling constant ) and the spin - relaxation rate can be obtained numerically .
both approaches recover the usual ey theory for weak soc and give that the spin - relaxation rate approaches the momentum - relaxation rate in the limit of strong soc .
we argue that this limit is realized in gold by analyzing spin relaxation data .
a calculation of the @xmath0-factor shows that the empirical elliott - relation , which links the @xmath0-factor and spin - relaxation rate , is retained even for strong soc .
the prospect of using electron spins as information carriers , ( a field known as spintronics ) renewed the interest in the fundamental desciption of spin relaxation in metals and semiconductors .
spintronics devices rely on the controlled creation and readout of a non - equilibrium net spin population .
spin relaxation in turn characterizes how rapidly the non - equilibrium spin population decays , knowledge and theoretical description of spin relaxation is therefore of central importance .
the first experiment in the field dates back to 1955 ( ref . ) , which reported the first electron spin resonance in metals . the first proper theoretical description of spin relaxation in metals with inversion symmetry was provided by elliott @xcite , which was later generalized to lower temperatures and for various relaxation mechanisms by yafet @xcite .
the elliott - yafet ( ey ) theory of spin relaxation is valid for metals and semiconductors ( metals in the following ) with i ) inversion symmetry , ii ) weak spin - orbit coupling ( soc ) , and iii ) low quasi - particle scattering rate @xcite .
when the inversion symmetry is broken , the so - called dyakonov - perel mechanism describes the spin relaxation @xcite .
the case of sizeable quasi - particle scattering was described before @xcite but strong soc has not been considered yet .
the conventional ey theory exploits that in the presence of inversion symmetry , the spin - up and spin - down states remain degenerate as a result of time - reversal invariance ( or kramers theorem ) until the later is broken by e.g. a magnetic field .
the presence of a nearby band gives rise to an admixture of the spin - up / down states in the conduction band , while the energy degenerancy is retained . as a result ,
the ey desciption is a four - band theory .
the admixed states read : @xmath1e^{i{\boldsymbol k}\cdot{\boldsymbol r } } , \label{eq - st1}\\ { | \widetilde{\downarrow } \rangle}_{\boldsymbol k } & = & \left[a^{\ast}_{-{\boldsymbol k}}\left({\boldsymbol r}\right ) |
\downarrow \rangle - b^{\ast}_{-{\boldsymbol k}}\left({\boldsymbol r}\right ) |
\uparrow \rangle\right]e^{i{\boldsymbol k}\cdot{\boldsymbol r}},\label{eq - st2}\end{aligned}\ ] ] where @xmath2 and @xmath3 are the pure ( unperturbed ) spin states and @xmath4 , @xmath5 are the perturbed bloch states . in the first order of the soc ,
the coefficients are given by the @xmath6 matrix element of the soc for the conduction and the near lying band , and the corresponding energy separation @xmath7 as : @xmath8 .
elliott showed with a first - order time dependent perturbation treatment @xcite that the usual momentum scattering induces spin transitions for the admixed states , i.e. a spin relaxation . with @xmath9 and @xmath10 used for the spin- and momentum - relaxation rates ( @xmath11 and @xmath12 are the corresponding relaxation times ) , respectively : @xmath13 where @xmath14 is a band structure dependent constant near unity .
elliott further showed that the magnetic energy of the admixed states is different from that of the pure spin - states , i.e. there is a shift in the electron @xmath0-factor : @xmath15 where @xmath16 is the free electron @xmath0-factor , @xmath17 is another band structure dependent constant near unity .
eqs . and
gives the so - called elliott relation @xmath18 which links three empirical measurables ; @xmath19 , @xmath20 , and @xmath21 .
in practice , the spin - relaxation time is obtained for metals from conduction electron spin resonance ( cesr ) measurements @xcite as : @xmath22 . here @xmath23 is the homogeneous esr line - width and @xmath24 is the electron gyromagnetic ratio .
the cesr resonance line position yields the @xmath0-factor shift .
monod and beuneu tested empirically @xcite the validity of eq . and found that the atomic soc induced energy splittings approximate well the appropriate matrix elements . for most elemental metals @xmath25
holds but for cs , cu , ag , and in it is about 0.1 and for other metals 0.2 ( pb ) , 0.3 ( hg and sn ) with a 0.9 as the largest value in au .
clearly , the validity of a first - order perturbation treatment of the soc for these cases is questionable .
this motivates us to revisit the ey theory for the case when the soc is not small compared to other energy scales ( kinetic energy or band separations ) . here
, we discuss the most general form of the soc hamiltonian which is applicable for the ey model .
we proceed with a simplified four - band hamiltonian and solve the problem of spin relaxation in the first order of the scattering but exactly for the soc .
we find that the conventional ey result is recovered for weak soc .
a calculation of the @xmath0-factor yields the elliott - relation for arbitrary strength of the soc .
a numerical calculation is also presented for the spin relaxation in the framework of a pseudopotential approximation for the conduction electrons , where soc is obtained directly from the potential together with an empirical scaling factor .
the numerical result qualitatively returns the result of the ey model calculations .
the most important prediction is that the spin scattering time approaches the momentum scattering for strong soc .
we revisit previous experimental data on au by monod and jnossy @xcite and show that this effect was observed already but it avoided the attention . |
the lie algebras @xmath3 of infinitesimal cr - automorphisms of various _ model_in the sense of beloshapka cauchy - riemann ( cr ) generic submanifolds @xmath4 of codimension @xmath1 and of cr dimension @xmath2 are key algebraic features which open the door to a potentially infinite number of new cartan geometries . indeed , the knowledge of @xmath5 and of its isotropy subalgebras @xmath6 at points @xmath7 strongly intervenes when one endeavours to build cartan connections associated to _ all _ _ geometry - preserving _ real analytic deformations @xmath8 of a chosen model .
it is well known that procedures due to cartan and to tanaka exist to perform such constructions ( _ see _
@xcite in a cr context ) , although the practical outcome appears most of the times to be delicate and unpredictable .
in addition , there has recently been an increasing interest towards complete classification of cr - generic submanifolds according to their algebras of infinitesimal cr - automorphisms .
notably , beloshapka and kossovskiy @xcite classified homogeneous cr - generic submanifolds @xmath9 of cr dimension @xmath10 , while a bit before , fels and kaup @xcite classified the levi - degenerate homogeneous @xmath11-nondegenerate hypersurfaces @xmath12 of dimension five . far beyond for what concerns the appearing ( co)dimensions , beloshapka and his school in the last decade devised general procedures to cook up _ model cr - generic submanifolds _ that , most often , have rigid polynomial defining equations .
but when some concrete equation of a cr - generic manifold is given , one unpleasant obstacle happens to be the _ complexity _ and the _ length _ of the computations that are required to attain the full lie algebras of infinitesimal cr - automorphisms ( _ see _
@xcite ) , an obstacle which motivates the present work . by @xmath2 and @xmath13 throughout this paper , we shall mean the cr dimension and the codimension of a real analytic local cr - generic submanifold @xmath14 passing through a reference point , say through the origin .
beloshapka in @xcite , introduced a significant class of cr - generic manifolds with several _ nice _ properties ( _ see _ _ e.g. _ theorem 14 in @xcite ) , that he denoted by @xmath15 and called _
universal models_. he also computed the algebras of infinitesimal cr - automorphisms associated to the simplest model @xmath16 in @xcite , and he derived some interesting stability results .
subsequently , shananina computed such algebras for the universal models @xmath17 with @xmath18 in @xcite and derived expected consequences too ( _ see _ theorem 1 , propositions 1 and 2 and corollary 1 of @xcite ) . finally , in @xcite , mamai studied lie algebras of infinitesimal cr - automorphisms associated to some universal models @xmath17 for @xmath19 , though without presenting details . as far as the authors are aware of , in cr dimension @xmath20 , no higher
codimensions have been explored . understandably , as much as the dimension or codimension of a cr - generic submanifold @xmath21 increases , the size and the complexity of the corresponding computations of @xmath22 growths rapidly , hence an automation would be welcome , even a partial one .
serendipitously , an important , recently renewed , much related subject has been extensively studied : _ differential ( computer ) algebra _ , _ cf .
_ @xcite . there
, one employs algebraic tools like the ones of grbner bases theory in order to solve systems of partial differential equations , or in order to find the lie symmetries of differential equations , a vast area too .
over the past few years , several relative packages have been developed within various computer algebra systems .
for instance , two maple packages were designed in this direction , firstly differentialalgebra by boulier , lazard , ollivier , petitot ( @xcite ) , and secondly diffgrob2 by mansfield ( @xcite ) . in this paper
, we aim to provide an effective algorithm in order to compute the lie algebras @xmath23 associated to model real analytic cr - generic submanifolds @xmath24 , by the valuable means of differential algebra , supplemented by some new operations . for this purpose
, we shall denote by @xmath25 the pde systems that have constant ( complex ) coefficients , precisely as the ones we shall encounter several times . since these systems admit complex - valued equations , we equip at first the fundamental _
ritt s reduction theorem _ with an operator , which we call the _ bar - reduction _ , and we extend it to gain the following conclusion , more appropriate to treat the arising complex - valued @xmath25 pde systems ( _ see _ theorem [ modritt - thm ] ) : ( extended ritt s reduction theorem ) consider a differential polynomial ring @xmath26 over the field of complex numbers , let @xmath27 be the set of its derivation operators , and let ` @xmath28 ' be a ranking over @xmath29 .
furthermore , assume that @xmath30 is a @xmath25 differential polynomial and let @xmath31 be a finite set of @xmath25 differential polynomials
. then , there exists @xmath32 , and for each @xmath33 , there exists @xmath34 and @xmath35 satisfying the following conditions : * @xmath36 * for each @xmath37 appearing in this summation we have : @xmath38 * no term of @xmath39 is a derivation of either @xmath40 or @xmath41 for each @xmath42 . using this result , we modify the _ rosenfeld - grbner _ algorithm to an algorithm we call the _ lrg _ algorithm ( algorithm [ lrg ] below ) in such a way that it works appropriately in the case of our @xmath25 complex - valued pde systems .
it is noteworthy that , for the sake of simplicity , we study only _ rigid _ cr - generic submanifolds @xmath43 ( _ see _ subsection [ mainstrategy ] for more details ) , because the arising pde systems _ are _ @xmath25 ; of course , a more general algorithm could be described for non - rigid @xmath44 , the pde system still being linear , though with non - necessarily constant coefficients .
specifically , we employ our _ lrg _ algorithm for computing _ explicitly _ the lie algebras of infinitesimal cr - automorphisms associated to the following three real analytic rigid cr - generic submanifolds @xmath45 , represented in coordinates @xmath46 as the graphs of the following _ shared _ six defining equations : @xmath47 together with : @xmath48 @xmath49 @xmath50 for this , we use the maple package differentialalgebra to carry out the necessary computations .
nevertheless , these computations involve complex integers and functions and hence it is not possible to perform directly this package in this respect .
that is why , we need also to utilize a new reduction called _ bar - reduction _ that uses complex conjugation to obtain a full remainder ( _ see _ section [ solving pde ] ) . executing long computations
, we achieve the following results .
of course , the main interest of our algorithmic partly automatize approach is to open the door to a wealth of other examples taking inspiration from beloshapka s universal models .
the lie algebras of infinitesimal cr - automorphisms @xmath51 , @xmath52 and @xmath51 of the three real analytic generic cr - generic submanifolds @xmath53 and @xmath54 of @xmath55 are of dimensions @xmath56 and @xmath57 , respectively and are generated by the @xmath58-linearly independent real parts of the following collections of holomorphic vector fields : @xmath59 @xmath60 @xmath61 the paper is organized as follows .
section [ differentialalgebra ] contains an overview of the necessary background concerning the theory of differential algebras . in section [ solving pde ] , we present the extended ritt s reduction algorithm and we show how to utilize the differential algebraic tools to resolve a system of partial differential equations . in section [ general description ] , we present a general method / strategy to compute the lie algebra of infinitesimal cr - automorphisms of _ arbitrary _ generic real analytic cr - generic submanifolds @xmath62 .
section [ computations ] is devoted to compute in detail the lie algebra of infinitesimal cr - automorphisms @xmath51 , associated to the first model @xmath63 .
however , we do not present the intermediate computations of @xmath64 and @xmath65 since they are similar to those of @xmath63 and offer no new aspect . tables of lie brackets appear at the end .
in this section , we present a brief overview of basic definitions , notation and results in differential algebra .
two extensive surveys of this subject are : @xcite .
an operator @xmath66 over the algebraic ring @xmath67 is called a _ derivation operator _ , if for each @xmath68 we have : @xmath69 a _ differential ring _ is a pair @xmath70 where @xmath67 is a ring equipped with a collection @xmath71 of commuting derivations operators over it , satisfying : @xmath72 for simplicity , we suppress the dependence on @xmath73 in the notation and denote a differential ring just by @xmath67 . if @xmath74 , then @xmath67 is called an _
ordinary _ differential ring ; otherwise it will be called _
partially_. an algebraic ideal @xmath75 of @xmath67 is called a _ differential ideal _ when it is closed under the action of derivations of @xmath67 , namely @xmath76 for each @xmath77 and @xmath78 .
the ring of polynomials @xmath79 $ ] over the variables @xmath80 with rational coefficients together with the set of operators @xmath81 is a differential ring .
let @xmath67 be a differential ring with @xmath71 .
here , we introduce a collection of notations in differential algebra via the following itemized list : * we denote by @xmath27 the free multiplicative commutative semigroup generated by the elements of @xmath73 , namely @xmath82 each element @xmath83 of @xmath27 is called a _ derivation operator _ of @xmath67 and furthermore the sum @xmath84 is called the _ order _ of @xmath85 .
then @xmath86 is said to be a derivative of @xmath87 of order @xmath88 . * for an arbitrary subset @xmath89 of @xmath67 , set @xmath90 .
it is the smallest subset of @xmath67 containing @xmath89 which is stable under derivation .
* an algebraic ideal of @xmath67 is called a _ differential ideal _ , if it is closed under the derivation operators .
we denote by @xmath91 and @xmath92 $ ] respectively , the smallest algebraic and differential ideals of @xmath67 containing @xmath89 .
in fact , @xmath92=(\theta s)$ ] .
this fact provides an algebraic approach to differential ideals which enables one to employ algebraic means . * for a field of characteristic zero @xmath93 , a _ differential polynomial ring _ : @xmath94\ ] ] is the usual commutative polynomial ring generated by @xmath29 over @xmath93 , where @xmath95 is the set of _ differential indeterminate_. * for two certain derivatives @xmath96 and @xmath97 of a same differential indeterminate @xmath98 , we denote by @xmath99 the _ least common derivative _ between @xmath100 and @xmath97 , easily seen to be : @xmath101 in this paper we let @xmath93 be a differential field of characteristic zero .
let @xmath102 be a differential polynomial ring with the set of indeterminates @xmath103 .
a _ ranking _
@xmath28 is an ordering over @xmath29 compatible with the derivation act over @xmath29 , in the sense that for each derivation @xmath104 and for each @xmath105 we have : * @xmath106 , * @xmath107 . for each @xmath108 and @xmath109 , a ranking @xmath28 for which the statement @xmath110 implies that @xmath111 is called _
orderly_. simultaneously ,
if the assumption @xmath112 gives @xmath113 , then @xmath28 is called _
elimination_. moreover , for a fixed ranking @xmath28 over @xmath29 and for a differential polynomial @xmath114 , the _ leader _ @xmath115 of @xmath116 is the highest derivative appearing in @xmath116 with respect to @xmath28 . if @xmath117 and @xmath118 is the degree of @xmath98 in the expression of @xmath116 then , the _ initial _
@xmath119 is defined to be the coefficient of @xmath120 in @xmath116 .
finally , @xmath120 is called the _ rank _ of @xmath116 , denoted by @xmath121 .
each differential polynomial ring @xmath102 can be considered as the conventional polynomial ring @xmath122 $ ] whose indeterminates are derivations of @xmath67 .
this enables one to use the conventional algebraic tools and get useful information about the differential polynomial ring and its differential ideals . in this section
, we employ the _ rosenfeld - grbner _ algorithm to discuss a system of partial differential equations , using algebraic operations . for a pde system @xmath123
, the rosenfeld - grbner algorithm presents the radical differential ideal generated by @xmath124 as an intersection of a finite number of differential ideals which are called _ regular differential ideals_. those are some differential ideals @xmath75 represented by a
_ canonical representative _ @xmath125 , _ i.e. _ a set of differential polynomials which depends only on @xmath75 and the given ranking . a canonical representative of the differential ideal @xmath75 helps to solve ideal membership problem , which is a key computational tool to analyze a pde system .
an implementation of the rosenfeld - grbner algorithm is available in the maple package diffalg , and it was recently renovated into the package differentialalgebra .
one of the main contributions of this paper is the use of rosenfeld - grbner algorithm followed by performing some further algebraic manipulations for considering our pde systems .
it is worth emphasizing that the pde systems that we deal with in this paper are _ linear and admit complex equations_. then , we have to equip the rosenfeld - grbner algorithm with a certain operator which enables to treat with such systems .
moreover , in the considerably significant class of _ rigid _ cr - generic submanifolds ( _ see _ the end of section [ general description ] for definition ) ,
the under consideration pde systems are not only complex and linear but also _ with constant coefficients _ and it is therefore reasonable to consider such systems more seriously . for brevity ,
let us call this type of systems by @xmath25 systems and also similarly , let us call each linear differential polynomial with constant coefficients @xmath25 polynomial one , in general .
for example , as an element of @xmath126 $ ] , the polynomial @xmath127 is linear while its coefficients are not constant . ] .
computation with the rosenfeld - grbner algorithm are comparatively less expensive .
furthermore , in this case _ there is no need longer to decompose the differential ideal generated by the system into regular differential ideals _ ( a complicated aspect of the general algorithm ) .
let us therefore recall the ritt s reduction algorithm , restricted to @xmath25 differential polynomials .
therefore , the reduction algorithm described here shall be a _ weak version of the ritt s reduction algorithm _ in comparison with the version of @xcite .
let us recall the definition of partial divisibility for differential polynomials .
consider two @xmath25 differential polynomials @xmath128 and @xmath129 .
we say that _
@xmath129 reduces @xmath128 due to the ritt s reduction algorithm _
, whenever there exists a certain derivation @xmath85 with @xmath130 . in this case
, the result of reduction is : @xmath131 when @xmath85 is proper , we call @xmath39 , the _ partial remainder _ of @xmath128 on division by @xmath129 .
one notices that in this definition , if @xmath132 then @xmath85 must be the identity element of @xmath27 and the ritt s reduction coincides with the conventional division algorithm for multivariate polynomial rings .
( ritt s reduction theorem ) [ ritt - thm ] consider a differential polynomial ring @xmath133 over a field @xmath93 of characteristic zero , let @xmath27 be the set of derivation operators and let ` @xmath28 ' be a ranking over @xmath29 .
furthermore , assume that @xmath30 is a @xmath25 differential polynomial and let @xmath31 be a finite set of @xmath25 differential polynomials
. then , there exists @xmath32 , and for each @xmath33 , there exists @xmath134 and @xmath135 satisfying the following conditions : * @xmath136 * @xmath137 , for each @xmath37 appearing in the summation , * no term of @xmath39 is a derivation of @xmath138 for each @xmath42 . here
, the differential polynomial @xmath139 is called the _ remainder _ of @xmath116 on division by @xmath31 . in order to prove this theorem ,
let us display first the ritt s reduction algorithm ensuing its assertion .
then , the correctness and termination of this algorithm proves theorem [ ritt - thm ] .
[ ritt ] the termination of this algorithm follows from the well - ordering property of @xmath28 ( _ cf . _
namely as one observes , the rank of @xmath140 decreases as one goes along the steps of the algorithm and hence , it terminates after a finitely many steps , when we will have @xmath141 .
now let us consider the correctness of the algorithm .
for this , we claim that the equality : @xmath142 holds at each step of the algorithm .
if we prove this claim then , the final value @xmath141 of @xmath140 can concludes the assertion . to prove the claim ,
we consider two cases : case 1 : : if a division occurs by a polynomial , say @xmath143 , then the right hand side of ( [ eq:1 ] ) is equal to : @xmath144 with @xmath145 .
therefore , it visibly is still equal to @xmath116 .
case 2 : : if no division arises , then the right hand side of ( [ eq:1 ] ) is equal to : @xmath146 and this is also equal to @xmath116 . in both cases ,
the equation ( [ eq:1 ] ) is satisfied at each step of the algorithm .
moreover , one convinces oneself that the last two assertions of the theorem hold according to the structure of this algorithm .
one should notice that , as long as the field @xmath93 is the one of real numbers @xmath58 , then the above version of the ritt s reduction algorithm works as well to compute the remainder of the division of a @xmath25 differential polynomial by a finite system of @xmath147 pde s .
however , the coefficients of the pde systems that we consider in this paper belong to the field @xmath148 with @xmath149 . in this case
, we need to perform also the _ complex conjugation _ to obtain a full remainder . for this , we need the following definition and theorem .
let @xmath150 be a differential polynomial ring with @xmath151 for @xmath152 as the unknown functions .
we define the _ bar operation _ @xmath153 by : * @xmath154 for each @xmath155 , * @xmath156 for each @xmath152 . as a matter of fact ,
the bar operator is compatible with the derivations ; namely , for each ( real ) @xmath157 and each @xmath152 , one has : @xmath158 this allows one to insert the bar reduction operator in the ritt s reduction algorithm .
[ modritt - thm ] ( extended ritt s reduction theorem ) consider a differential polynomial ring @xmath26 over the field of complex numbers , let @xmath27 be the set of its derivation operators , and let ` @xmath28 ' be a ranking over @xmath29 .
furthermore , assume that @xmath30 is a @xmath25 differential polynomial and let @xmath31 be a finite set of @xmath25 differential polynomials .
then , there exists @xmath32 , and for each @xmath33 , there exists @xmath34 and @xmath35 satisfying the following conditions : * @xmath36 * for each @xmath37 appearing in this summation we have : @xmath38 * no term of @xmath39 is a derivation of either @xmath40 or @xmath41 for each @xmath42 . here , the differential polynomial @xmath39 is called the _ full - remainder _ of @xmath116 on division by @xmath31 .
the proof of this theorem is similar to that of theorem [ ritt - thm ] .
let us display the following algorithm like algorithm [ ritt ] extended by the bar reduction .
[ mod - ritt ] as mentioned before , to solve a pde system @xmath124 , we use the rosenfeld - grbner algorithm to decompose the radical of @xmath159 $ ] , into some new pde systems , presented by explicit generators .
these generators have novel properties which leads to do a complete analysis of @xmath124 .
the main rosenfeld - grbner algorithm as presented in @xcite requires some recursive loops to construct the mentioned decomposition .
however , as we consider the rosenfeld - grbner algorithm in the special case of the @xmath25 pde systems , _
there is no need longer to decompose the differential ideal generated by the system into regular differential ideals_. so , it is convenient to provide an adapted and computationally simpler version of the rosenfeld - grbner algorithm to deal with just @xmath25 pdes . let us call this algorithm by _
lrg _ which stands for the _ * l*_incons _ * r*_osenfeld-_*g*_rbner algorithm .
first , we need the definition of @xmath73-polynomial similar in spirit to that in grbner bases theory which plays a crucial role .
consider two @xmath25 differential polynomials @xmath128 and @xmath129 with @xmath160 , @xmath161 .
then , the @xmath73-_polynomial _ of @xmath128 and @xmath129 is defined as : @xmath162 where @xmath163 .
the aim of calculating the @xmath73-polynomial of two differential polynomials is in fact to remove their leading derivatives to obtain ( probably ) a new leading derivative .
now , let us describe the lrg algorithm .
if @xmath164 is a subset of a differential ring @xmath67 ( in fact , a pde system ) then , @xmath159 $ ] denotes the smallest differential ideal of @xmath67 , containing @xmath124 and _ closed under complex conjugation_. [ lrg ] the following theorem shows the termination and correctness of the algorithm .
the following statements hold : * lrg algorithm terminates in a finite number of steps . *
if @xmath165 is the canonical representative of @xmath159 $ ] then , any full - remainder of a @xmath25 differential polynomial @xmath166 on division by @xmath165 is zero if and only if @xmath167 $ ] .
\(a ) the termination of the algorithm is guaranteed by the ritt - raudenbush basis theorem ( the analogue of the hilbert basis theorem for polynomial rings ) . according to this theorem ,
since @xmath168 is noetherian with respect to the radical differential ideals , then @xmath67 is too ( _ cf . _
@xcite ) .
now , using _ reductio ad absurdum _
, let us assume that the algorithm does not terminate for a finite set @xmath124 .
thus , we have an ascending chain of ideals @xmath169 \subset [ \sigma_2 ] \subset \cdots$ ] which does not stabilize where @xmath170 is the set of leading derivatives of the differential polynomials at the @xmath171-th step of the execution of the algorithm ( by the @xmath171-th step we mean computing the @xmath171-th new polynomial and adding it to @xmath165 )
. one can observe that @xmath170 contains @xmath25 polynomials and therefore @xmath172 $ ] is a radical ideal , namely a radical differential ideal , then the radical of a differential ideal is a radical differential ideal ( _ cf . _
@xcite , proposition 2.1 ) . ] .
this contradicts the noetherianity of @xmath67 and so , proves the termination .
\(b ) to prove the correctness of the algorithm , it is sufficient to show that if @xmath167 $ ] is a @xmath25 differential polynomial then its full - remainder on division by @xmath165 is zero . since @xmath167 $ ] , then one can conclude by rosenfeld s lemma ( _ see _ @xcite , chap .
iii , sect .
8 , lemma 5 ) , that a partial remainder of @xmath116 , say @xmath173 , on division by @xmath165 belongs to @xmath174 .
on the other hand , the set of @xmath73-polynomials contains also s - polynomials ( _ cf . _ @xcite)note that the polynomials , under consideration in this paper , are @xmath147 .
moreover , in extendedrittreduction algorithm , we consider the complex conjugation of any computed polynomial .
these imply that @xmath175 according to the buchberger s criterion ( _ see _ @xcite , theorem 5.48 )
. therefore , any full - remainder of @xmath116 will be equal to zero . at the end of this section ,
let us illustrate with the help of an example , how one can employ the maple package differentialalgebra to handle and solve a @xmath25 pde system .
consider the following @xmath25 pde system @xmath176 $ ] : @xmath177 where @xmath178 is the field of fractions of the polynomial ring @xmath179 $ ] .
first , we must call the desired package to make it available : .... [ > with(differentialalgebra ) ; .... and continue with the following three steps . * in this step , we set : @xmath180 furthermore , we define the differential ring @xmath181 $ ] equipped with a ranking as follows .
+ .... [ > r:=differentialring(blocks=[[u , v ] ] , derivations=[x , y ] ) ; .... + this ring has @xmath182 as unknown functions of the variables @xmath183 and also , it admits the set of derivations @xmath184 .
moreover , here the picked ranking is orderly , namely the following statement holds : @xmath185 for two arbitrary elements @xmath186 of @xmath187 . *
next , we employ the rosenfeld - grbner algorithm to compute a canonical representative for @xmath159 $ ] , : + .... [ > cp : = rosenfeldgroebner([p1 , p2 , p3 ] , r ) ; .... + now to see the equations of , for example , the first component of @xmath188 , we enter the following command : + .... [ > equations(cp[1 ] ) ; .... + the result of this line is : @xmath189 * finally , we find the formal power series solutions corresponding to the functions @xmath182 around the point @xmath190 , up to the order @xmath191 : + .... powerseriessolution(cp , order = 3 ) ; .... + and as the results it returns : @xmath192 which is in fact the solution set of the above pde system @xmath124 .
it is worth noting that , if a pde system contains some complex conjugates of unknown functions , then we can use only the maple package differentialalgebra and without implementing lrg algorithm to analyze it . for this purpose ,
we associate to each unknown function a _ tag variable _ as its complex conjugate , and we add it in the definition of the base differential ring .
furthermore , we add the complex conjugate of each input differential polynomial to @xmath124 .
then , the result of this approach is the same as the output of lrg algorithm .
however , the complexity of this computation is growing ( and higher than lrg algorithm ) when we add a new tag variable .
hereafter , by @xmath193 we mean a real analytic cr - generic submanifold of cr - dimension @xmath194 and of codimension @xmath195 ; recall ( @xcite ) that _ cr - genericity _ means that @xmath196 , where @xmath197 is the standard complex structure ( multiplication by @xmath171 ) . in order to compute @xmath198 for an explicitly given such submanifold @xmath199 ,
it is most convenient to work with _ complex defining equations _ of the specific shape ( @xcite ) : @xmath200 where the coordinates @xmath201 are centered at the origin @xmath202 and where @xmath203 , so that all @xmath204 . a general @xmath205 vector field having holomorphic coefficients : @xmath206
is called an _ infinitesimal cr - automorphism _ of @xmath207 , whenever it is _ tangent _ to it , namely whenever @xmath208 . concretely and
more precisely , the condition that a holomorphic vector field @xmath209 belongs to @xmath210 means that for @xmath211 the differentiated equation : @xmath212 = \\ & = \overline{\sf x}\,\big[\overline{w}_j+w_j - \overline{\xi}_j(\overline{z},z , w ) \big ] + { \sf x}\ , \big[\overline{w}_j+w_j - \overline{\xi}_j(\overline{z},z , w ) \big ] \\ & = \overline{w}^j(\overline{z},\overline{w } ) - \sum_{i=1}^n\,\overline{z}^i(\overline{z},\overline{w})\ , \frac{\partial\overline{\xi}_j}{\partial \overline{z}_i } ( \overline{z},z , w ) + \\ & \ \ \ \ \ + w^j(z , w ) - \sum_{i=1}^n\ , z^i(z , w)\ , \frac{\partial\overline{\xi}_j}{\partial z_i}(\overline{z},z , w ) - \sum_{l=1}^k\,w^l(z , w)\ , \frac{\partial\overline{\xi}_j}{\partial w_l}(\overline{z},z , w ) \\ \endaligned\ ] ] should vanish for every @xmath213 . on the other hand , this condition holds if and only if , after extrinsic complexification and replacement of @xmath214 by @xmath215 , the @xmath195 power series obtained in @xmath216 vanish identically , and this yields the _ tangency equations _ : @xmath217 [ first step ] the holomorphic vector field : @xmath218 is an infinitesimal cr - automorphism of a generic real analytic cr - generic submanifold @xmath219 represented in coordinates @xmath220 as the graph of the @xmath195 complex defining functions : @xmath221 if and only if its coefficients @xmath222 and @xmath223 satisfy the @xmath195 equations .
according to proposition [ first step ] , to find the infinitesimal cr - automorphisms @xmath209 associated to a cr - generic submanifold @xmath207 of @xmath224 , it is necessary and sufficient to determine the holomorphic functions @xmath222 and @xmath223 satisfying the tangency equations . then , the main question that immediately arises here is to ask : _ how can one specify such functions ? _ to answer this question , we focus our attention on providing an effective algorithm . to this aim
, first we introduce the expansions of the coefficients of such a sought @xmath225 with respect to the powers of @xmath226 : @xmath227 where the @xmath228 and the @xmath229 are local holomorphic functions .
we will show that the identical vanishing of the @xmath195 equations in @xmath230 is equivalent to a certain ( in general complicated ) linear system of partial differential equations involving the @xmath231 , the @xmath232 , the @xmath233 and the @xmath234 . inserting these expansions with respect to the powers of @xmath226 in the tangency equations ,
we get : @xmath235 since in these equations , @xmath236 is the argument both of all the @xmath237 and of all the @xmath238 appearing in the second line , one should arrange that the same argument @xmath236 takes place inside the functions @xmath239 and @xmath240 appearing in the first line . thus , one is led , for an arbitrary converging holomorphic power series @xmath241 , to apply the well known basic infinite taylor series formulae under the following slightly artificial form : @xmath242 when one does this , one transforms the first lines of the previous @xmath195 equations as follows : @xmath243 but still , we must expand and reorganize everything in terms of the powers @xmath244 of @xmath245 . at first
, we must do this for the multipowers : @xmath246 to begin with , let us denote the @xmath245-power series expansion of @xmath247 by : @xmath248 with the understanding that the coefficients of the expansion of @xmath249 would be denoted plainly @xmath250 , without @xmath251 sign .
reminding @xmath252 , we adopt the convention that in this right - hand side , the @xmath253 for @xmath254 comes not from @xmath255 itself , but just from the supplementary first - order term @xmath256 .
thus , denoting : @xmath257 we may expand explicitly the exponentiated product under consideration , and the intermediate , detailed computations read as follows : @xmath258 \\ & = \sum_{\alpha\in\mathbb n^n}\,\sum_{\beta\in\mathbb n^n}\ , \overline{z}^\alpha\,z^\beta \bigg [ \sum_{\alpha^1+\cdots+\alpha^k=\alpha \atop \beta_1+\cdots+\beta^k=\beta}\ , \sum_{\alpha_1 ^ 1+\cdots+\alpha_{\gamma_1}^1=\alpha^1 \atop \beta_1 ^ 1+\cdots+\beta_{\gamma_1}^1=\beta^1}\ , \cdots\ , \sum_{\alpha_1^k+\cdots+\alpha_{\gamma_k}^k=\alpha^k \atop \beta_1^k+\cdots+\beta_{\gamma_k}^k=\beta_k } \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \
\overline{\xi}_{1,\alpha_1 ^ 1,\beta_1 ^ 1}^\sim(w ) \cdots \overline{\xi}_{1,\alpha_{\gamma_1}^1,\beta_{\gamma_1}^1}^\sim(w ) \cdots\cdots\ , \overline{\xi}_{k,\alpha_1^k,\beta_1^k}^\sim(w ) \cdots
\overline{\xi}_{k,\alpha_{\gamma_k}^k,\beta_{\gamma_k}^k}^\sim(w ) \bigg ] \\ & = : \sum_{\alpha\in\mathbb n^n}\,\sum_{\beta\in\mathbb n^n}\ , \overline{z}^\alpha\,z^\beta\ , \mathcal{a}_{\alpha,\beta,\gamma } \big ( \big\ { \overline{\xi}_{\widehat{j},\widehat{\alpha},\widehat{\beta}}^\sim(w ) \big\}_{\widehat{j}\in\mathbb n,\widehat{\alpha}\in\mathbb
n^n,\widehat{\beta}\in\mathbb n^n } \big ) , \endaligned\ ] ] where we introduce a collection of certain polynomial functions @xmath259 of all the @xmath260 that appear naturally in the large brackets of the penultimate equality , namely where we set : @xmath261 at present , coming back to the @xmath195 equations we left momentarily untouched , we see that in them , five sums are extant and we now want to expand and to reorganize properly each one of these sums as a @xmath262-power series of the form : @xmath263 for the sum in , we therefore compute , changing in advance the index @xmath264 to @xmath265 : @xmath266 . \endaligned\ ] ] the computations for the second sum in are the same : @xmath267 .
\endaligned\ ] ] the third sum in is already almost well written , for we indeed have , if we denote by @xmath268 the zero - multiindex : @xmath269 , \endaligned\ ] ] where @xmath270 if @xmath271 and equals @xmath10 if @xmath272 .
to transform the fourth sum in , we must at first compute , for each @xmath273 ( and for each @xmath274 ) , the first - order partial derivatives @xmath275 from , which gives , if we denote simply by @xmath276 the multiindex @xmath277 of @xmath278 with @xmath10 at the @xmath171-th place and zero elsewhere : @xmath279 thanks to this , the fourth sum in may be reorganized as wanted : @xmath280 . \endaligned\ ] ] lastly , in order to transform the fifth sum in , we must at first compute , for each @xmath281 ( and for each @xmath274 ) , the first - order partial derivatives @xmath282 , and to this aim , we start by rewriting from : @xmath283 whence it immediately follows : @xmath284 thanks to this , the fifth sum in , too , may be reorganized appropriately : @xmath285 \\ & = \sum_{\alpha\in\mathbb n^n}\,\sum_{\beta\in\mathbb n^n}\ , \overline{z}^\alpha\,z^\beta\ , \bigg [ -2\,\delta_\alpha^{\bf 0 } \cdot w^{j,\beta}(w ) - \sum_{l=1}^k\ , \sum_{\beta'+\beta''=\beta}\ , \frac { \partial\overline{\xi}_{j,\alpha,\beta''}^\sim(w ) } { \partial w_l}\ , w^{l,\beta'}(w ) \bigg ] .
\endaligned\ ] ] summing up these five reorganized sums appearing in as a double sum @xmath286 , and equating to zero all the obtained coefficients , , , and , we deduce the following fundamental statement .
[ theorem ] let @xmath207 be a local generic real analytic cr - generic submanifold of @xmath287 having positive codimension @xmath1 and positive cr dimension @xmath2 which is represented , in local holomorphic coordinates @xmath288 centered at the origin @xmath202 by @xmath195 complex defining equations of the shape : @xmath289 with @xmath290 , and introduce the power series expansion with respect to the variables @xmath291 : @xmath292 for every multiindex @xmath293 , every multiindex @xmath294 and every multiindex @xmath295 , introduce also the explicit universal polynomial : @xmath261 then a general holomorphic vector field : @xmath296 is an infinitesimal cr - automorphism of @xmath207 belonging to @xmath297 , namely it has the property that @xmath298 is tangent to @xmath207 _ if and only if _ , for every @xmath299 , for every @xmath293 and for every @xmath300 , the following linear holomorphic partial differential equation : @xmath301 which is _ linear _ with respect to the partial derivatives : @xmath302 together with its conjugate , are satisfied identically in @xmath303 by the four families of functions : @xmath304 depending only upon the @xmath195 holomorphic variables @xmath305 . according to theorem [
theorem ] , finding the sought infinitesimal cr - automorphisms @xmath209 is equivalent to solve the linear pde system constructed by the equations with the unknowns @xmath306 and afterwards , finding the expressions of the holomorphic coefficients @xmath222 and @xmath223 of @xmath209 for @xmath307 and @xmath308 , according to the formulae and thanks to the achieved solution .
then , we can choose the following strategy to compute the desired infinitesimal lie algebra @xmath309 associated to a specific real analytic cr - generic manifold @xmath219 : * constructing the @xmath195 fundamental equations . * expanding these equations according to the formulaes , and . * extracting the coefficients of each @xmath310 for @xmath311 and constituting the linear homogeneous pde system of the partial differential equations , introduced in theorem [ theorem ] . * solving the obtained system by the means of techniques in differential algebra theory . * substituting the solution of the pde system into the formulae to find the holomorphic functions @xmath312 and @xmath313 as the coefficients of the infinitesimal cr - automorphisms @xmath209 .
it is worth noting that the already introduced linear pde system defined underlying the differential ring : @xmath314\ ] ] admits _ complex _ linear equations and hence at the fourth step of the above strategy , it is not possible to employ the rosenfeld - grbner algorithm directly for solving it .
but , it is feasible to equip this algorithm with the conjugate operator bar - reduction for being able to treat the complex equations .
nevertheless , one should notice that such a linear pde system is not necessarily a @xmath25 one , since its coefficients may contain some powers of the variables @xmath315 .
however , a careful look at the equations and reveals that if the @xmath195 defining functions @xmath316 are _ independent _ of the variables @xmath317 , then the concerned linear pde system has constant complex coefficients , namely it will be a @xmath25 pde system .
such cr - generic submanifolds @xmath193 which are defined as the graph of the @xmath195 defining complex equations : @xmath318 are usually called the _ rigid submanifolds_.
they constitute a wide and considerably significant class of cr - generic submanifolds ( _ see _ @xcite ) .
hence , in the case of the rigid real analytic generic cr - generic submanifolds , one can employ the computationally much simpler algorithm _ lrg _ ( algorithm [ lrg ] ) to perform the fourth step of the above strategy .
in the sequel , using this method , we compute the lie algebras of infinitesimal cr - automorphisms associated to three rigid real analytic generic cr - generic submanifolds of @xmath319 .
the aim of the current section is to compute the lie algebras of infinitesimal cr - automorphisms , associated to the three rigid real analytic cr - generic submanifolds @xmath53 and @xmath321 of @xmath322 of cr dimension @xmath10 , represented in coordinates @xmath323 by .
we give in detail the computations of @xmath51 and since the remaining computations of @xmath52 and @xmath65 are fairly similar , then shall not report them in detail . for the first model @xmath63 , represented by the eight real analytic equations : @xmath324 a holomorphic vector field @xmath325 is tangent to @xmath63 if and only if the restriction of the real vector filed @xmath326 to each of the above eight defining equations vanishes identically .
in other words , if and only if the holomorphic functions @xmath327 and @xmath328 , @xmath329 , and their conjugates enjoy the following eight equations ( _ cf . _ ): @xmath330_{w=\overline{w}+\xi } , \\ & 0\equiv\big[w^2-\overline{w}^2 - 4\,i\,z\,\overline{z}z-2\,i\ , \overline{z}^2\,z- 2\,i\,z^2\overline{z}-4\,i\,z\overline{z}\overline{z}\big]_{w= \overline{w}+\xi } , \\ & 0\equiv\big[w^3-\overline{w}^3 - 4\,z\overline{z}z+2\,\overline{z}^2z-2 \,z^2\overline{z}+ 4\,z\,\overline{z}\overline{z}\big]_{w=\overline{w}+\xi } , \\ & 0\equiv\big[w^4-\overline{w}^4 - 6\,i\,z^2\overline{z}z-2\,i \overline{z}^3z-2\,i\,z^3\overline{z}- 6\,i\,z\overline{z}^2\overline{z}\big]_{w=\overline{w}+\xi } , \\ & 0\equiv\big[w^5-\overline{w}^5 - 6\,z^2\overline{z}z+2\,\overline{z}^3z-2 \,z^3\overline{z}+ 6\,z\overline{z}^2\overline{z}\big]_{w=\overline{w}+\xi } , \\ & 0\equiv\big[w^6-\overline{w}^6 - 4\,i\,z\overline{z}^2z-4\,i\,z^2 \overline{z}\overline{z}\big]_{w=\overline{w}+\xi } , \\ & 0\equiv\big[w^7-\overline{w}^7 - 8\,i\,z^3\overline{z}z-2\,i\ , \overline{z}^4{z}-2\,i\,z^4\overline{z}- 8\,i\,z\overline{z}^3\overline{z}\big]_{w=\overline{w}+\xi } , \endaligned\ ] ] @xmath331_{w=\overline{w}+\xi}. \endaligned\ ] ] these functions are real analytic and hence one may expand them with respect to the powers of @xmath226 ( _ cf . _ ): @xmath332 our current aim is to find a closed form expression for the holomorphic functions @xmath333 and @xmath223 by using their corresponding taylor series .
one sees simplified expressions of these functions in the following lemma : after expansion , the first equation reads : @xmath336 + \\ & + \sum_{k\in\mathbb n}\,\overline z^k\big[-\overline w^1_k(\overline w_1,\overline w_2,\ldots,\overline w_8)-2\,i\,z\,\overline z_k(\overline w_1,\overline w_2,\ldots,\overline w_8)\big ] .
\endaligned\ ] ] then , we expand further each holomorphic function @xmath337 and @xmath338 according to the formulae ( _ cf . _ ): @xmath339 chasing the coefficient of @xmath340 for every @xmath341 after further expansion , we therefore see that the first two lines give absolutely no contribution , and that from the third line , it only comes : @xmath342 , which is what was claimed about the @xmath343 .
next , chasing the coefficient of @xmath344 for every @xmath345 , we get @xmath346 .
the seven remaining families of vanishing equations @xmath347 for @xmath348 with @xmath349 , with @xmath350 and with @xmath351 are obtained in a completely similar way by looking at the second to eighth equations of .
after this extensive simplification , we try to determine expressions of the remaining @xmath352 holomorphic functions @xmath353 in the above lemma . substituting in the already obtained expressions for the nine functions @xmath354 , the fundamental equations change into the form : @xmath355 @xmath356 @xmath357 @xmath358 @xmath359 @xmath360 @xmath361
@xmath362 now we arrive at the point where we should expand the functions @xmath363 and @xmath364 appearing in the above eight equations using the taylor series .
afterwards , the coefficients of @xmath365 in these equations should be extracted and set equal to zero , identically .
this is equivalent to solve the @xmath25 pde system constructed by these coefficients underlying the differential ring @xmath366 $ ] , equipped with the functional conjugation operator .
doing so , we have extracted these coefficients as much as it was needed and obtained the following system of 63 equations with 33 unknowns @xmath353 , which are in fact the indeterminates of the differential ring @xmath67 .
here by @xmath367 we mean the coefficient of @xmath365 in the fundamental equation ( @xmath368 ) for @xmath369 : @xmath370 we employ the maple package differentialalgebra , performing the approach described in section [ solving pde ] based on the rosenfeld - grbner algorithm to solve such a @xmath25 pde system , and we obtain the following general solution : @xmath371 @xmath372 for some twelve real variables @xmath373 .
putting these functions into the general expression @xmath374 of the desired infinitesimal cr - automorphisms gives their general form .
one can check easily that such a parametrized holomorphic vector field enjoys the tangency condition @xmath375 . picking the coefficients of the above twelve real variables in this general expression ,
provides twelve @xmath58-linearly independent infinitesimal cr - automorphisms @xmath376 which constitute a basis for the desired lie algebra @xmath51 . before presenting them , we recall that a lie algebra @xmath377 is called _ graded _ in the sense of tanaka whenever it admits a gradation like : @xmath378 satisfying : @xmath379\subset\frak g_{i+j}.\ ] ] [ theorem-1 ] the lie algebra of infinitesimal cr - automorphisms @xmath51 of the rigid real analytic cr - generic submanifold @xmath380 is @xmath381-dimensional , generated by the twelve @xmath58-linearly independent holomorphic vector fields : @xmath382 furthermore , it is graded of the form : @xmath383 with @xmath384,with @xmath385 , with @xmath386 , with @xmath387 , with @xmath388 , and with @xmath389 and together with the following table of commutators : , _ explicit expression of cartan s connections for levi - nondegenerate 3-manifolds in complex surfaces , and identification of the heisenberg sphere _ , cent .
eur . j. math .
, * 10*(5 ) , pages 18011835 , 2012 .
, _ basic principles of mechanical theorem proving in geometries .
, _ journal of sys .
sci . and math .
, * 4*(3 ) , pages 207235 , 1984 .
( see also journal of automated reasoning , * 2*(4 ) , pages 221252 , 1986 ) .
, _ some recent advances in mechanical theorem proving of geometries . ,
_ * 29 * automated theorem proving : after 25 years , contemporary mathematics , american mathematical society , providence , rhode island , pages 235242 , 1984 . | we perform _ detailed _ computations of lie algebras of infinitesimal cr - automorphisms associated to three specific model real analytic cr - generic submanifolds in @xmath0 by employing differential algebra computer tools mostly within the maple package differentialalgebra in order to automate the handling of the arising highly complex linear systems of pde s . before treating these new examples which prolong previous works of beloshapka , of shananina and of mamai
, we provide general formulas for the explicitation of the concerned pde systems that are valid in arbitrary codimension @xmath1 and in any cr dimension @xmath2 .
also , we show how ritt s reduction algorithm can be adapted to the case under interest , where the concerned pde systems admit so - called _ complex conjugations_. |
quantum bit commitment ( qbc ) @xcite is a two - party cryptography including two phases . in the commit phase , alice ( the sender of the commitment )
decides the value of the bit @xmath0 ( @xmath1 or @xmath2 ) that she wants to commit , and sends bob ( the receiver of the commitment ) a piece of evidence , e.g. , some quantum states .
later , in the unveil phase , alice announces the value of @xmath0 , and bob checks it with the evidence .
an unconditionally secure qbc protocol needs to be both binding ( i.e. , alice can not change the value of @xmath0 after the commit phase ) and concealing ( bob can not know @xmath0 before the unveil phase ) without relying on any computational assumption .
qbc is recognized as an essential primitive for quantum cryptography , as it is the building block for quantum multi - party secure computations and more complicated post - cold - war eramulti - party cryptographic protocols @xcite .
unfortunately , it is widely accepted that unconditionally secure qbc is impossible @xcite-@xcite , despite of some attempts toward secure ones ( e.g. , hejpa , hepra , hearxiv , hearxiv2,qbc75,qi92 and the references therein ) .
this result , known as the mayers - lo - chau ( mlc ) no - go theorem , was considered as putting a serious drawback on quantum cryptography .
( note that cheat - sensitive qbc @xcite is not covered , as its security goal is defined differently from that of the qbc studied in the current paper . )
very recently , we proposed a qbc protocol using orthogonal states hejpa , where the density matrices do not satisfy a crucial condition on which the mlc no - go theorem holds .
the protocol is based on the goldenberg - vaidman ( gv ) quantum key distribution ( qkd ) scheme @xcite , which makes use of the mach - zehnder interferometer involving symmetric beam splitters .
koashi and imoto qi917 pointed out that the gv qkd scheme can be simplified by replacing the symmetric beam splitters with asymmetric ones . here
we will apply the same idea to propose a simplified version of our qbc protocol , so that it can be more feasible and efficient . in the next section ,
we introduce some basic notations and settings used throughout the paper .
the koashi - imoto ( ki ) qkd scheme will be briefly reviewed in section iii .
then we propose our qbc protocol in section iv , and analyze its security in section v. the feasibility will be discussed in section vi . section vii summarizes the conclusion and gives some remarks .
an example of the technical attack mentioned in section vi will be provided in the appendix .
generally , in both qkd and qbc the two participants are called alice and bob . but similar to @xcite , in our current qbc protocol , the behavior of bob is more like that of the eavesdropper rather than the bob in qkd . to avoid confusion , here we use the names in the following way . in qkd
, the sender of the secret information is called alice , the receiver is renamed as charlie instead of bob , and the external eavesdropper is called eve . in qbc , the sender of the commitment is alice , the receiver is bob , and there is no eve since qbc merely deals with the cheating from internal dishonest participants , instead of external eavesdropping . as our main interest is focused on the theoretical possibility of secure qbc
, we will only consider the ideal case where no transmission error occurs in the communication channels , nor there are detection loss or dark counts , etc .
our qbc proposal is inspired by the ki qkd scheme @xcite , which makes use of the mach - zehnder interferometer illustrated in fig .
1 . let @xmath3 and @xmath4 denote the reflectivity and transmissivity of the asymmetric beam splitters @xmath5 and @xmath6 , with @xmath7 and @xmath8 .
alice encodes the bit values @xmath9 and @xmath2 she wants to transmit to charlie , respectively , using two orthogonal states@xmath10here @xmath11 is the @xmath12 photon fock state for the arm @xmath13 .
that is , each @xmath14 or @xmath15 is split into two localized wave packets , and sent to charlie separately in quantum channels @xmath16 and @xmath17 , respectively ; thus single photon nonlocality is presented .
this is done by sending a single photon either from the source @xmath18 ( sending @xmath19 ) or from the source @xmath20 ( sending @xmath21 ) , then splitting it with the beam splitter @xmath5 made of a half - silvered mirror ( note that polarizing beam splitters are not recommended due to the security problem addressed at the end of section 6 of @xcite ) . to ensure the security of the transmission , the wave packet in channel @xmath17 is delayed by the storage ring @xmath22 , which introduces a sufficiently long delay time @xmath23 so that this wave packet will not leave alice s site until the other wave packet in channel @xmath16 already entered charlie s site .
thus the two wave packets of the same photon are never present together in the transmission channels .
this makes it impossible for eve to prepare and send charlie a perfect clone _ on time _ if she waits to intercept and measure both wave packets , even though @xmath14and @xmath24 are orthogonal , because she has to decide what to resend to channel @xmath16 before she can receive anything from channel @xmath17 . on the other hand ,
when no eavesdropping occurs , charlie can distinguish @xmath14 and @xmath25 unambiguously by adding a storage ring @xmath26 to channel @xmath16 whose delay time is also @xmath23 , while introducing a phase shift @xmath27 to channel @xmath17 .
the two wave packets of the same photon will then recombine and interfere on the beam splitter @xmath6 , which is identical to @xmath5 .
thus the complete apparatus of alice and charlie forms a mach - zehnder interferometer , so that @xmath28 ( @xmath24 ) will always make the detector @xmath29 ( @xmath30 ) click with certainty , allowing charlie to decode the transmitted bit correctly .
any mismatched result between alice s transmitted state and charlie s measurement will immediately reveals the presence of eve @xcite . comparing with the gv qkd scheme @xcite
, the key difference is that @xmath31 and @xmath6 in the ki scheme are asymmetric beam splitters , while the gv scheme uses symmetric ones .
the advantage of this modification is that the sending time of each photon can be fixed and publicly known beforehand , while in the gv scheme it has to be random and kept secret from eve until the security check .
an important fact that will be useful for our qbc protocol is that , since the ki scheme is unconditionally secure , it is clear that eve s intercept - resend attack will always be detected with a nontrivial probability .
that is , if she intercepts a state sent from alice and decodes a nontrivial amount of information , then the state she resends to charlie can not always make charlie s correct detector click at the right time with the probability @xmath32 , no matter what is eve s strategy on preparing the resent state .
there will always be a nontrivial probability that the resent state will be detected by the wrong detector ( the one that should not click when no eavesdropping presented ) , or at the wrong time ( which is earlier or later than the correct arrival time when the state is not intercepted ) .
please see @xcite for the rigorous proof , as well as some examples showing why eve s strategies do not work .
as illustrated in fig . 2 , to build a qbc protocol upon the above ki qkd scheme , we treat charlie s site as a part of alice s , so that the two parties merge into one .
alice sends out a bit - string @xmath33 encoded with the above orthogonal states , whose value is related to the bit @xmath0 she wants to commit .
then she receives the states herself . meanwhile , let bob take the role of eve .
his action shifts between two modes . in the _ bypass _ mode
, he simply does nothing so that the corresponding parts of the states return to alice intact . in the _ intercept _ mode , he applies the intercept - resend attack .
that is , he intercepts the state and decodes the corresponding bit ( which can be done using the same device as that of charlie s ) , and prepares a fake state to resend to alice on time .
while there could be many strategies for bob to prepare the resent state , we use @xmath34 to denote the lower bound of the average probability for his resent state to be caught in alice s check . as we mentioned above
, the unconditional security of the ki qkd scheme guarantees that @xmath35 can not always equal exactly to zero for both @xmath28 and @xmath24 even when bob uses the optimal strategy .
therefore , alice is able to estimate the upper bound of the frequency of the presence of the intercept mode , to limit bob from intercepting the whole string @xmath33 , so that the value of the committed bit @xmath0 can be made concealing .
meanwhile , since @xmath36 , at the end of the commit phase there will be some bits of the string become known to bob , while alice does not know the exact position of all these bits .
thus she can not alter string @xmath33 freely at a later time , making the protocol binding .
the complete qbc protocol is described below .
the _ commit _ protocol : \(1 ) bob chooses a binary linear @xmath37-code @xmath38 @xcite and announces it to alice , where @xmath39 , @xmath40 , @xmath41 are agreed on by both alice and bob .
\(2 ) alice chooses a nonzero random @xmath39-bit string @xmath42 and announces it to bob .
this makes any @xmath39-bit codeword @xmath43 in @xmath38 sorted into either of the two subsets @xmath44 and @xmath45 . here
@xmath46 .
\(3 ) now alice decides the value of the bit @xmath0 that she wants to commit .
then she chooses a codeword @xmath33 from @xmath47 randomly .
\(4 ) alice encodes each bit of this specific @xmath33 as @xmath48 where the state @xmath49 is defined by equation ( [ eqpsi ] ) .
she sends bob the two wave packets of the same state separately in channels @xmath16 and @xmath17 , with the storage ring @xmath22 on channel @xmath17 introducing a delay time @xmath50 known to bob .
\(5 ) for each of alice states @xmath51 ( @xmath52 ) , bob chooses the intercept mode with probability @xmath53 ( @xmath54 ) and the bypass mode with probability @xmath55 .
if bob chooses to apply the bypass mode , he simply keeps channels @xmath16 and @xmath17 intact so that the state sent from alice will be returned to her detectors as is .
else if bob chooses to apply the intercept mode , he uses the same measurement device as that of alice s , to measure @xmath49 so that he can decode @xmath56 with certainty .
meanwhile , he prepares another state and sends it back to channels @xmath16 and @xmath57 at the right time , i.e. , the time which can ensure that it reaches alice s detectors at a time that looks as if bob were applying the bypass mode .
there are many different strategies how bob sends this state ( thus we left this part of bob s device as a black box in fig .
one of the simplest ways is to use the same device that alice uses for sending her state .
more strategies will be discussed in section v.a .
it will be shown there that all these strategies can not hurt the security of the protocol , so that bob are allowed to use any of them here . as stated above , in any strategy
there is a nonvanishing probability that bob s resent state does not equal to alice s original @xmath51 since he has to send it before completing the measurement on @xmath51 ( to be further explained in section v.a too ) .
let @xmath35 be the lower bound of this probability for all these strategies .
\(6 ) alice uses the same device that charlie used in the ki qkd scheme , to measure the output of the quantum channels from bob .
she counts how many times her measurement results do not match the states she sent , and denotes it as @xmath58 . from step ( 5 )
it can be shown that @xmath59 . thus alice can estimate the upper bound of the probability of bob choosing the intercept mode as @xmath60 .
alice agrees to continue with the protocol if she finds @xmath54 , which means that the number of @xmath56 s known to bob is @xmath61 .
otherwise she concludes that bob is cheating .
the _ unveil _ protocol : \(7 ) alice announces the values of @xmath0 and the specific @xmath43 she chose .
\(8 ) bob accepts the commitment if @xmath62 and @xmath33 is indeed a codeword from @xmath38 , and every @xmath56 agrees with the state @xmath49 he detected in the intercept mode .
for easy understanding , here we will first give a heuristic explanation of the security of the protocol . a more general theoretical proof will be provided in section v.b .
the binary linear @xmath37-code @xmath38 used in the protocol can simply be viewed as a set of classical @xmath39-bit strings .
each string is called a codeword .
this set of strings has two features .
\(a ) among all the @xmath63 possible choices of @xmath39-bit strings , only a particular set of the size @xmath64 ( @xmath65 ) is selected to form this set .
\(b ) the distance ( i.e. , the number of different bits ) between any two codewords in this set is not less than @xmath41 ( @xmath66 ) .
feature ( a ) puts a limit on alice s freedom on choosing the initial state @xmath67 , since each @xmath49 ( @xmath68 ) can not be chosen independently . instead
, the string @xmath69 formed by the indices @xmath56 s can only be chosen among the codewords .
meanwhile , feature ( b ) guarantees that if alice wants to change the string @xmath33 from one codeword into another so that the value of her committed @xmath0 can be altered , she needs to change at least @xmath41 bits of the codeword @xmath33 .
but among the @xmath39 states she sent to bob , there are only @xmath70 states which she knows for sure that bob has applied the intercept mode . for each of the rest @xmath71 states
, her measurement result always matches the state she sent , so that she can not distinguish unambiguously whether bob has applied the bypass mode or the intercept mode .
if it was the intercept mode , bob already knew the corresponding bit @xmath56 of the codeword @xmath33 , so that alice s altering it will be caught inevitably .
rigorously speaking , among these @xmath72 states , the probability that bob has applied the intercept mode is@xmath73therefore alice s altering one bit of the codeword will stand the probability @xmath74 to be detected , and her probability of altering @xmath75 bits without being detected will be @xmath76 . alternatively , alice may prepare @xmath77 initially in a state where @xmath33 is not a codeword .
instead , it is half - way between two codewords , so that changing @xmath78 will be sufficient to turn it into one of the codewords . in this case
, her probability of escaping the detection will be increased to @xmath79 .
nevertheless , in either way the probability drops exponentially as @xmath41 increases .
thus it can be made arbitrarily close to zero .
on the other hand , feature ( a ) also guarantees that the number of different codewords having less than @xmath80 bits in common increases exponentially with @xmath40 .
this is the key that makes our protocol secure against bob , as alice can bound the frequency @xmath53 that bob applies the intercept mode , which in turns bounds the number of the bits known to bob so that he can not know alice s actual choice of the codeword .
the reason is that in step ( 5 ) of the protocol , no matter what is the strategy that bob used in his intercept mode to prepare the resent state , it can be shown that his resent state can not _
always _ equal to the @xmath51 he received from alice , as long as he wants to ensure that the time it reaches alice s detectors will show no difference than the case where he were applying the bypass mode .
that is , either bob s resent state in the intercept mode will inevitably make alice s detectors click earlier or later than the time it does when the bypass mode were used instead , or the resent state will be different from @xmath51 with a nonvanishing probability @xmath35 . to see why this is the case ,
let us first consider the strategy where bob prepares the resent state using the same device that alice uses for sending her @xmath51 .
suppose that the first wave packet of alice s @xmath51 enters bob s site via channel @xmath16 ( before entering his storage ring ) at time @xmath81 ( which should be agreed on by both alice and bob beforehand ) .
then the second wave packet of @xmath51 will enter bob s site via channel @xmath17 at time @xmath82 due to the existence of the storage ring in alice s sending device ( see fig .
2 ) . as there is also another storage ring in alice s measurment device
, bob must send the first wave packet of his resent state into channel @xmath16 at time @xmath81 , and send the second wave packet into channel @xmath17 at time @xmath82 , so that they can reach alice s measurement device at a time that looks like bob is running the bypass mode . here for simplicity , we assume that the time for the wave packets to travel the rest part of the channels ( other than the storage rings ) is negligible . that is , the first wave packet of bob s state needs to be sent before the second wave packet of alice s @xmath49 reaches him .
otherwise his resent state will reach alice s detectors later than the expected arrival time when no interception occurs .
therefore , by the time bob prepares the resent state , he has not received alice state completely yet so he does not know the form of @xmath83 .
thus his resent state can not match alice s state with probability @xmath32 , and he can not make them match by his local operations acting solely on the second wave packet of his resent state ( the one to be sent via channel @xmath17 ) after the first wave packet in channel @xmath84 was already sent .
on the other hand , suppose that bob waits until @xmath82 so that alice s @xmath51 enters his measurement device completely , and prepares the resent state following the measurement result .
then even though the form of the states will match perfectly , the resent state will reach alice s detectors much later than it should when bob uses the bypass mode , because the storage ring in channel @xmath16 of alice s measurement device will delay the corresponding wave packet .
this will make alice aware that bob is using the intercept mode too .
alternatively , there can be another strategy where bob simply sends all wave packets of his state simultaneously via one of the channels alone , e.g. , through channel @xmath16 at time @xmath81 or through channel @xmath17 at time @xmath85 .
but he will not be able to guarantee which of alice s detectors will click with certainty .
though here we have only analyzed the above few strategies as examples , the result is common .
that is , in any strategy potentially existed , there will be a nonvanishing probability ( let @xmath35 denote the lower bound of the probability for all strategies ) that alice will find a mismatched result between the @xmath51 she sent and her measurement on the state she received from bob , or the arrival time of bob s state is different from what it should be in the bypass mode .
this is because , as mentioned in section iv , the role of bob in our protocol is actually the same as that of eve in the ki qkd scheme @xcite .
if there is a strategy which is not bounded by the above result , then it can be utilized to fulfill a successful eavesdropping to the ki qkd scheme , which will conflict with the existing proof of the unconditional security of the scheme @xcite . as a consequence ,
whenever bob applies the intercept mode , alice can distinguish it from the bypass mode with the probability @xmath35 . then as described in step ( 6 ) , by counting the number @xmath58 of the mismatched results , alice can estimate the upper bound of the probability of bob choosing the intercept mode as @xmath60 .
as she agrees to continue only when @xmath54 , it is guaranteed that during the commit phase , bob knows only @xmath61 bits of @xmath33 . since feature ( a ) of the binary linear @xmath37-code @xmath38 ensures that the number of different codewords having @xmath86 bits in common increases exponentially with @xmath40 , the potential choices for @xmath33 are too much for bob to determine whether @xmath33 belongs to the subset @xmath87 or @xmath88 .
thus the amount of information bob gained on the committed bit @xmath0 before the unveil phase can be made arbitrarily close to zero by increasing @xmath40 .
taking both alice s and bob s cases studied above into consideration , we can see that fixing @xmath89 and @xmath90 while increasing @xmath39 will then result in a protocol secure against both sides .
while it is hard to find a completely general proof like those for qkd qi70 showing that the protocol is secure against all cheating strategies that may potentially exist , here it can be shown that our protocol is at least secure against the specific cheating strategy proposed in the mlc theorem . as pinpointed out in section 4 of @xcite , while there are many variations and extensions of the mlc no - go theorem @xcite-@xcite , their proofs all have the following common features .
\(i ) the reduced model .
according to the no - go proofs , any qbc protocol can be reduced to the following model .
alice and bob share a quantum state in a given hilbert space .
each of them performs unitary transformations on the state in turns .
all measurements are performed at the very end .
( ii ) the coding method .
the quantum state corresponding to the committed bit @xmath0 has the form@xmath91here the systems @xmath92 and @xmath93 are owned by alice and bob respectively .
\(iii ) the concealing condition . to ensure that bob s information on the committed bit is trivial before the unveil phase
, any qbc protocol secure against bob should satisfy@xmath94where @xmath95 is the reduced density matrix corresponding to alice s committed bit @xmath0 , obtained by tracing over system @xmath96 in equation ( [ coding ] ) .
note that in some presentation of the no - go proofs ( e.g. , @xcite ) , this feature was expressed using the trace distance or the fidelity instead of the reduced density matrices , while the meaning remains the same .
\(iv ) the cheating strategy .
as long as equation ( [ eqconcealing ] ) is satisfied , according to the hughston - jozsa - wootters ( hjw ) theorem ( a.k.a
. the uhlmann theorem , etc . )
@xcite , there exists a local unitary transformation for alice to map @xmath97 into @xmath98 successfully with a high probability @xcite .
thus a dishonest alice can unveil the state as either @xmath99 or @xmath100 at her will with a high probability to escape bob s detection .
consequently , a concealing qbc protocol can not be binding . following the method of the mlc theorem to reduce our protocol into an equivalent form where alice and bob perform unitary transformations in turns on the quantum state they share
, we can see that our commit protocol is essentially the following 3-stage process .
\(i ) alice prepares and sends bob a system @xmath101 containing @xmath39 qubits @xmath83 ( @xmath68 ) .
\(ii ) bob prepares an @xmath39-qubit system @xmath102 and an @xmath39-qutrit system @xmath103 .
then he performs an operation @xmath104 on the combined system @xmath105 .
\(iii ) bob sends system @xmath102 to alice . in this process
, the initial state of each qubit in @xmath102 is chosen at bob s choice , and kept secret from alice .
the system @xmath106 is for storing the result of bob s measurement on @xmath101 .
the three orthogonal states of each qutrit in @xmath106 are denoted as @xmath107 , @xmath108 and @xmath109 , where @xmath110 ( @xmath108 ) means that the measurement result of @xmath51 is @xmath111 ( @xmath112 ) , and @xmath109 means that @xmath49 is not measured so that the result remains unknown .
all the @xmath39 qutrits in system @xmath106 are initialized in @xmath113 .
the operation @xmath114 is defined by the mode that bob chooses for each qubit in @xmath101 , where each @xmath115 ( @xmath68 ) is a @xmath116-particle operator acting on the @xmath117-th qubits / qutrit of @xmath102 , @xmath101 and @xmath106 .
if bob chooses the bypass mode for the @xmath117-th qubit of @xmath101 , then@xmath118here @xmath119 is a @xmath120-qubit permutation operator which interchanges the states of the @xmath117-th qubits of @xmath101 and @xmath102 , and @xmath121 is the identity operator that keeps the @xmath117-th qutrit of @xmath103 unchanged . or if bob chooses the intercept mode for the @xmath117-th qubit of @xmath101 , then@xmath122here @xmath123 is the identity operator on the @xmath117-th qubit of @xmath102 , @xmath124 ( @xmath125 ) is an unitary transformation on the @xmath117-th qutrit of @xmath106 which can map the state @xmath113 into @xmath126 ( @xmath127 ) .
that is , applying @xmath128 is equivalent to measuring the @xmath117-th qubit of @xmath101 and storing the result in the @xmath117-th qutrit of @xmath106 , while keeping the @xmath117-th qubit of @xmath102 unchanged . among all the possible forms of the operation @xmath104 , bob chooses one of those that can pass the security check in step ( 6 ) of our original protocol in section iv ( i.e. , the number of @xmath128 in @xmath104 should be @xmath61 ) , and keeps his choice secret from alice .
now we will show why the specific cheating strategy in the mlc theorem does not apply to our protocol . with the above equivalent description , we can see that after the commit phase , the density matrix of the systems shared between alice and bob is in the form @xmath129 , where @xmath130 , @xmath131 and @xmath132 are the density matrices for the systems @xmath102 , @xmath101 and @xmath106 , respectively .
note that here the first @xmath39 qubits are now owned by alice , i.e. , they serve like the system @xmath92 in equation ( [ coding ] ) even though system @xmath102 was prepared by bob .
meanwhile , the rest qubits and qutrits now belong to bob and serve like the system @xmath93 in equation ( [ coding ] ) , even though system @xmath101 was prepared by alice .
let @xmath133 ( @xmath134 ) denote the initial density matrix of system @xmath101 prepared by alice , that can unveil the committed bit as @xmath1 ( @xmath135 ) . according to the mlc theorem ,
now the goal for an dishonest alice is to find a local unitary transformation @xmath136 acting on the @xmath39 qubits at her side alone , that can map @xmath137 into @xmath138 .
that is , @xmath136 should satisfy@xmath139u_{a}^{\dagger } = u_{b}(\rho ^{\beta } \otimes \rho _ { 1}^{\alpha } \otimes \rho ^{\gamma } ) u_{b}^{\dagger } .\]]applying @xmath140 on both sides of this equation , we yield@xmath141recall that @xmath104 was kept secret from alice , thus she has to choose an @xmath142 which is independent of @xmath104 .
meanwhile , the right side of equation ( [ ua ] ) is independent of @xmath104 too .
therefore , equation ( ua ) can not be satisfied in general , unless @xmath136 always commutes with the specific @xmath104 s that bob may choose in the protocol . in this case
@xmath143 , and equation ( [ ua ] ) becomes@xmath144 in many previous qbc protocols ( e.g. @xcite ) there is @xmath145 .
then the hjw theorem guarantees that such a local @xmath136 exists , so that alice can alter the value of her committed bit . this is why the cheating strategy in the mlc theorem is always successful in such protocols .
however , as shown from equation ( eqpsi ) , in our protocol @xmath14 and @xmath15 are orthogonal .
thus the state @xmath67 corresponding to a specific codeword @xmath33 is orthogonal to the state corresponding to any other codeword .
consequently , our qbc protocol satisfies @xmath146 , which is a crucial difference from previous insecure protocols . in this case
the hjw theorem does not apply , making alice unable to find an local transformation @xmath136 acting on her system ( i.e. , system @xmath147 ) alone while satisfying equation ( [ ua2 ] ) . for this reason ,
our protocol is immune to the specific cheating strategy proposed in the mlc no - go theorem . as the system @xmath101 alice sent to bob in our qbc protocol satisfies @xmath148 , at the first glance it seems that bob can simply perform a measurement @xmath149 to project the state of @xmath150 into the hilbert spaces supported by @xmath151 ( @xmath152 ) and thus learn the value of alice s committed @xmath0 .
but this is not true , because as shown above , the density matrix of the systems shared between alice and bob after the commit phase is @xmath129 .
that is , bob is required to perform the operation @xmath104 on system @xmath101 .
this @xmath104 is incommutable with the measurement @xmath149 on @xmath101 for distinguishing @xmath153 and @xmath134 , since equation ( [ u bypass ] ) indicates that @xmath104 contains permutation operators which act on not only system @xmath101 , but also other systems .
therefore , @xmath104 and @xmath149 can not be both applied on @xmath101 .
bob can only choose one of them .
the timing of the sending of the states in the protocol also prevents bob from keeping system @xmath102 unsent until he receives the entire system @xmath101 .
thus he needs to decide on the fly which operation to apply .
once he applied @xmath104 , then the state of @xmath101 will be disturbed so that bob will lose the chance to apply @xmath149 on it for decoding @xmath0 .
moreover , the difference between @xmath104 and @xmath149 is detectable to alice . according to step ( 5 ) of our protocol
, an @xmath104 is considered legitimate if it includes the operator @xmath154 ( i.e. , bob is using the bypass mode ) for @xmath155 times . on the contrary , because the minimal distance between the codewords is @xmath41 , the number of @xmath154 in @xmath149 must be less than @xmath41 . otherwise as a basic property of the binary linear @xmath37-code
, the number of the possible codewords having less than @xmath80 bits in common will be at the order of magnitude of @xmath156 so that such an @xmath149 can provide only trivial information on the value of alice s committed @xmath0 .
but as elaborated in section v.a , the unconditional security of the ki qkd scheme @xcite ensures that , for each of alice s state @xmath49 which bob applies the intercept mode , his resent state can equal to @xmath51 with the probability @xmath157 only , where there is always @xmath158 no matter which resend strategy bob uses .
when the number of the applied bypass mode is less than @xmath41 , alice will find that the number of mismatched result between her received state and the original @xmath49 she sent will be @xmath159 .
if she takes @xmath160 as stated in step ( 6 ) of the protocol , then she will find that @xmath161 .
thus she knows that bob was attempting to apply the operation @xmath149 instead of a legitimate @xmath104 .
namely , while alice s committed state satisfies @xmath162 so that it is distinguishable to bob , alice s security check in the protocol requires bob to perform an operation @xmath104 , which is incommutable with the operation @xmath149 for distinguishing @xmath133 and @xmath134 .
the protocol thus becomes concealing against bob .
in the above we focused only on the theoretical possibility of evading the mlc no - go theorem .
but we can see from fig .
2 that our protocol is also very feasible , as the required experimental technology is already available today @xcite . also ,
as the secret random sending time is no longer required , the commit phase will take less time than that of our previous protocol @xcite , and the total number of photons that bob needs to send in the intercept mode will be significantly reduced . more rigorously ,
suppose that the minimal time for bob to shift between the intercept and bypass modes is @xmath163 .
when both protocols choose the same @xmath39 as the length of the codewords , our previous protocol @xcite requires bob to send @xmath164 ( @xmath165 .
note that @xmath53 was denoted as @xmath101 in @xcite ) photons in total , and the duration time of the commit phase is @xmath166 .
but in the current protocol , the photon number is reduced to @xmath167 , and the duration time is @xmath168 . as it was suggested in section 3 of @xcite that a typical choice of the parameters is @xmath169 , we can see that the current protocol can be @xmath170 times more efficient than our previous one @xcite .
there is another advantage of removing the need to keep the sending time secret at first .
that is , now alice and bob can determine the binary linear @xmath171-code @xmath38 and decide on the sending time of the states @xmath172 ( @xmath68 ) beforehand , so that no more classical communication is needed at all during the commit phase , unless one of them finds the other participant cheating and announces to abort .
therefore , it not only reduces the communication cost significantly , but also makes it easier for security analysis and comparison with the mlc theorem , which was generally deduced in a theoretical model where classical communication is not presented directly . nevertheless , under practical settings , some more security checks should be added against technical attacks .
especially , the physical systems implementing the states may actually have other degrees of freedom , which leave rooms for alice s cheating .
for example , she may send photons with certain polarization or frequency , so that she can distinguish them from the photons bob sends in the intercept mode .
in this case , bob and alice should discuss at the beginning of the protocol , to limit these degrees of freedom to a single mode . in step ( 5 )
when bob chooses the intercept mode , he should also measure occasionally these degrees of freedom of some of alice s photons , instead of performing the measurement in the original step ( 5 ) . then if alice wants to send distinguishable photons with a high probability so that they are sufficient for her cheating , she will inevitably be detected .
moreover , when bob applies the bypass mode , he should add phase shifters to both channels @xmath16 and @xmath17 to introduce the same phase shift in both channels so that an honest alice will not be affected ( to be explained in the appendix ) , while the amount of this phase shift is randomly chosen and kept secret from alice , so that the counterfactual attack described in the appendix can be defeated .
note that all these technical attacks and the corresponding modifications are due to the imperfection of the physical systems implementing the protocol .
they should not be messed with the theoretical possibility of evading the mlc no - go theorem .
in conclusion , inspired by the ki simplified version @xcite of the gv qkd scheme @xcite , we improved our previously proposed qbc protocol @xcite , and proved that it is immune to the specific cheating strategy used in the mlc no - go theorem of unconditionally secure qbc .
the key reason is that the density matrices of alice s states corresponding to the committed values @xmath9 and @xmath2 are orthogonal to each other , making her unable to find the local unitary transformation for the cheating by using the hjw theorem . meanwhile , the protocol remains concealing against bob because he is required to perform another operation , which is incommutable with the measurement for distinguishing the density matrices .
however , there may potentially exist innumerous cheating strategies other than the one in the mlc theorem .
it is natural to ask whether our protocol can be unconditionally secure against all these strategies .
a rigorous evaluation of the upper bound of the cheating probability is also needed .
for example , our protocol involves a probability @xmath35 which denotes the lower bound of the average probability for bob s resent state in step ( 5 ) to be caught in alice s check .
the exact value of @xmath35 should be determined by the rigorous quantitatively security analysis of the ki qkd scheme .
unfortunately , such a value was not yet provided in the literature @xcite . in turns ,
the parameters @xmath40 and @xmath41 in the binary linear @xmath37-code @xmath38 used in our protocol need to be chosen according to @xmath35 .
therefore it remains unknown whether the cheater can have a nontrivial probability of success if these parameters are improperly chosen .
we would like to leave these questions open for future researches .
though the current qbc protocol and the one in @xcite have similarities in many ways , the underlying origins of their security against bob are somewhat different .
while both protocols are immune to bob s cheating because they are based on unconditionally secure qkd schemes , as pointed out in @xcite , the gv qkd scheme can actually be viewed as utilizing three orthogonal states two photon states and one vacuum state .
its security is provided by the random sending time of the photons . on the contrary
, the ki qkd scheme does not require the vacuum state and the secret sending time .
the security is guaranteed by the fact that the eavesdropper can not fake the states with certainty owe to the use of the asymmetric beam splitters .
similarly , the security of the qbc protocol in @xcite against bob is based on alice s random sending time that remains secret before the last step of the commit phase , while in our current qbc proposal , it is because bob can not fake the states with certainty when he runs the intercept mode .
therefore our current protocol is more than merely a simplification on the presentation .
the work was supported in part by the nsf of china , the nsf of guangdong province , and the foundation of zhongshan university advanced research center .
as we mentioned in sec . 6 , under practical settings our qbc protocol may need some modifications against technical attacks .
here we give such an example .
recently a cheating strategy against counterfactual qkd protocols qi801,qi1026 was proposed @xcite . unlike general intercept - resend attacks in which measurements are performed on the quantum states carrying the secret information , in this strategy the cheater makes use of quantum counterfactual effect to detect the working modes of the devices of other participants .
thus it was named the counterfactual attack @xcite . here
we will skip how it applies to qkd protocols , while focus only on its impact on our qbc protocol .
( f2 ) when the paths @xmath173 and @xmath0 are adjusted correctly , two wave packets coming from paths @xmath173 and @xmath0 , respectively , will interfere and combine together , and enter path @xmath33 with certainty .
an ideal fbs that can realize these functions faithfully does not exist in principle .
thus it is called fictitious .
for example , devices with the functions ( f2 ) and ( f3 ) may not accomplish the function ( f1 ) perfectly , i.e. , a photon coming from path @xmath33 could pass the devices with a nontrivial probability , making the attack detectable . however , fbs can be implemented approximately by using an infinite number of ordinary bs @xcite . in practice , the number of bs involved in the implementation has to be finite .
but if the deviation from an ideal fbs is too small to be detected within the capability of available technology , then the attack could become a real threat .
suppose that an ideal fbs is available to a dishonest alice in our qbc protocol .
when she is supposed to send bob a state in step ( 4 ) , she runs both the fbs system in fig .
3 and the apparatus in the honest protocol ( i.e. , the one shown in fig .
2 ) simultaneously in parallel , with path @xmath0 of the fbs system connecting to both the input and output of bob s channel @xmath16 ( or both the input and output of bob s channel @xmath17 ) . the apparatus in fig .
2 works as usual so that the protocol can be executed as if she is honest , while the fbs system serves as a probe to detect bob s mode .
according to the function ( f2 ) of the fbs , whenever bob applies the bypass mode in step ( 5 ) , the wave packets of a photon alice sent to the fbs will be returned from both paths @xmath173 and @xmath0 so that the detector @xmath174 will click with certainty . on the other hand , whenever bob applies the intercept mode , an ideal fbs can guarantee that @xmath174 will never click as path @xmath0 is actually blocked . therefore alice can learn bob s mode unambiguously . since bob
does not know the state @xmath51 alice sends when he applies the bypass mode , alice can lie about the value of the corresponding @xmath56 freely , thus alters her committed @xmath0 in the unveil phase .
nevertheless , it is easy to defeat this counterfactual attack . as pointed out in ref .
@xcite , bob s randomizing the optical length of path @xmath0 is sufficient to destroy the interference effect in the fbs system .
therefore in our protocol , bob can simply add extra phase shifters ( other than the one shown in fig .
2 ) to both channels @xmath16 and @xmath17 when he applies the bypass mode , to introduce the same phase shift @xmath175 in both channels , where the value of @xmath176 is randomly chosen and kept secret from alice , and can vary for different @xmath49 . in this case , after passing bob s apparatus , alice s initial states @xmath14 and @xmath21 will become , respectively , we can see that @xmath178 ( @xmath179 ) differs from @xmath28 ( @xmath24 ) merely by a global factor @xmath180 .
it is well known that such a global factor is not detectable .
in fact , in our case the interference pattern of the two wave packets meeting at the beam splitter of alice s measurement device is determined by their relative phase difference . no change will be detected when they both receive a phase shift @xmath175 .
thus an honest alice who uses the original apparatus described in fig . 2 will still detect @xmath181 ( @xmath182 ) as @xmath14 ( @xmath15 ) , even though she does not know the value of @xmath176 .
on the other hand , if alice wants to apply the above counterfactual attack , without knowing @xmath176 she can not know how to adjust path @xmath173 to ensure @xmath174 in fig .
3 clicking with certainty . consequently
, there will be times that alice does not know which mode bob is running . then the number of @xmath56 s that she can alter will be limited , which is insufficient to change the committed @xmath0 as long as the value of @xmath183 in our qbc protocol is properly chosen .
99 bennett , c.h . , brassard , g. : quantum cryptography : public key distribution and coin tossing . in : proceedings of ieee international conference on computers , systems , and signal processing , pp .
ieee , new york ( 1984 ) brassard , g. , crpeau , c. , jozsa , r. , langlois , d. : a quantum bit commitment scheme provably unbreakable by both parties . in proceedings of 34th annual ieee symposium on foundations of computer science , pp .
ieee , los alamitos ( 1993 ) crpeau , c. : what is going on with quantum bit commitment ? in : proceedings of pragocrypt 96 : 1st international conference on the theory and applications of cryptology .
czech technical university publishing house , prague ( 1996 ) dariano , g.m . : the quantum bit commitment : a complete classification of protocols ( shortened version of quant - ph/0209149 ) .
in : proceedings of qcm&c .
rinton press , boston . also available as .
e - print .
quant - ph/0209150v1 ( 2002 ) dariano , g.m . ,
kretschmann , d. , schlingemann , d. , werner , r.f . : reexamination of quantum bit commitment : the possible and the impossible .
a * 76 * , 032328 ( 2007 ) .
e - print .
quant - ph/0605224v2 ( 2006 ) chiribella , g. , dariano , g.m .
, perinotti , p. , schlingemann , d.m . ,
werner , r.f . : a short impossibility proof of quantum bit commitment .
a * 377 * , 1076 ( 2013 ) .
e - print .
arxiv:0905.3801v1 ( 2009 ) li , q. , li , c.q . ,
long , d .- y . , chan , w.h .
, wu , c .- h . : on the impossibility of non - static quantum bit commitment between two parties .
quantum inf . process . * 11 * , 519 ( 2012 ) .
e - print .
arxiv:1101.5684v1 ( 2011 ) | we simplified our previously proposed quantum bit commitment ( qbc ) protocol based on the mach - zehnder interferometer , by replacing symmetric beam splitters with asymmetric ones .
it eliminates the need for random sending time of the photons ; thus , the feasibility and efficiency are both improved .
the protocol is immune to the cheating strategy in the mayers - lo - chau no - go theorem of unconditionally secure qbc , because the density matrices of the committed states do not satisfy a crucial condition on which the no - go theorem holds . |
the covering spectrum of a complete length space @xmath0 is a geometric invariant introduced by sormani and wei @xcite that measures the size of one dimensional holes in @xmath0 by isolating the lengths of certain closed geodesics that are minimal in their free homotopy class .
more specifically , the covering spectrum can be computed by considering a certain family @xmath1 of regular coverings of @xmath0 , where @xmath2 covers @xmath3 for @xmath4 .
the covering spectrum of @xmath0 then consists of those values of @xmath5 where the isomorphism type of the cover changes .
that is , the covering spectrum is the collection of @xmath5 s where there is a `` jump '' in the step function @xmath6 .
one can see that a `` jump '' at time @xmath5 corresponds to the appearance of a `` hole '' of circumference @xmath7 or , in the case of manifolds , a non - trivial free homotopy class whose shortest geodesic is of length @xmath8 ( cf .
proposition [ prop : newgeo ] ) . as an example
, we follow ( * ? ? ?
* example 2.5 ) and consider the flat @xmath9 torus @xmath10 , where @xmath11 denotes the circle of circumference @xmath12 .
one can see through the covering spectrum algorithm ( see [ csa ] ) that the @xmath5-covers of @xmath0 are given by @xmath13 ( for @xmath14 ) , @xmath15 ( for @xmath16 ) and @xmath17 ( for @xmath18 ) .
hence , the covering spectrum of @xmath0 is given by @xmath19 .
each of these numbers is the circumference of a `` hole '' in this torus and also half the length of the minimal closed geodesics in the free homotopy classes associated to the standard generators @xmath20 and @xmath21 in @xmath22 . as there is a long - standing interest in the relationship between the laplace spectrum of a manifold and its length spectrum ( i.e. , the collection of lengths of closed geodesics ) , it is natural to wonder about the mutual influences between the laplace spectrum and the covering spectrum
specifically , one would like to know if the covering spectrum is a spectral invariant . in (
10.5 ) , sormani and wei demonstrated that certain isospectral manifolds arising through sunada s method @xcite share the same covering spectrum , suggesting that the covering spectrum might at least be an invariant of this technique .
however , through an analysis of the group theory and geometry underlying the covering spectrum we constructed sunada - isospectral manifolds with distinct covering spectra in dimensions three and higher ( * ? ? ?
8) , thus showing that the covering spectrum is not a spectral invariant .
indeed , we recall that a triple of finite groups @xmath23 , where @xmath24 and @xmath25 are subgroups of @xmath26 , is said to be a _ gassmann - sunada _
triple if for any conjugacy class @xmath27 the sets @xmath28 and @xmath29 are of the same order .
if @xmath23 is a gassmann - sunada triple where @xmath26 acts freely and isometrically on a riemannian manifold @xmath30 , then sunada s theorem ( theorem [ thm : sunadapesce ] ) tells us that the quotient riemannian manifolds @xmath31 and @xmath32 are isospectral , and the isospectral pairs arising in this manner are said to be _ sunada - isospectral_. sunada s theorem provided the first systematic method for constructing isospectral manifolds and along with its numerous variants
accounts for many of the isospectral pairs in the literature ( see @xcite for a retrospective on sunada s method ) . in @xcite
we discovered an analogous algebraic condition and theorem concerning the covering spectrum .
specifically , we showed the following .
[ thm : dgs ] let @xmath23 be a triple of finite groups , where @xmath24 and @xmath25 are subgroups of @xmath26 , and suppose @xmath26 acts freely on a closed manifold @xmath30 . if @xmath30 is simply - connected and of dimension at least @xmath33 , then the following are equivalent : 1 .
@xmath24 and @xmath25 are _ jump equivalent _ subgroups of @xmath26 ; that is , for any subsets @xmath34 that are stable under conjugation we have @xmath35 in which case we say that @xmath23 is a _
jump triple_. 2 . for every @xmath26-invariant riemannian
metric on @xmath30 , the quotient riemannian manifolds @xmath31 and @xmath36 share the same covering spectrum .
we then went on to produce examples of gassmann - sunada triples that are not jump triples .
as a consequence we were able to construct sunada - isospectral manifolds with distinct covering spectra in all dimensions greater than or equal to three . however , our method for building such examples appears to fail in dimension two as the arguments underlying theorem [ thm : dgs ] depend in part on the fact that all free homotopy classes in dimension three and higher can be represented by simple closed curves . in this article
we take a different approach in order to establish the following theorem showing that the covering spectrum is not a spectral invariant of a surface .
[ thm : main ] for each integer @xmath37 there exist pairs of closed connected sunada - isospectral surfaces of genus @xmath38 with distinct covering spectra . to prove the result above we construct certain cayley graphs associated to a well - known gassmann - sunada triple ( * ? ? ?
11.4.7 . ) , and endow them with suitably chosen edge lengths so that the resulting length spaces @xmath39 and @xmath40 have distinct covering spectra .
we then build continuous families of riemannian manifolds @xmath41 and @xmath42 that are pairwise sunada - isospectral for each @xmath43 and such that for each @xmath44 the family @xmath45 converges with respect to the gromov - hausdorff metric to @xmath46 as @xmath47 tends to zero .
then , since the covering spectrum is continuous under gromov - hausdorff convergence ( * ? ? ?
* corollary 8.5 ) , it follows that for @xmath47 sufficiently small @xmath48 and @xmath49 are sunada - isospectral manifolds with distinct covering spectra .
currently , it is not clear to us whether any of the aforementioned pairs consist of riemann surfaces .
it would appear to be an interesting problem to determine whether the covering spectrum is a spectral invariant among such spaces .
the outline of this article is as follows . in section
[ sec : covering ] we will formulate the precise definition of the covering spectrum , present a useful algorithm for computing it , and discuss its behavior under gromov - hausdorff convergence . in section [ sec : cayley ] we recall the notion of a cayley graph , we produce the length spaces @xmath39 and @xmath40 mentioned above , and show their covering spectra are distinct . in section [ sec :
surfaces ] we review how the cayley graphs relate to the isospectral problem , and we finish the proof of theorem [ thm : main ] as outlined .
in this section we will recall the definition of the covering spectrum as given in @xcite . since the primary focus of this paper will be the covering spectra of compact semi - locally simply - connected length spaces we will only formulate our definition in this setting .
however , we note that in @xcite we extended this definition of the covering spectrum to arbitrary metric spaces .
we will begin by recalling the notion of a filtration of a group .
let @xmath26 be a group .
a _ filtration _ @xmath50 of @xmath26 is a family @xmath51 of subgroups of @xmath26 , where @xmath52 ranges over an ordered index set @xmath53 , such that @xmath54 whenever @xmath55 .
the _ jump set _ of the filtration is the subset of @xmath53 given by @xmath56 [ exa : filtration ] let @xmath26 be a group and @xmath57 a class function ; i.e. , @xmath58 is constant on the conjugacy classes of @xmath26 .
then @xmath58 induces a filtration @xmath59 , where @xmath60 such filtrations will prove to be useful in our discussion of the covering spectrum .
we now observe that given a connected and compact semi - locally simply - connected length space @xmath61 its metric structure induces a natural filtration ( indexed by the positive reals ) on its fundamental group @xmath62 . indeed , for each @xmath63 let @xmath64 be the open cover of @xmath0 consisting of all the open @xmath5-balls in @xmath61 .
then let @xmath65 be the subgroup of @xmath62 generated by classes of the form @xmath66 $ ] , where @xmath67 \to x$ ] is a loop that is completely contained in some open @xmath5-ball , @xmath68 is a path from @xmath69 to @xmath70 and @xmath71 is its reverse .
the subgroup @xmath65 is normal in @xmath62 and the conjugacy classes of the generators @xmath66 $ ] of @xmath65 represent the free homotopy classes of loops that have representatives completely contained in a @xmath5-ball .
in this fashion we obtain a filtration @xmath72 of @xmath62 .
let @xmath61 be a connected and compact semi - locally simply - connected length space .
we then define
the _ covering spectrum _ of @xmath61 to be the set @xmath73 now , since each subgroup @xmath65 of @xmath62 is normal , the galois theory of covering spaces ( * ? ? ?
* theorem 2.5.13 ) tells us that we may associate to each subgroup a connected regular covering space @xmath74 known as the _ @xmath5-cover _ of @xmath0 .
the @xmath5-cover has the property that if @xmath75 is any covering space that evenly covers every @xmath5-ball of @xmath0 , then @xmath76 is a covering space of @xmath77 ( cf .
* lemma 2.5.11 ) ) .
hence , we obtain a tower @xmath78 of regular covers of @xmath0 and the covering spectrum detects when we have a change in the isomorphism type of the cover .
that is , @xmath79 if and only if for all @xmath80 , @xmath81 and @xmath76 are non - isomorphic as covering spaces . in the event that we are considering a compact riemannian manifold @xmath82
we see that the covering spectrum also contains information about the length spectrum .
[ prop : newgeo ] let @xmath82 be a closed riemannian manifold
. then 1 .
@xmath83 is a subset of the length spectrum .
2 . if @xmath84 , then @xmath8 is the length of the shortest ( smoothly ) closed geodesic in @xmath82 that has a lift to @xmath85 that is not a closed loop . in order to compute the length spectrum it will be useful to introduce the minimum marked length map . let @xmath61 be a connected and compact semi - locally simply - connected length space with fundamental group @xmath86 and simply - connected universal cover @xmath87 .
then the _ minimum marked length map _
@xmath88 is given by @xmath89 ) \equiv \inf_{\tilde{x } \in { \widetilde{x } } } \tilde{d}(\tilde{x } , \gamma \cdot \tilde{x}),\ ] ] where @xmath86 acts on @xmath90 via deck transformations .
the minimum marked length map @xmath88 assigns to each @xmath91 \in \pi_1(x)$ ] the length of the shortest closed geodesic in the free homotopy class determined by @xmath91 $ ] .
since free homotopy classes correspond to conjugacy classes in @xmath86 we see that @xmath58 is a class function .
hence , by example [ exa : filtration ] @xmath58 induces the filtration @xmath92 and we have the following . [ prop : filter ] let @xmath61 be a connected and compact semi - locally simply - connected length space with minimum marked length map @xmath93 .
then @xmath94 the above leads to the following algorithm for computing the covering spectrum of a compact semi - locally simply - connected length space as found in @xcite and @xcite .
[ csa ] let @xmath61 be a compact semi - locally simply - connected length space with minimum marked length map @xmath88 .
then let @xmath95 where we continue until @xmath96 .
now since @xmath0 is a compact length space we know by ( * ? ? ?
* lemma 4.6 ) that the image of @xmath58 is closed and discrete .
it then follows from the fact that @xmath86 is finitely generated that the above process stops in finitely many steps at some @xmath97 , and by the previous proposition we see @xmath98 for an application of the covering spectrum algorithm to the case of flat tori the reader is encouraged to see ( * ? ? ?
* example 4.7 ) .
in particular , this example shows that for a flat @xmath99-torus , the covering spectrum need not be equal to the successive minima of the corresponding lattice @xmath100 in @xmath101 .
we conclude this section by recalling that in proposition [ prop : newgeo ] we observed that for any compact length space @xmath0 we have @xmath102 , where @xmath103 denotes the length spectrum of @xmath0 .
now , it is well known that the length spectrum does not behave well under gromov - hausdorff convergence : lengths may disappear or suddenly appear in the limit ( cf .
* examples 8.1 - 8.3 ) ) .
however , in section [ sec : cayley ] we will use the fact that , in contrast , the covering spectrum is continuous under gromov - hausdorff convergence .
[ prop : ghconv ] if @xmath46 is a sequence of compact length spaces converging to a compact length space @xmath77 in the gromov - hausdorff sense , then the covering spectra converge in the hausdorff sense as subsets of @xmath104 .
that is , if @xmath105 , then @xmath106 where @xmath107 denotes the hausdorff distance between subsets of @xmath104 .
hence , twice the covering spectrum determines a part of the length spectrum of a compact length space that behaves well under gromov - hausdorff convergence .
we refer the reader to sections 7 and 8 of @xcite for the relevant details on gromov - hausdorff convergence and the proof of the proposition above .
in this section we will introduce certain cayley graphs and associated length spaces . in the next section
we will use these examples to construct isospectral surfaces with distinct covering spectra . by a _ graph
_ we will mean a quadruple @xmath108 where @xmath109 and @xmath110 are the sets of vertices and edges ( respectively ) , and @xmath111 and @xmath112 are maps from @xmath110 to @xmath109 associating to each edge its origin and target .
in other words , we consider directed graphs with possible self - edges and multiple edges between the same vertices .
a morphism @xmath113 from a graph @xmath114 to a graph @xmath115 is a pair of maps @xmath116 and @xmath117 for which @xmath118 and @xmath119 .
if @xmath26 is a group acting _ on the right _ on a set @xmath109 , and @xmath120 is a subset , then we can consider the ( _ generalized _ ) _ cayley graph _ @xmath121=(v , e , o , t)$ ] associated to @xmath109 and @xmath122 , where @xmath123 with @xmath124 and @xmath125 .
note that we can label the edges by @xmath122 according to their second coordinate this will be referred to as the _ color _ or _ type _ of the edge and that for every @xmath126 every vertex has a single incoming and outgoing edge of type @xmath127 .
if for @xmath109 we take a finitely generated group @xmath26 with the @xmath26-action by right multiplication and we let @xmath122 be a set of generators of @xmath26 , then we obtain the usual cayley graph @xmath128 $ ] of the group @xmath26 ( cf .
the group @xmath26 then has a left - action by graph automorphisms on @xmath128 $ ] where @xmath26 acts on the vertices by left multiplication .
the action preserves the labeling of the edges by @xmath122 . for any subgroup @xmath129 of @xmath26 the quotient graph @xmath130)$ ]
is then naturally isomorphic to the generalized cayley graph @xmath131 $ ] .
the graph @xmath131 $ ] is also known as the _ schreier graph _ associated to @xmath132 and @xmath122 ( cf .
4 ) ) . in our application , the group @xmath26 will be the group @xmath133 , which we will consider with its right action on the fano plane , the projective plane over the field @xmath134 of @xmath135 elements .
the set @xmath136 of points of the fano plane is labeled by non - zero row vectors of length @xmath33 over @xmath134 , which are multiplied on the right by matrices in @xmath26 . a second set @xmath137 with right @xmath26-action that we will consider is the set of lines in the fano plane .
we can label @xmath137 by the same set of non - zero vectors in @xmath138 where the line through two distinct points @xmath139 is labeled by the unique nonzero vector @xmath140 so that @xmath141 under the standard inner product @xmath142 on @xmath138 ; see figure [ fig : fano ] . by computing the stabilizer in @xmath26 of the element represented by @xmath143 in each of @xmath136 and @xmath137
we see that we have @xmath26-set isomorphisms @xmath144 , and @xmath145 , where @xmath146 let us now consider the subset @xmath147 consisting of the two matrices @xmath148 the action of @xmath149 on the fano plane is given by dotted arrows in figure [ fig : fano ] , while the action of @xmath150 is rotation by 120 degrees counterclockwise . by rearranging the vertices we see that the cayley graph @xmath151 $ ] is the graph in figure [ fig : graph1 ]
where the dotted arrows indicate edges associated to @xmath149 and the solid arrows indicate edges associated to @xmath150 . by tracking the action of @xmath149 and @xmath150 on the lines of the fano plane ( or on the three points each line contains )
we construct the cayley graph @xmath152 $ ] as in figure [ fig : graph2 ] .
these graphs are also given in ( * ? ? ?
11.4.7 ) and @xcite . at ( 58:2 ) @xmath153 ; ( 0,0 ) circle ( 2 cm ) ; ( 90:4 ) ( 210:4 ) node[pos=0.75,above left]@xmath154 ; ( 90:4 )
( 330:4 ) node[pos=0.75,above right]@xmath155 ; ( 90:4 ) ( 90:-2 ) node[pos=0.5,left]@xmath156 ; ( 330:4 ) ( 210:4 ) node[pos=0.75,below , inner sep=3pt]@xmath143 ; ( 210:4 )
( 210:-2 ) node[pos=0.43,below right , inner sep=0pt]@xmath157 ; ( 330:4 ) ( 330:-2 ) node[pos=0.47,above right , inner sep=1pt]@xmath158 ; ( z100 ) at ( 90:4 ) @xmath143 ; ( z010 ) at ( 210:4 ) @xmath155 ; ( z001 ) at ( 330:4 ) @xmath154 ; ( z101 ) at ( 30:2 ) @xmath157 ; ( z110 ) at ( 150:2 ) @xmath158 ; ( z011 ) at ( 270:2 ) @xmath156 ; ( z111 ) at ( 0,0 ) @xmath153 ; ( z100 ) edge[dotted , bend right ] node ( z110 ) ( z110 ) edge[dotted , bend right ] node ( z111 ) ( z111 ) edge[dotted , bend right ] node ( z101 ) ( z101 ) edge[dotted , bend right ] node ( z100 ) ( z011 ) edge[dotted , loop below ] node ( z011 ) ; ( z010 ) edge[dotted , bend right ] node ( z001 ) ; \(4 ) @xmath158 ; ( 1 ) [ above right of= 4 ] @xmath157 ; ( 3 ) [ below right of= 4 ] @xmath156 ; ( 0 ) [ above right of= 3 ] @xmath153 ; ( 6 ) [ right of= 0 ] @xmath143 ; ( 2 ) [ above right of= 6 ] @xmath155 ; ( 5 ) [ below right of= 6 ] @xmath154 ; \(4 ) edge [ loop right , dotted ] node ( 4 ) edge [ ] node ( 3 ) ( 1 ) edge [ ] node ( 4 ) edge [ dotted ] node ( 6 ) ( 3 ) edge [ dotted ] node ( 0 ) edge node ( 1 ) ( 0 ) edge [ loop left ] node ( 0 ) edge [ dotted ] node ( 1 ) ( 6 ) edge node ( 2 ) edge [ dotted ] node ( 3 ) ( 2 ) edge node ( 5 ) edge [ bend left , dotted ] node ( 5 ) ( 5 ) edge [ bend left , dotted ] node ( 2 ) edge node ( 6 ) ; = [ inner sep=2pt , circle , draw , fill = white ] \(3 ) @xmath153 ; ( 0 ) [ right of= 3 ] @xmath156 ; ( 4 ) [ above right of= 0 ] @xmath158 ; ( 6 ) [ below right of= 0 ] @xmath157 ; ( 5 ) [ right of= 4 ] @xmath154 ; ( 2 ) [ right of= 6 ] @xmath155 ; ( 1 ) [ above right of= 2 ] @xmath143 ; \(0 ) edge [ dotted , bend right ] node ( 3 ) edge [ bend right ] node ( 6 ) ( 3 ) edge [ loop left ] node ( 0 ) edge [ bend right , dotted ] node ( 0 ) ( 1 ) edge [ loop right , dotted ] node ( 1 ) edge node ( 2 ) ( 2 ) edge [ bend left ] node ( 5 ) edge [ dotted ] node ( 5 ) ( 4 ) edge [ dotted ] node ( 6 ) edge [ bend right ] node ( 0 ) ( 5 ) edge [ dotted ] node ( 4 ) edge node ( 1 ) ( 6 ) edge [ dotted ] node ( 2 ) edge [ bend right ] node ( 4 ) ; the _ geometric realization _ of a graph @xmath114 is the topological space that one obtains by glueing closed intervals onto the set @xmath109 as follows : for each @xmath159 one adds an interval @xmath160 $ ] identifying @xmath161 with @xmath162 and @xmath163 with @xmath164 .
if one designates a positive real number @xmath165 as the length of an edge @xmath166 then one can give the interval assocated to @xmath166 length @xmath165 and in this way one obtains a length space .
for our cayley graph @xmath167 $ ] we will assign a length @xmath168 to all edges associated with @xmath149 and @xmath169 to all edges associated with @xmath150 , and we thus obtain a length space @xmath170
. the group @xmath26 acts by isometries on @xmath170 and it acts freely , i.e. , no non - trivial element of @xmath26 has a fixed point on @xmath170 .
the quotient length space @xmath171 is the length space @xmath39 associated to @xmath151 $ ] and @xmath172 is the length space @xmath40 associated to @xmath152 $ ] .
[ prop : distinct ] if @xmath173 , then @xmath174 lies in the covering spectrum of @xmath175 but not in the covering spectrum of @xmath176 .
we see from the assumption that @xmath177 represent the lengths of the non - trivial minimal loops of length below @xmath178 .
these lengths form the first column of the table in figure [ fig : table ] . the second and third column in the table lists closed loops in @xmath39 and @xmath40 of the given lengths . in this table @xmath179
denotes the path along the edge associated to @xmath149 leaving from the vertex with label @xmath154 .
one has to inspect the appropriate graph to see what the endpoint of @xmath179 is : in @xmath39 it is @xmath155 and in @xmath40 it is @xmath158 .
paths are composed according to the following convention : if the endpoint of a path @xmath180 is the starting point of a path @xmath181 then @xmath182 runs through @xmath180 at double speed on the interval @xmath183 $ ] and then through @xmath181 on @xmath184 $ ] . by inspecting the graphs in figures [ fig : graph1 ] and [ fig : graph2 ] one deduces that every closed loop in @xmath39 or @xmath40 that is of length below @xmath185 , is freely homotopic to either a constant loop , or to a loop in the table , or to the inverse of a loop in the table in figure [ fig : table ] .
note also that the lengths of the loops in the table are minimal within their free homotopy class . [ cols="<,^,^",options="header " , ] [ fig : table ] denoting the minimum marked length map for @xmath40 by @xmath58 one sees that @xmath186 is the normal subgroup of @xmath187 generated by all homotopy classes of loops that are freely homotopic to one of @xmath188 , which is also equal to @xmath189 for any @xmath190 .
we deduce that @xmath191 is not a jump of the filtration @xmath192 , so that by proposition [ prop : filter ] the number @xmath174 is not in the covering spectrum of @xmath40 . to complete the proof
we will show that the class @xmath12 in @xmath193 of any loop in @xmath39 based at @xmath69 , that is freely homotopic to @xmath194 , does not lie in the normal subgroup @xmath195 of @xmath193 generated by the classes of all loops based at @xmath69 that are freely homotopic to one of @xmath196 . to see this , consider the topological quotient space @xmath197 of @xmath39 that one obtains by contracting the loops @xmath196 to a point .
then @xmath195 lies in the kernel of the induced group homomorphism @xmath198 , but one sees from the graph in figure [ fig : graph1 ] that @xmath12 does not map to the trivial class .
since @xmath194 has length @xmath191 , and this length is minimal in its free homotopy class , it follows with proposition [ prop : filter ] that @xmath174 lies in the covering spectrum of @xmath39 .
suppose that @xmath26 is a group and @xmath122 is a subset of @xmath26 .
for any set @xmath109 with a right action of @xmath26 we defined the cayley graph @xmath121 $ ] with vertex set @xmath109 and edge set @xmath110 consisting of an edge from @xmath199 to @xmath200 for each @xmath126 and @xmath201 .
we can construct a surface out of this graph as follows ( cf .
* chapter 11 ) ) .
suppose that we have a compact connected surface of genus @xmath202 .
if @xmath203 , then we add @xmath99 handles to the surface , and on this surface @xmath30 we choose a riemannian metric .
we now cut open the handles to obtain a compact surface @xmath204 with @xmath205 border components @xmath206 , where for each @xmath207 a homeomorphism @xmath208 is given so that identifying @xmath209 with @xmath210 for all @xmath126 gives us @xmath30 . in figure
[ fig : genus2 ] this is illustrated for @xmath211 and @xmath212 . with this building block
we can now make a surface as follows .
take the surface @xmath213 and for every edge from an element @xmath214 to @xmath215 of color @xmath126 we identify @xmath216 with @xmath217 using the homeomorphism @xmath208 .
we will denote this quotient by @xmath218 $ ] .
it is a compact surface , and it is not hard to see that if @xmath121 $ ] is connected , then the genus of @xmath218 $ ] is @xmath219 .
a riemannian metric on @xmath30 then gives rise to a riemannian metric on @xmath218 $ ] and it follows that any color - preserving graph automorphism of @xmath121 $ ] induces an isometry of @xmath218 $ ] .
thus , for any subgroup @xmath129 of @xmath26 we find that @xmath129 acts freely and by isometries on @xmath220 $ ] and the manifolds @xmath221 $ ] and @xmath222 $ ] are isometric .
this set - up allows us to produce isospectral manifolds by sunada s method .
we recall that a triple of groups @xmath23 where @xmath223 is said to be a _ gassmann - sunada triple _ if for any conjugacy class @xmath224 we have @xmath225 .
this group theoretic notion identifies the configurations of isospectral manifolds in the sense of the following theorem .
the fact that @xmath228 implies @xmath229 is the well - known theorem of sunada @xcite and the reverse implication follows from the work of pesce @xcite . to apply sunada s method to our manifolds associated to cayley graphs ,
suppose that @xmath24 and @xmath25 are two subgroups of @xmath26 so that @xmath230 is a gassmann - sunada triple .
the theorem above then implies that @xmath231 $ ] and @xmath231 $ ] are isospectral .
we can now finish the proof of our main result . the group @xmath133 and the subgroups @xmath24 and @xmath25 of the previous section form a gassmann - sunada triple .
taking @xmath232 , so @xmath212 we see that @xmath233 $ ] and @xmath234 $ ] are connected compact isospectral surfaces of genus @xmath235 .
we will now vary the choice of the riemannian metric on @xmath30 . for each @xmath236
we choose a metric @xmath237 on @xmath30 so that the handle associated to @xmath149 , thought of as a cylinder , is isometric to @xmath238\times { \mathbb r}/\epsilon { \mathbb z}$ ] , and the handle associated to @xmath150 is isometric to @xmath239\times { \mathbb r}/\epsilon { \mathbb z}$ ] , and the diameter of the complement of the two handles is below @xmath47 . as @xmath47 tends to zero , the metric spaces @xmath240 then converge in the gromov - hausdorff sense to a bouquet of two circles of circumferences @xmath241 and @xmath242 .
we obtain induced riemannian metrics @xmath243 on @xmath244 $ ] for @xmath245 and by the remark following theorem [ thm : sunadapesce ] the manifolds @xmath246,g_1^\epsilon)$ ] and @xmath247,g_2^\epsilon)$ ] are isospectral for every @xmath47 . for both @xmath248 and @xmath249
the spaces @xmath250,g_i^\epsilon)$ ] now converge as @xmath251 to the space @xmath252 in the gromov - hausdorff metric . by gromov - haussdorff continuity of the covering spectrum ( proposition [
prop : ghconv ] ) and the fact that @xmath39 and @xmath40 have distinct covering spectra , as shown in proposition [ prop : distinct ] , it now follows that for @xmath47 sufficiently small the isospectral surfaces @xmath246,g_1^\epsilon)$ ] and @xmath247,g_2^\epsilon)$ ] have distinct covering spectra as well .
let @xmath23 be a non - trivial gassmann - sunada triple , and @xmath120 .
suppose we can choose lengths @xmath253 for all @xmath126 such that the induced length space structures on the geometric realizations of the cayley graphs @xmath254 $ ] and @xmath255 $ ] give rise to distinct covering spectra .
then every connected closed surface @xmath30 with distinct marked handles labeled by @xmath122 has a riemannian metric so that @xmath256 $ ] and @xmath257 $ ] are isospectral surfaces with distinct covering spectra . | the covering spectrum is a geometric invariant of a riemannian manifold , more generally of a metric space , that measures the size of its one - dimensional holes by isolating a portion of the length spectrum . in a previous paper we demonstrated that the covering spectrum is not a spectral invariant of a manifold in dimensions three and higher . in this article
we give an example of two isospectral cayley graphs that admit length space structures with distinct covering spectra . from this
we deduce the existence of infinitely many pairs of sunada - isospectral surfaces with unequal covering spectra .
funded in part by the european commission under contract mrtn - ct-2006 - 035495 ] research partially supported by nsf grant dms-0204648 ] research partially supported by nsf grant dms-0605247 and a career enhancement fellowship from the woodrow wilson national fellowship foundation ] |
radio sources from the 3crr sample @xcite whose nuclei are quasars tend to be associated with classical double [ frii , @xcite ] structures .
this sample contains no examples of radio sources clearly having the other characteristic structure , the fri type ( where the brightness decreases with increasing distance from the core ) , powered by quasar nuclei .
this , together with a consistent picture from similar samples , led some to postulate that there is a deep association between the nature of the central engine and whether the radio - structure is type fri or frii @xcite . in principle
, a survey like 3crr is hindered from finding fri radio structures associated with quasars because of two effects .
first , as pointed out by @xcite , fris have lower radio luminosities than the classical double friis ( see fig.[fig : rqq_fr ] ) ; this means they are not found at high redshifts in a bright flux - limited sample .
second , the rarity of quasars means that searching the requisite large volume of space necessarily involves going out to high redshifts .
an alternative means of finding quasars associated with fri radio structures is to begin with an optical survey of quasars and then make follow - up radio observations to establish the nature of their structures . in the case of the bqs sample of quasars
@xcite the radio structures , when prominent , are only of the classical double frii type .
observations to date show the other quasars appear to have weak , though compact @xcite , radio emission associated with their nuclei .
this indicates a seeming gap in radio luminosity for quasars of a given optical luminosity , as pointed out by e.g.@xcite : at radio luminosities corresponding to the fris from the 3crr sample , there are no bqs quasars .
the absence of evidence of quasars with these radio luminosities ( the radio luminosities of fri radio galaxies in 3crr ) was taken by @xcite as evidence of the absence of these objects in the universe .
this led them to postulate that torus opening angles in fris are too small to observe a quasar nucleus : for most angles to the line - of - sight , the nucleus would be obscured by the torus , while for very small angles to the line - of - sight the nuclear emission would be that of a strongly core - boosted bl lac .
the wider opening angles of frii sources permit a significant fraction of these objects to be observed as quasars .
thus , the seeming absence of fri quasars was taken by @xcite as evidence that the nature of the torus is closely linked to the frii / fri transition .
a possible counter to their claim was the discovery by @xcite of an fri radio structure whose nuclear identification , while not a bona fide quasar , exhibits broad lines . with a projected linear size of @xmath4mpc it seemed most unlikely that this object was being viewed at a very small , favourable angle to the line - of - sight .
examples of fri sources having bona fide quasar nuclei would undermine @xcite s theory .
deep , low - frequency , radio surveys are a means of sampling the high-@xmath5 universe necessary to find rare quasars . however , in the low - radio luminosity regime where fri sources are found , current surveys remain challenged by sensitivity . but despite the limited sky coverage of redshift surveys of deep low - frequency radio sources to date , a low - radio luminosity source from the 38-mhz selected 8c sample @xcite , which we find to have an fri structure over 280kpc in extent , is associated with a well - studied quasar which is optically extremely powerful ( * ? ? ?
* unreddened @xmath6 ) . what are the chances that optically - selected quasars will be found to have associated fri structures ?
given that optically - powerful quasars are too rare to populate the nearby universe and the consequences of redshift on surface brightness discussed in
[ sec : redshift ] , the short interferometric snapshots which comprise most observations of these objects to date are inadequate to reveal any such structures .
psychology presumably plays its part in discouraging observers , or at least proposal referees , from making deep radio observations of seemingly ` radio - quiet ' quasars . in
[ sec : meas ] we briefly describe what our deep observations of e1821 + 643 have revealed . in [ sec : results ] we discuss the size and luminosity of its elongated emission and consider the stability in jet - axis direction that is implied by the large - scale radio emission . in
[ sec : redshift ] we consider the detectability , given current technology , of the extended emission if the quasar were located at @xmath7 instead of its true value @xmath8 @xcite . in [ sec : sourcephysics ] we discuss the challenges this quasar presents to the implications of the radio bimodality in quasars , the radio - optical correlation ( e.g. * ? ? ?
* ; * ? ? ?
* ) and the assumed mapping between jet power and observable radio luminosity .
we assume that @xmath9 is the hubble constant in units of @xmath10 , and that @xmath11 and @xmath12 .
the vla in its bna configuration was used to observe e1821 + 643 at 18 cm for 133 minutes on 14 september 1995 ; the primary flux calibrator was 3c48 and the phase calibrator was 1842 + 681 .
the data were reduced using standard procedures using aips .
fig.[fig : images ] shows that at 18 cm low - surface brightness emission ( which contributes to a total of 47.7mjy associated with the quasar ) is extended over 45@xmath13 ( 280kpc ) , way beyond the confines of its host galaxy @xcite .
previously , extended radio emission had only been known to lie approximately co - spatially with its host galaxy @xcite . despite the somewhat elongated nature of the radio emission @xcite found , they thought it likely that this emission was associated with star formation within the galaxy ;
this was concluded because of the proximity of this quasar to the seemingly tight line of correlation between the far - infra - red and radio luminosities of seyfert galaxies and radio - quiet quasars ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the increased radio emission we have discovered lifts this object further away from that tight correlation .
the total luminosity of this source at 18 cm is @xmath14@xmath15 . at the resolution of the 18 cm image ,
the core component accounts for 15.3mjy / beam , i.e. half of the luminosity in the extended emission .
its linear extent is 280kpc in the assumed cosmology , and its width is 60kpc .
thus in terms of radio size , luminosity , the ratio of core to total luminosity , host galaxy and cluster environment @xcite , e1821 + 643 is like other fris .
it seems likely that the jet observed on sub - arcsec scales by @xcite produces the extended emission .
fig.[fig : images ] shows their merlin image in which the jet appears to curve from the line joining the core to e , round towards f. this e f direction is parallel to a
b on larger scales , which in turn probably represents more recent emission than that along the direction delineated by c d .
the axis of the inner merlin jet indicates the most recent directionality of the jet - axis .
thus , two clues from these images taken together may indicate that the jet is precessing : i ) the curvature of the sub - arcsec jet and ii ) the similarity in direction of the jet - axis a b on scales of 10s of arcsec compared with e
f on sub - arcsec scales . this would be in exact analogy to the way the precessing jets of ss433 @xcite give rise to a projected - helical trace on the sky .
it is possible that , as in ss433 , the precession of the jet - axis in this radio - quiet quasar arises from the central engine being a binary .
the changes in jet - axis direction may be the manifestation of the active nucleus containing two super - massive black holes in orbit round one another @xcite .
the outermost ends of the extended emission do not appear to end in shocks ( as might be evinced by an abrupt , rather than a gradual , decrease in surface brightness ) .
this might suggest that the extended emission has grown no faster than the ambient sound speed of the igm .
the sound speed in an ionized gas at temperature @xmath16 k is given by @xmath17 where @xmath18 is the speed of light ; for @xmath19k this gives an assumed transonic advance speed of @xmath20 .
if this were the constant speed at which the plumes had emerged from the radio - quiet quasar , then this would have commenced @xmath21yrs ago so the direction of the jet - axis appears to have remained within a 7degree cone for at least @xmath22yrs . with the a
b distance being somewhat under half the c d distance , this still gives that the characteristic time for the stability of the jet axes is @xmath23yrs .
the precession period of ss433 is 163days , @xmath23 times shorter than the likely period of e1821 + 643 ; thus the ratio of precession periods in these very different objects would be similar to the ratio of their masses .
we now consider an estimate of the stability timescale of the optical emission : @xcite found extended line emission 2.5@xmath13 ( 7.5kpc ) from the nucleus in e1821 + 643 , with line ratios typical of nuclear photoionization , suggesting stability over @xmath24yrs .
the width of this emission perpendicular to the mean jet - axis is similar to the width of the radio emission , suggesting that the radio source may be interacting with this gas .
we now consider whether extended radio emission is likely to be a general characteristic of radio - quiet quasars or whether e1821 + 643 is simply an exceptional case .
radio maps of other radio - quiet quasars from e.g. @xcite and @xcite had an on - source time of only a few minutes , compared with 2hrs for the image we present here : longer integration time brings with it not just a simple increase in signal - to - noise ratio but also substantially better _
uv_-plane coverage .
have previous observations of radio - quiet quasars simply had insufficient surface - brightness sensitivity for diffuse extended radio structures to be detected at even moderate redshifts ?
we performed an experiment similar to one performed by @xcite in a different context : we took our image of e1821 + 643 and calculated what it would look like if instead of being located at the redshift of 0.297 @xcite , it were located at the maximum redshift at which an otherwise identical but unreddened e1821 + 643 could be detected by the bqs survey , namely @xmath25 . for constant @xmath26 over physical sizes exceeding the pixel size ( in kpc at any considered @xmath5 ) , the surface brightness will vary with redshift as @xmath27 ( where @xmath28 is the spectral index ) , thus the surface brightness of the object at @xmath29 would be lower than if it were at @xmath8 by a factor of 5.4 .
the estimated source structure is shown in the left panel of fig.[fig : images ] : at the risk of stating the obvious _ there would be no reason to deduce the presence of 100kpc - scale emission on the basis of such an image_. for sources at moderate to high redshift , it is difficult to observe any low - luminosity extended emission .
the fri radio galaxies in 3c may differ from e1821 + 643 in that the presence or absence of examples clearly indicating precession requires merlin - scale imaging .
the fri radio galaxies in 3c definitely differ from e1821 + 643 in that their optical luminosities are @xmath30 lower , though their radio luminosities are similar . _
do any other ` radio - quiet ' quasars have 100-kpc - scale jets _ ( manifested as fris ) ?
figure[fig : rqq_fr ] clearly demonstrates that with plausible radio spectral indices , and given current sensitivity at low - frequency , the relevant part of parameter space is not sampled , so for the time being this question remains unanswered , though it is possible that some do have such jets ( e.g. those with elliptical hosts ) and some do not ( e.g. those with spiral hosts ) .
only if the answer to this question turns out to be negative ( after deeper exposures at low-@xmath5 with existing facilities , and ultimately with e - merlin , e - vla , lofar and ska , have sampled this parameter space ) would it be possible to posit a bimodality , rather than a continuity , in the luminosities of the jet output of quasars .
implicit in the work of @xcite , @xcite and @xcite was the view that there is a physical dichotomy between ` radio - loud ' objects ( which channel power into jets of the same order as is released by accretion ) and ` radio - quiet ' objects ( whose jet powers are a negligible fraction of the bolometric accretion luminosity ) .
this view does not rest easily with the optically - powerful e1821 + 643 since it has nearly all of the attributes ( [ sec : results ] ) of a low - jet - power fri radio source and is therefore intermediate between high - power friis and classically ` radio - quiet ' objects .
one way of retaining a physical dichotomy would be to postulate a different mapping between jet power and observed low - frequency radio luminosity in ( at least ) this fri .
could fri - like structures from optically - powerful nuclei emerge from different physical processes ?
we speculate that the radio luminosity of some radio - quiet quasars may be reduced by significant precession in their jet - axes .
this precession could arise if insufficient time has elapsed for orbiting black holes in the central engine to have coalesced .
the correlation with radio luminosity of the structural classifications of @xcite is one of the most persistent and robust correlations in astronomy : there are _ no _ examples of highly radio - luminous fris .
this suggests that high jet power is necessary for a highly collimated non - dissipative jet which is ultimately capable of forming compact hotspots and a characteristic frii structure .
as lower jet powers are considered , a jet may be more likely to disrupt and dissipate within a few kpc of the cores , characteristic of fris [ see e.g. @xcite ] , but giving a low luminosity source . the density and inhomogeneity of the environment into which the jet is expanding determines the exact threshold value of jet power above which a jet can give rise to an frii and below which a jet will give rise to an fri . for a given jet power ,
if the jet axis is precessing , this may favour the disruption and hence dissipation of the jet , lowering the likelihood of an frii .
our discovery of an fri radio structure associated with an unusually - nearby optically - powerful quasar , together with consideration of the inadequecy of existing observations of optically - similar objects to reveal similar radio structures , overturn previous assertions that quasar nuclei can not be associated with fri radio structures .
the _ large - scale _ radio emission of e1821 + 643 , elongated way beyond the quasar s host galaxy , is powered by oppositely directed jets which appear to precess on timescales which are scaled up in approximate proportion with the mass of the central engine compared to the galactic radio jetted source , ss433 .
we speculate that the central engine in this quasar may be composed of two black holes which have yet to coalesce ( cf .
any precession in the jet - axis of a radio source may smear out its extended radio emission and promote dissipation in the jet , affecting its potential to form an frii structure , hindering its detectability at radio wavelengths .
the ubiquity of precession as a means by which lower radio luminosity is manifested from highly - accreting yet ` radio - quiet ' quasars remains to be established .
this is essential to developing our understanding of the quasar phenomenon , and beckons the emerging generation of radio telescopes such as e - merlin , e - vla , lofar and ska .
thanks the royal society for a university research fellowship .
the vla is a facility of the nrao operated by associated universities , inc .
, under co - operative agreement with the nsf .
merlin is a uk national facility operated by the university of manchester on behalf of pparc .
we thank the referee for helpful comments . | we present a deep image of the optically - powerful quasar e1821 + 643 at 18 cm made with the very large array ( vla ) .
this image reveals radio emission , over 280@xmath0kpc in extent , elongated way beyond the quasar s host galaxy .
its radio structure has decreasing surface brightness with increasing distance from the bright core , characteristic of fri sources @xcite .
its radio luminosity at 5ghz falls in the classification for ` radio - quiet ' quasars ( it is only @xmath1 see e.g. * ? ? ?
* ) .
its radio luminosity at 151mhz ( which is @xmath2 ) is at the transition luminosity observed to separate fris and friis .
hitherto , no optically - powerful quasar had been found to have a conventional fri radio structure . for searches at low - frequency
this is unsurprising given current sensitivity and plausible radio spectral indices for radio - quiet quasars .
we demonstrate the inevitability of the extent of any fri radio structures being seriously under - estimated by existing targetted follow - up observations of other optically - selected quasars , which are typically short exposures of @xmath3 objects , and discuss the implications for the purported radio bimodality in quasars . the nature of the inner arcsec - scale jet in e1821 + 643 , together with its large - scale radio structure , suggest that the jet - axis in this quasar is precessing ( cf .
galactic jet sources such as ss433 ) .
a possible explanation for this is that its central engine is a binary whose black holes have yet to coalesce .
the ubiquity of precession in ` radio - quiet ' quasars , perhaps as a means of reducing the observable radio luminosity expected in highly - accreting systems , remains to be established . |
j. s. bell , physics ( n.y . ) 1 , 195 ( 1964 ) .
a. aspect , p.grangier , g. roger , phys.rev.lett 47 , 460(1981 ) ; _ ibid .
_ 49 , 91(1982 ) ; p.g .
kwiat , k. mattle , h. weinfurter , a. zeilinger , _ ibid .
_ 75 , 4337(1995);w .
tittel , j. brendel , h. zbinden , n.gisin , _ ibid .
_ 81 , 3563(1998 ) ; g. weihs , t. jennewein , c.simon , h. weinfurter , a. zeilinger , _ ibid .
_ 81 , 5039(1998 ) ; a. aspect , nature ( london ) 398 , 189(1999 ) . | using path integrals we express the quantum nonlocality of ab - effect type in the form of singularity . the gauge - fixing term in path integrals
induce the ab effect in ordinary scattering processes .
this means that all scattering processes are accompanied by nonlocal effect .
the formulae are then extended to theory of fields that additionally include a scalar potential .
it turns out that the degree of freedom of nonlocality in quantum fields is just the degree of the ghosts .
furthermore , renormalization method can be related to this type of nonlocal effect .
* e - mail address : [email protected] _ introduction . _
the nonlocality in quantum mechanics has long been a hot topic in the past decades , and up to date there has been no experiment contradicting the nonlocality ; it refers to the correlation between two particles separated in space such as entanglement derived from the bell theory [ 1 ] and well confirmed in many experiments [ 2 ] .
all these experiments used massless photons as carriers of the states , and the nonlocality is of the bell type .
the study of nonlocality has been also extended to a single photon , which is in a superposition state of two space - separated states such as @xmath0 and @xmath1 , as in the situation when a photon passes through a two - slit plane ( diffraction ) .
this type of nonlocality is called hardy type [ 3 , 4 ] .
the conventional method of studying this type of states is to introduce two vacuum states @xmath2 and @xmath3 at two local regions for @xmath4 and @xmath5 [ 5 - 7 ] .
then the correlation of states @xmath0 and @xmath1 is obvious when one couples them to entangled state @xmath6 @xmath7 @xmath8 .
however , some authors argued that [ 8 ] for a single fermion , the above method is not effective any longer because the massive fermion manifests its nonlocality in a completely different way . since its nonlocality ca nt be transformed to explicit correlation as above , one ca nt measure it at location @xmath4 and @xmath5 and then evaluate how it violates the bell inequalities . through a tedious analysis
, the authors arrived at the conclusion that the only type of nonlocality for a fermion ( except the collapse ) wave is of the ab - effect [ 9 ] type .
the ab effect for fermions has been well demonstrated by experiments [ 10 , 11 , 12 ] .
it appears when a charged particle winds around a magnetic flux completing closed _
n _ ( integer ) loops . to understand the effect in more general circumstances [ 13 ] ,
one may ask what will happen if a charged particle is scattered by a very thin flux such as a spin moment .
the answer to this question provides us with some more useful observables of nonlocality other than the familiar scattering results .
the scattering amplitudes and phase shifts have been studied quantitatively [ 14 , 15 ] , and distinguishing feature is that the scattering result loses the axile symmetry and dependence on the magnetic component appears .
but , how does the nonlocal effect take place * simultaneously * with the scattering process remains elusive as does not make sense to require the charged particle to move around the flux for exactly _
n _ closed loops in a certain plane and then come back to fulfill the scattering process involving other interactions . in this paper we will demonstrate that the nonlocal effect takes place when and only when scattering occurs .
first , we prove that the aharonov - bohm effect can be reproduced if we introduce a singularity to the feynman path integral for ( 2 + 1)-dimensional quantum electrodynamics . secondly , the condition of ( 2 + 1)-dimension can be removed by considering the gauge transformation and gauge - fixing condition . to this end
, the nonlocality like ab - effect can be described using the same lagrangian with only an additional singularity . in above expression
, we have viewed the fermion that is with magnetic moment as rest singularity , the other fermion is initial and final particle that being scattered .
this view makes us aware that any scattering process is accompanied by the ab - like nonlocality .
the resultant formulae can be easily generalized to quantum field and thus to the non - abelian situation .
it is showed that the degree of nonlocality is relevant to the degree of freedom of ghost fields .
the application to the renormalization group is sketched . in
what follows , we employ the path integral method of feynman to interpret the propagating process of the wave function [ 16 ] .
for instance , using the kernel @xmath9 to describe how the wave function @xmath10 has evolved from all states @xmath11 at the moment @xmath12 : @xmath13 and @xmath14e^{is}\ ] ] where @xmath15 $ ] denotes all possible paths , @xmath16 @xmath17 is the action along a certain path , and @xmath18 is the corresponding lagrangian . in calculation , the following two properties of the kernel are very useful . in quantum mechanics level ,
the main contribution to the kernel comes from the paths that nearly make the action @xmath19 in an classical extremum @xmath20 [ 16 ] up to a normalizing factor , i.e. @xmath21if the @xmath22 in @xmath19 is a quadratic form . and due to the kinetic energy @xmath23 , few situations violate the quadratic requirement .
although this simplification originally appeared only as a mathematical technique , it can also simplify our physical consideration evidently , which will be validated later .
another feature of kernel is that _
amplitudes for events occurring in succession in time multiply_. [ 16 ] . for two such succeeding events @xmath24 and @xmath25
, we have @xmath26 \(i ) _ expressing ab effect in feynman integral with a singularity _ : we do the integral of kernel as usual for any lagrangian , and then evaluate the contribution from the singularity and its neighborhood . by virtue of eq .
( 3 ) , we can perceive the removal of singularity as a succeeding event after the just finished integral .
this is equal to the case that a particle walks backward along all the original paths near the singularity after the particle has already completed all the possible paths .
henceforth we call this sort of walking backward as a * return * mechanism .
so , in physical sense , the redundant integral performed around the singularity is removed not by subtracting but by multiplying another kernel .
although the events are in successive order and the latter motion is along the same original paths , the integral over the position space should be written out using different variables for distinguishability .
@xmath27e^{i\int\nolimits_{t_b}^{t_c}l(q^{\prime } , % \dot q^{\prime } ) { \bf d}t}\int [ { \bf d}q]e^{i\int\nolimits_{t_a}^{t_b}l(q,% \dot q){\bf d}t } \nonumber \\ \ & = & \int [ { \bf d}q][{\bf d}q^{\prime } ] e^{is+is^{\prime } } \end{aligned}\ ] ] the measure @xmath15 $ ] stands for all the paths and @xmath28 $ ] represents all the paths covering the neighborhood of the singularity , see graph 1 .
it will be clear later that the action @xmath19 is associated with scattering calculation and @xmath29 associated with nonlocal effect . in general , we are only concerned with the effect of singularity on the wave function in eq.(1 ) .
in such a case we only need to evaluate the first kernel @xmath30e^{is^{\prime } } $ ] .
now suppose an ideal dimensional ( 2 + 1 ) situation in which a low energy electron is scattered by a neutron : a plane is predetermined and the magnetic moment of neutron is vertical to the plane .
a general lagrangian has a simple form as @xmath31 @xmath32 , in which @xmath33 and @xmath34 are scalar and vector potential respectively . for neutron , we know @xmath35 and make @xmath36 .
then it is straightforward to carry out the integral @xmath37 using the above - mentioned first property .
but the result is not what we have expected for the singularity . in a plane with singularity ,
the contribution to the kernel ca nt be approximated using only one classical extremum of @xmath38 .
it is obvious that the paths encompassing a singularity ca nt be topologically invariably shrunk to one classical path .
this point will be clearer with the consideration of gauge fixing condition_. _ using the above mentioned @xmath39 we may get @xmath40_{\text{left}}+[{\bf d}q^{\prime } ] _ { \text{% right}})e^{is_{cl}^{\prime } } \sim e^{is_{cl}^{\text{left}}}+e^{is_{cl}^{% \text{right}}}\text{.}\ ] ] in the integral of action @xmath41 , the second term @xmath42 can just afford the phase difference @xmath43 required by ab - effect .
however , there yet left the first term to be treated in consistency . by a similar process ,
we obtain @xmath44 where @xmath45 is the momentum , and thus the phase difference is @xmath46 ( @xmath47 is differential area element ) , which actually vanishes when the momentum is constant .
obviously when a momentum is split into two on one side and merge on another side of the singularity , the direction of momenta along the two paths will change . using the method of difference
, we can prove that @xmath48 is non - vanishing ( graph 2 ) , so is the nontrivial difference .
this difference from @xmath49 is expected to be responsible for the phase shift due to diffraction , which should be held even when the interaction @xmath50 is absent .
the interaction of scalar potential @xmath33 can be included ( neutron replaced by proton ) without affecting the above result since it is path - independent .
\(ii ) _ generalizing the expression of nonlocality_. in order to generalize the expression , it is necessary to consider the gauge transformation @xmath51 , and gauge fixing condition , e.g. coulomb gauge @xmath52 . concerning the gauge transformation , since the scalar function @xmath53 can be arbitrarily determined
, it can directly reduce the degree of freedom of @xmath54 by one .
therefore , the vector @xmath54 can directly be chosen as @xmath55 , and thus in a plane .
on the other hand , we know that any two - particle scattering must occur in a plane which ca nt be predetermined but actually exists .
so as an effective choice ( equivalent way ) , make all the @xmath54 s are parallel to the scattering plane .
so , the above assumption for ab effect that the scattering must happen in a plane can actually be removed .
additionally , @xmath52 suggests that the vector @xmath54 behaves like a tangent of magnetic lines without divergence ; and hence form closed loops . to this end , the condition of ab - effect , in a plane and close loop , is automatically satisfied by two - particle scattering with regard to the gauge transformation and gauge fixing condition .
these two constraints reduce the freedoms of vector @xmath56 from three to one , identical to the freedom along a loop .
we note that the freedom for gauge transformation is the very nonlocal degree of the fermion . on an alternative viewpoint ,
since the choice of function @xmath53 for the gauge transformation @xmath57 is arbitrary , @xmath54 can be replaced by @xmath58 at every point in a continuous manner .
thus the discussion of @xmath54 also applies to @xmath58 , i.e. the @xmath58 is in the scattering plane with @xmath59 . a change of @xmath60 in @xmath54 transform the wave function @xmath61 to @xmath62
. the movement will not be trivial if a singularity exists for it produces observable interference stripe of the ab - effect . in qed
the coulomb gauge is equivalent to the lorentz gauge @xmath63 , so the above argument _ can be extended to include scalar potential_. we do this at the cost of giving up the time arrow , and the plane in which scattering and diffraction occur is now a 3-dimensional super plane .
the condition @xmath59 becomes @xmath64 , accordingly .
similarly , the two constrains @xmath65 and @xmath66 will reduce the freedoms of @xmath67 from four to two . in this way
, the scattering considered here is automatically extended to the general cases of a two - fermion scattering . to extend the above argument to field theory , let s write down the gauge condition in terms of faddeev and popov form
[ 17 ] . in qed ,
the form is @xmath68\int [ d\theta ] \delta ( \partial _ \mu ( a^\mu ) ^\theta ) = 1,\ ] ] where @xmath69 $ ] is the jacobian @xmath70 which is independent of @xmath71 in abelian case , and hence gauge invariant .
eq .
( 8) is demanded to be gauge invariant , so the gauge invariance of @xmath72\delta ( \partial _ \mu ( a^\mu ) ^\theta ) $ ] automatically restores the result @xmath73 .
the gauge condition @xmath74 can be controlled in experiments by specifying the final outcome , but the resultant path of @xmath75 is a pure gauge property that ca nt be fixed by using local experiments , i.e. the function @xmath53 ca nt be determined locally .
if considering the integral for the grassman variables @xmath76 , @xmath77det@xmath5 [ 18 ] , @xmath69 $ ] in eq .
( 8) can be expressed using degree [ 18 ] of freedom of the ghost as @xmath78[{\cal d}c]e^{i\int d^4x\,\bar c(x)\box c(x)}\ ] ] in the case of qed the above integral only contains dynamical term that can be absorbed into normalization constant .
we reserve its explicit form here to see clearly that the freedom of ghosts ( @xmath79 and @xmath80 ) is just the nonlocal degree of freedom . and
the degree of freedom of an electron is 4 , larger than the ghosts . from amplitude @xmath81 it is straightforward to extend the above discussion to field theory .
notice the form of eq.(1 ) and the expansion property of quantum mechanics @xmath82 , every element of the s - matrix can be written as @xmath83 now the integral measurement should be replaced by the fermion and boson fields instead of configuration space . for qed , @xmath84 has the form @xmath85[{\cal d}{\em \psi } ] [ { \cal d}{\em a}]e^{is}$ ] , where the lagrangian in @xmath19 now becomes @xmath86 with @xmath87 and @xmath88 .
a gauge - invariant lagrangian made by imposing gauge fixing condition and adding the ghost [ 19 , 20 ] has a general form @xmath89 a lagrangian with ghost field is convenient to do brst transformation to get identities used in renormalization . here
we skip the details along this context .
notice that the term @xmath90 in eq .
( 12 ) and the term @xmath91 in eq.(8 ) are both responsible for gauge invariance after the gauge fixing term having been added [ 19 ] . comparing the freedom of @xmath92 and ghost field @xmath93 , it is found that the ghost degree is very the nonlocal degree . in qed , this degree of freedom is 2 , one for scalar , one for vector .
the above formulae can be further extended to the non - abelian case ( e.g. qcd ) by only changing @xmath94 and @xmath95 to @xmath96 and @xmath97 , in which the @xmath98 is the generator of the gauge group and @xmath99 is the corresponding structure constant .
the terms in lagrangian of @xmath100 is the same as @xmath101 ; however , the integral measurement should be independent .
furthermore , at the neighborhood of singularity , the fermion wave function and thus the fermion field should vanish . here
let s again concentrate on the degree of freedom for nonlocality . for non - abelian field , eq.(8 )
should be changed to @xmath102\delta ( g(a^\alpha ) ) \text{det}(\frac{\delta g(a^\alpha ) } { % \delta \alpha } ) = 1\text{,}\ ] ] in which the jacobian det@xmath103 is related to @xmath104 and ca nt be taken out of the integral as @xmath105 $ ] in qed . for qcd ,
the gauge transformation is @xmath106 from this expression we can get the analogous result of qed in lorentz gauge : @xmath107 similarly eq.(10 ) takes the form : @xmath108[{\cal d}{\it c}]e^{i\int d^4x\,\bar c(x)(-\partial ^\mu d_\mu ) c(x)}\ ] ] accordingly gauge - invariant lagrangian becomes [ 18 ] @xmath109 the fact that the degrees of the ghost fields are responsible for the nonlocal degrees is directly extended from qed . in qcd , the degree of freedom for ghost fields is 16 . here
the discussion of nonlocality is more complex for the ghost fields are coupled with the vector field , which is one of the main features in non - abelian field theory .
the relationship between the wilson loop and berry phase [ 21 ] belongs to 2-dimensional out of 4-dimension case of the spatial degrees .
and the degree of quarks here is @xmath110 , smaller than @xmath111 .
it may be of this reason that quarks are confined .
\(iii ) _ new hamiltonian_.
let s start from the eq.(4 ) .
since the two lagrangians in actions @xmath19 and @xmath29 have the same forms , the same integrand will certainly induce the same integral resultant function , in which only boundary values of @xmath112 and @xmath113 are different . for example , if resultant function is @xmath114 and @xmath115 , then for the first action , @xmath116 , and for the second , @xmath117 .
to finish the integral of [ @xmath118 and [ @xmath119 , let s divide the time interval into many parts as , @xmath120\sim [ t_1,t_2,\cdots , t_l,\cdots , t_m,\cdots , t_n]$ ] .
then the integral measurement [ @xmath118 changes with the time division to @xmath121=dq_2dq_3\cdots dq_l\cdots dq_m\cdots dq_{n-1}$ ] . for the division of @xmath122 $ ]
, we determine the intervals of time so coincident with the time interval between the @xmath123 $ ] as to reflect the * return * mechanism : if @xmath124\sim [ t_1^{\prime } , t_2^{\prime } , \cdots , t_n^{\prime } ] $ ] , then make @xmath125 , @xmath126 , @xmath127 , @xmath128 , @xmath129 , which means @xmath130=dq_{m-1}dq_{m-2}\cdots dq_{l+1}$ ] . and
thus applying the second property of feynman integral and multiplying every integral results together yield @xmath131[{\bf d}q^{\prime } ] e^{is+is^{\prime } } \nonumber \\ \ & = & \int dq_2dq_3\cdots dq_l\cdots dq_m\cdots dq_{n-1}e^{i(s_{12}+s_{23}+\cdots + s_{n-1,n})}\times \nonumber \\ & & \
\ \int dq_{m-1}dq_{m-2}\cdots dq_{l+1}e^{i(s_{m-1,m-2}+s_{m-2,m-3}+\cdots + s_{l+2,l+1})}\end{aligned}\ ] ] where @xmath132 represents the integral for action between @xmath133 $ ] .
it should also be noted that , according to the * return * mechanism and our design of division , @xmath134[for @xmath135@xmath136[for @xmath84]@xmath138 .
equally , if the integrand for @xmath139 is @xmath140(@xmath141 is a singularity region ) , then there will be no paths between the region @xmath142 $ ] .
in other words , the paths that contribute nothing to the kernel will only add a factor @xmath143 to the kernel .
so @xmath144or @xmath145 and is absorbed in normalized factor .
for this reason the integral of eq.(18 ) has an equivalent expression , @xmath146[{\bf d}q^{\prime } ] e^{is+is^{\prime } } = \int [ { \bf d}% q]e^{i(s - s_d)}=\int [ { \bf d}q]e^{i\int ( { \frak l}-{\frak l}^{\prime } ) dt}\ ] ] now the integral measurement is over all the region of space , and @xmath147 is the part which is nontrivial only in singular region .
the same analysis applies equal to quantum fields : changing the integral measurement and extending the variables in @xmath148 and @xmath149 directly to fields quantities , and make fermion fields vanish at the singularity point correspondingly .
eq.(11 ) suggests that future calculations may consist of two steps .
first , we calculate the scattering amplitude @xmath84 as usually done ; then , we consider the @xmath100 to include nonlocal effect . to calculate the correction of higher order than the tree level , then , the amplitude @xmath84 and @xmath150 have to be considered together . for the case of ultraviolet divergence in qed and the infrared behavior in qcd , the * return * region for @xmath151 may just be the forbidden region for @xmath84 , whose contribution should be cut off for its non - physical meaning . in this sense
, we make the lagrangian in @xmath147 as the counter terms for that in action @xmath19 in @xmath101 , and the lagrangian has the same form as that in @xmath19 with only some renormalization constants multiplied to the corresponding terms .
the new lagrangian means a new hamiltonian , which contributes only to nonlocal phase as in berry phase , and does nt contribute to the transitional amptitude . in this sense ,
the renormalization and the nonlocality have been unified to an expression . here
we will not reiterate the lengthy procedure of renormalization calculations .
it is easy to recognized that the cutoff for renormalizationare determined by the nonlocal region which we are really concerned . knowing the geometry of this region can enable us to distinguish what is physical from what is non - physical , and justify the effective region that renormaliztion group can be properly used . using the relation @xmath152 can only get a rough estimation in purterbative situation .
further investigations on this issue are under way . in above argument
we have employed the case that electrons(or protons , as in ref [ 13 ] ) are scattered by neutron , but with the generalized formulae , there is no constraints for the involved particles .
the argument can be readily extended to any scattering of two particles that carry both flux and charge[14 , _ commence ( i ) _ ] . in summary , in this paper we argue that any scattering process must be accompanied by the nonlocal effect which may induce resultant observables . and it is also achieved that the degree of freedom of nonlocality is equivalent to the degree of the ghost field , and also the gauge degree of freedom .
the understanding of the renormalization along this context is also sketched .
though having been unified in the same path integral frame , the nonlocality ( diffraction or ab - effect ) is really different from the scattering .
the former is sensitive to measurement and thus unobservable locally , but the latter can be detected anyway .
now the dissertation can be unified using the gauge transformation even in higher - dimension unitary space . |
the mond scheme ( modified newtonian dynamics ) was put forth by milgrom in ref .
@xcite in order to phenomenologically explain two basic facts concerning spiral galaxies without resorting to the concept of hidden dark matter : the asymptotic flatness of the rotation curves of spiral galaxies and the tully - fisher law which is a well - defined relationship between the rotation velocity in spiral galaxies and their luminosity . viewed as a modification of gravity . ]
, mond predicts that the gravitational acceleration @xmath7 felt by a particle in the field of a distribution of mass is a_g=,where @xmath8 is the newtonian acceleration , @xmath9 is an acceleration scale which different , independent ensembles of observations set to @xmath10 m s@xmath11 , and @xmath12 is an interpolating function which approximates 1 for @xmath13 , i.e. for accelerations larger than @xmath9 ; for @xmath14 @xmath15 , so that in such a strongly mondian regime @xmath16
. here we wish to investigate the deep newtonian regime ( @xmath13 ) in view of recent advances in planetary orbit determination occurred for the inner planets of the solar system . for a quite general class of interpolating functions ,
@xmath12 can be cast into the form ( x)1-k_0()^n , which yields a modified gravitational acceleration a_ga_n.note that the most commonly used expressions for @xmath12 , i.e. ( x)=,and ( x)=,reduce to for @xmath17 and @xmath18 , respectively , in the appropriate limit . for a recent review of many aspects of mond as various attempts to theoretically justify it see ref .
@xcite and ref .
it can be shown that affects the orbital motion of a test particle in the field of a central mass @xmath19 with a secular rate of the longitude of the pericenter = -a^2n-+ ( e^2)q a^z , where @xmath20 , @xmath21 and @xmath22 are the orbit s semimajor axis and eccentricity , respectively , and q- , z2n - .the expression of can be fruitfully used in conjunction with the corrections to the known netonian / einsteinian secular rates of the perihelia of the inner planets of the solar system phenomenologically estimated as least - square solve - for parameters in ref .
@xcite by fitting a planetary data set spanning almost one century with the dynamical force models of the epm2004 ephemerides .
the same procedure was followed also for some of the outer planets @xcite . since such models fully include newtonian and einsteinian gravity , such estimated extra - precessions account , in principle , for all unmodelled physical effects possibly present in nature like , e.g. , mond .
if and when other groups will independently estimate their own extra - precessions of perihelia it will be possible to enlarge and enforce the present test . for a search of other mond - like effects in the solar system see ref .
@xcite , while the possibility of testing a mondian violation of the newton s second law in a terrestrial environment is discussed in ref . @xcite . by assuming that mond does not violate the equivalence principle , i.e. @xmath1 and @xmath2 are not body - dependent
, it is possible to consider for a generic pair of planets a and b the ratio of their perihelion rates getting does not experience spatial variations , according to the mondian point of view for which modifications of newtonian gravity does not depend on distance but on acceleration only . of course , our test based on the ratio of the perihelia is valid for the case @xmath23 . ]
= ^z.by defining , and _
n^z , it is possible to construct _
n-_n;if , for a given value of @xmath2 , the quantity @xmath24 , computed with the extra - rates of perihelia of ref .
@xcite estimated without including any exotic acceleration with respect to standard newton - einstein one in the suite of the dynamical force models used to fit the data , turns out to be incompatible with zero within the errors , i.e. if @xmath25 , that value of @xmath2 must be discarded .
note that our test makes sense for @xmath26 , as it is just the case from galactic data .
the uncertainty in @xmath27 can be conservatively assessed as _ n+ _
n , with ||,_nz_n.the linear sum of the individual errors in accounts for the existing correlations among the estimated perihelia corrections , which reach a maximum of about 20@xmath28 for mercury and the earth ( pitjeva , private communication 2005 ) . by choosing a = mars and b = mercury , from table [ tavola ] .semimajor axes @xmath21 , in au ( 1 au@xmath29 m ) , and phenomenologically estimated corrections to the newton - einstein perihelion secular rates , in arcseconds per century ( ) , of the inner @xcite and some of the outer @xcite planets . also the associated errors are quoted : they are in m for @xmath21 ( see ref .
@xcite ) and in for @xmath30 @xcite . for the semimajor axes they are the formal , statistical ones , while for the perihelia of the inner planets they are realistic in the sense that they were obtained from comparison of many different solutions with different sets of parameters and observations ( pitjeva , private communication 2005 ) .
the errors quoted here for the perihelia of the outer planets are the formal ones re - scaled by a factor 10 in order to get realistic estimates for them . [
cols="^,^,^,^,^",options="header " , ] [ tavola ] we get figure [ figura1 ] @xmath31 for mars and mercury and @xmath3 . as can be noted , @xmath32 for all values of @xmath2 .
, width=491,height=415 ] in which we plot @xmath31 for @xmath3 .
it turns out that the corrections of order @xmath33 to are negligible in the sense that their inclusion in the calculation does not alter the results for @xmath31 .
moreover , @xmath34 is far smaller than @xmath35 even by
re - scaling the formal errors in the semimajor axes by a factor 10 or more . as can be noted ,
@xmath36 is always incompatible with zero at much more than @xmath4 level , thus ruling out the interval [ 1,2 ] for @xmath2 .
figure [ figura2 ] refers to a = earth , b = mercury : it yields the same conclusions .
@xmath31 for the earth and mercury and @xmath3 . as can be noted , @xmath32 for all values of @xmath2.,width=491,height=415 ]
it can be shown that the same holds also for a = mars , b = venus and a = earth , b = venus .
it maybe interesting to note that , although the errors in the perihelion rates quoted in table [ tavola ] are not the mere , formal ones , should one decide to re - scale them by a factor 10 our conclusions would remain unchanged , apart from the number of @xmath37 which would pass from about 180 to 18 ( a = mars , b = mercury ) , or from 60 to 6 ( a = earth , b = mercury ) for @xmath38 . another way to tackle the problem is to construct a @xmath39-like quantity @xmath40 defined as follows .
the role of the observables @xmath41 is played by the ratios of the perihelia precessions @xmath42 for all the @xmath43 pairs a / b and b / a of planets , including also the gaseous giant ones whose data are quoted in table [ tavola ] . the computed , or predicted , quantities @xmath44 are the ratios @xmath45 for the corresponding pairs of planets .
the errors @xmath46 are @xmath47 because of the negligible impact of the uncertainties of the semimajor axes , while we will assume the number of different pairs constructed , i.e. 30 , for the number of degrees of freedom @xmath48 .
thus , we obtain = _ k=1 ^ 30 10 ^ 3 , n=2,which largely confirms our previous conclusion .
it turns out that @xmath49 also for @xmath50 .
a major outcome of our analysis is that @xmath38 is neatly ruled out independently of @xmath1 and @xmath9 , contrary to what obtained in ref .
@xcite in which only the perihelion of mars was used by keeping fixed @xmath51 ( and using the commonly accepted value @xmath10 m s@xmath11 ) .
the results of the analysis presented here seem to suggest that a reconsideration of the matching between the deep newtonian and mondian regimes should be looked for .
however , caution is in order because the present analysis is based upon the extra - precessions of perihelia estimated by only one team ; it would be of great importance to have at disposal corrections to the perihelia rates determined independently by other groups as well . | in this letter we investigate the deep newtonian regime of the mond paradigm from a purely phenomenological point of view by exploiting the least - square estimated corrections to the secular rates of the perihelia of the inner and of some of the outer planets of the solar system by e.v .
pitjeva with the epm2004 ephemerides . by using @xmath0 for the interpolating mondian function , and by assuming that @xmath1 , considered body - independent so to avoid violations of the equivalence principle , experiences no spatial variations throughout the solar system we tightly constrain @xmath2 with the ratios of the perihelion precessions for different pairs of planets .
we find that the range @xmath3 is neatly excluded at much more than @xmath4 level .
such a test would greatly benefit from the use of extra - precessions of perihelia independently estimated by other groups as well .
# 1 # 1 # 1#2_#1^#2 # 1eq .
( [ # 1 ] ) # 1#2eq .
( [ # 1])-eq .
( [ # 2 ] ) # 1eq .
( [ # 1 ] ) # 1#2eq .
( [ # 1])-eq .
( [ # 2 ] ) # 1 # 1#2 # 1#2 # 1#2 } \def\ear{\end{aligned}\ ] ] # 1#2#1#2 # 1 # 1[#1 ] 2@xmath5 # 1 * * + + + lorenzo iorio + _ viale unit@xmath6 di italia 68 , 70125 + bari , italy + tel./fax 0039 080 5443144 + e - mail : [email protected]_
pacs : 04.80.-y ; 04.80.cc ; 95.10.ce ; 95.10.eg + keywords : experimental studies of gravity ; experimental tests of gravitational theories ; celestial mechanics ; orbit determination and improvement |
* * * * he purpose of this paper is to discuss von neumann s @xcite rejection method to generate a random variable @xmath2 under the random bit model . in this model
, we have access to an infinite sequence of independent random bernoulli@xmath3 bits , and are interested in the complexity as measured by the number of bits used until halting . for integer - value random variables @xmath2 , the entire story is known .
knuth and yao @xcite obtained lower bounds on the expected number of bits for the exact simulation of @xmath2 , and exhibited optimal algorithms . on the other hand , for random variables @xmath2 with a density on @xmath1 , no exact algorithm exists , since any output of an algorithm delivers a function of a ( possibly random ) number of random bits .
thus , it is necessary to introduce the notion of an @xmath0-approximation ( see section [ sect : bisection ] ) using wasserstein @xmath4-distance ( devroye and gravel @xcite and rachev and rschendorf @xcite ) . a naive application of the rejection method one of the most often used methods in simulation leads to errors and inconsistencies .
to deal with this , we introduce the notion of an oracle that computes the supremum and infimum of a function over any rectangles of @xmath1 ( see section [ sect : compactcase ] ) .
the oracle can be used in conjunction with a quadtree partition of the space to design a rejection algorithm that is guaranteed to deliver an @xmath0-approximation ( sections [ sect : compactcase ] and [ sect : noncompacta1 ] ) .
we show that is valid whenever @xmath5 is riemann - integrable and derive expected complexity bounds ( in terms of the number of random bits consumed ) that are close to the universal lower bounds of @xcite .
we believe that random number generation libraries should offer the possibility of specifying any @xmath0no matter how small as an input .
current practice entirely disregards the effects of approximations . in this paper and a companion paper @xcite
, we take a first small step towards this goal .
applications in physics ( see @xcite who requires and develops @xmath0-accurate normal generators ) , quantum computing ( see brassard , devroye , and gravel @xcite ) , and other areas of science demand very precise computations .
much more is needed of course , especially when many random variables are combined in scientific computation how does an @xmath0-approximation propagate through a system , for example ?
consider two probability vectors @xmath6 , and @xmath7 , where @xmath8 for a finite constant @xmath9 .
then von neumann s rejection method can be reformulated as follows , if @xmath9 is explicitly known : [ algo : orinatumsupolov ] generate @xmath2 according to @xmath10 . generate @xmath11 uniformly on @xmath12 $ ] .
@xmath2 the expected number of iterations in this algorithm is @xmath9 .
the returned random variable , @xmath2 , has distribution @xmath6 .
knuth and yao @xcite showed that to generate @xmath2 , the expected number of random bits required by any algorithm is at least the entropy of @xmath2 , @xmath13 they also exhibited an algorithm that requires an expected number of bits not exceeding @xmath14 .
note that given @xmath2 , the decision @xmath15 can be made using @xmath16 expected random bits because to decide @xmath17 given @xmath2 follows a geometric law with parameter @xmath18 ; we compare the @xmath19 bit of @xmath11 with the @xmath19 bit of @xmath20 until both do nt agree for integers @xmath21 ( see , e.g. , devroye and gravel @xcite ) .
the expected number of random bits needed by this implementation is not more than @xmath22 this is usually quite far from the lower bound of knuth and yao , @xmath23 .
it is worth to mention an application of the rejection method in the bit model to the simulation of physical phenomena and to communication complexity in brassard , devroye , and gravel @xcite .
the remainder of the paper is concerned with the case in which @xmath2 has a density @xmath5 . since such @xmath2
can not be generated exactly by any algorithm , it has to be approximated in some manner by a discrete random variable , and thus we require an appropriate and as it turns out , nontrivial
generalization of algorithm [ algo : orinatumsupolov ] .
the building block for continuous random generation is the bisection algorithm , which is mathematically equivalent to an algorithm given in devroye and gravel @xcite .
we develop a version here that is convenient for later use .
the analysis of theorem [ thm_bisect ] below is new .
consider an algorithm for generating a random variable @xmath2 with density @xmath5 on @xmath1 that has the following property : for a given @xmath24 , it outputs a random variable @xmath25an @xmath0-approximation of @xmath2such that there exists a coupling of @xmath26 with @xmath27 where @xmath28 is the @xmath4-distance in @xmath1 .
it is understood that necessarily , @xmath29 is discret since it is a function of a ( random ) finite number of random bits .
we call @xmath29 an @xmath0-approximation of @xmath2 . devroye and gravel @xcite showed that if @xmath30 is the random number of bits needed by any algorithm for generating such an approximation @xmath29 , then @xmath31 where @xmath32 is the differential entropy of @xmath5 , @xmath33 the lower bound is valid under a technical condition known as rnyi s condition ( see rnyi @xcite and csiszr @xcite ) , namely that the entropy of the discrete random variable @xmath34 , which takes values on the grid of all integer - valued vectors of @xmath1 , be finite .
the lower bound ( [ lower_bnd ] ) serves as a guide to calibrate and compare the rejection algorithms presented in this paper .
it is especially crucial to match its main term , , without an extra multiplicative constant .
we require an auxiliary set of results on bisection algorithms for generating a random variate that is an @xmath0-approximation of a random variable @xmath2 with a continuous ( not necessarily absolutely continuous ) distribution function @xmath35 on a compact interval @xmath36 $ ] of length @xmath37 .
the bisection algorithm [ bisecto_algo ] assumes that we have access to both @xmath35 and @xmath38 .
@xmath39=[0,1]$ ] @xmath40 @xmath29 @xmath41 .
@xmath43 $ ] @xmath44=\big[g(a),g(z)\big]$ ] @xmath45 $ ] @xmath46=\big[g(z),g(b)\big]$ ] [ thm_bisect ] for the bisection algorithm [ bisecto_algo ] applied to any distribution with support on an interval of length @xmath47 , and halted as soon as an interval of length less than or equal to @xmath48 is reached , we have 1 .
if @xmath29 denotes the center of the halting interval , then there exists a coupling between @xmath2 and @xmath29 such that @xmath49 .
2 . if @xmath30 denotes the number of bits used , then @xmath50 where @xmath51 .
we note here that in general this bound can not be improved by more than @xmath52 bits .
just consider the uniform distribution on @xmath53 $ ] . since all intervals have length to @xmath54 after @xmath55 were used , we have @xmath56 the bisection method yields , very naturally , a full binary tree , i.e. , one in which all internal nodes have two children .
each internal node corresponds to a subinterval of @xmath36 $ ] of length greater than @xmath48 , the root represents the original interval @xmath36 $ ] of length @xmath47 , and leaves represent intervals of length less than or equal to @xmath48 that cause an exit . upon exit ,
the random variable @xmath29 can be coupled with @xmath2 such that @xmath49 , because at every iteration , the random binary choice picks the correct interval , @xmath57 $ ] or @xmath58 $ ] , with the correct probability @xmath18 .
one could thus define @xmath2 as the limit of @xmath59 when the algorithm is run without halting . since @xmath29 is the midpoint of an interval of length at most @xmath48 that also contains @xmath2
, we must have @xmath60 .
this shows part ( i ) . to prove part ( ii ) , let us denote the set of leaves by @xmath61 , and the set of internal nodes ( i.e. , all non - leaf nodes ) by @xmath62 .
the depth of node @xmath63 is denoted by @xmath64 .
it is of course possible that @xmath62 and @xmath61 are both infinite .
however , one has that in all cases , @xmath65 because the leaves form a non - overlapping covering of @xmath36 $ ] so that the bisection method a random walk started at the root and halted when a leaf is reached always stops . also , @xmath66 in the next chain of inequalities , we define @xmath67 i.e. , the number of internal nodes at depth @xmath68 in the tree .
using kraft s inequality ( see , e.g. , cover and thomas @xcite ) , which states that for any binary tree , @xmath65 we have , with @xmath69 denoting the set of ancestors @xmath63 and @xmath70 denoting the set of descendants of @xmath71 , that , since @xmath72 , @xmath73 observe that at depth @xmath68 , all intervals associated with nodes are disjoint , and thus , since each interval node corresponds to an interval strictly larger than @xmath48 , @xmath74 also , @xmath75 because we have a binary tree .
hence , @xmath76 we deduce , using @xmath77 that @xmath78 to see this , treat the cases @xmath79 and @xmath80 separately .
in this section , we assume that @xmath5 is riemann - integrable and supported on @xmath12^{d}$ ] . note that this is equivalent to the assumption that @xmath5 is almost - everywhere continuous , bounded , and supported on @xmath12^{d}$ ] .
our algorithm requires an oracle a black box that for any closed rectangle @xmath81 gives @xmath82 if @xmath83 , then that oracle coincides with a standard function evaluation . without the possibility of computing infimum and supremum of the density @xmath5 over compact subintervals of the domain of @xmath5 , sampling absolutely continuous distribution using the rejection method seems to be impossible in total generality .
we will also track complexity in terms of the number of uses of the oracle before halting , and call it @xmath30 .
one use of the oracle reveals @xmath84^{d}}{f(x)},\ ] ] which is a finite number by assumption ( riemann - integrable functions are bounded by definition ) . at once
, we have a simple bound for applying the rejection method : @xmath85 .
algorithm [ orig_vnabc ] shows the original algorithm by von neumann @xcite ( see also devroye @xcite ) of the standard rejection algorithm .
generate @xmath2 uniformly on @xmath12^{d}$ ] .
generate @xmath11 uniformly on @xmath12 $ ] .
@xmath2 since we can not generate @xmath2 and @xmath11 with infinite precision , at least two modifications are needed .
one modification is to take into account the precision @xmath0 desired for @xmath2 , and another modification is to take into account that bits of @xmath11 are generated sequentially .
appendix [ app_b ] , which is in this article for pedagogical purpose , gives more insights on wrong and naive modifications that someone could be tempted to do .
we can consider the initial rectangle @xmath86^{d}\times [ 0,c],\ ] ] and its @xmath87 child rectangles defined by the @xmath87 quadrants centered at the center @xmath88 of @xmath89 : @xmath90 in the data structure literature , such a partition , when applied recursively , leads to a quadtree ( see , e.g. , samet @xcite ) .
any subsequent rectangle can be split again about its center point , and so forth , in the manner of infinite quadtree @xmath91 on @xmath92 as illustrated by figures [ fig_2dim_decomp_for_quadtree ] and [ fig_quadtree_decomp_example ] . in von neumann s algorithm , deciding if @xmath93 for @xmath94 is equivalent to finding a rectangle @xmath95 in the quadtree @xmath91 with the property that either @xmath96 or @xmath97 however , this must be done carefully , without overlapping rectangles .
thus , we must trim @xmath91 , and associate the exiting rectangles @xmath95 with the leaves .
thus , @xmath98 and @xmath99 below , we will see that riemann - integrability of @xmath5 suffices for this decomposition .
when a rejecting rectangle is found , the procedure is repeated .
when an accepting rectangle is found , say , @xmath100}\times [ \alpha,\beta],\ ] ] it suffices to generate @xmath101}$ ] such that @xmath102 where @xmath103 is uniform on @xmath104}$ ] and coupled with @xmath29 .
it is easy to see by the triangle inequality that @xmath29 is then coupled with @xmath2 having density @xmath5 such that @xmath105 to achieve ( [ xstar_xepsilon ] ) , use bisection for each dimension separately . by theorem [ thm_bisect ] ,
the expected number of bits needed is bounded by @xmath106 for @xmath107 .
if @xmath108 , then bisection requires no bit , so that we conclude that the expected number of bits does not exceed @xmath109 we note that the checks @xmath110 and @xmath111 can be carried out using our oracle .
let @xmath112 then ( [ check_one ] ) holds if @xmath113 , and ( [ check_two ] ) holds if @xmath114 .
@xmath115^{d}\times\big[0,\mathop{\sup_{}}_{x\in[0,1]^{d}}{f(x)}\big]$][init_line_reject_algo ] @xmath116 call the oracle that returns @xmath117 and @xmath118 which are in turn used by the following branching statement .
@xmath119 @xmath120 @xmath121 select one vertex @xmath71 of @xmath95 uniformly at random and replace @xmath95 by the rectangle with @xmath71 and @xmath122 as opposing vertices .
goto line ( [ init_line_reject_algo ] ) use bisection to generate an @xmath0-approximation @xmath29 of a uniform variable in .[bisecto_sub_routine_line_reject_algo ] @xmath29 [ thm_quadtree ] let @xmath5 be a riemann - integrable density on @xmath12^{d}$ ] . then the quadtree @xmath91 partitions @xmath123^{d}\times [ 0,\sup{f}]}$ ] in a collection of leaf rectangles @xmath95 for which ( [ accept_rect ] ) and ( [ reject_rect ] ) hold .
thus , algorithm [ algo_reject_compact ] halts with probability one and delivers an @xmath0-approxiation @xmath29 of @xmath2 , a random variable with density @xmath5 .
let @xmath124 be the number of iterations of the algorithm before halting .
in other words , @xmath124 is the depth of the leaf rectangle @xmath95 reached by randomly walking down the quadtree .
that walk costs @xmath125 random bits .
we show that @xmath126 and thus , ( [ accept_rect ] ) and ( [ reject_rect ] ) must hold . in the partition of @xmath89 into @xmath127 level-@xmath128 rectangles ( each of lebesgue measure @xmath129[0pt]{\small$\textstyle \frac{1}{2^{k}}$}}{\raisebox{0.04ex}[0pt][0pt]{\small$\textstyle \frac{1}{2^{k}}$}}\cdots{\raisebox{0.04ex}[0pt][0pt]{\small$\textstyle \frac{1}{2^{k}}$}}{\raisebox{0.04ex}[0pt][0pt]{\small$\textstyle \frac{c}{2^{k}}$}}$ ] ) , there are @xmath130 cells @xmath95those for which we can not decide that are `` visited '' by @xmath5 , i.e. for which @xmath131 and @xmath132 then @xmath133 for every rectangle @xmath95 , let us define its projection , @xmath134 , onto @xmath1 , i.e. @xmath135 then , for a fixed dimension , we can group the @xmath136 cells @xmath134 with the same projection and verify that of these @xmath136 cells , at most @xmath137 are visited by @xmath5 .
since there are @xmath138 rectangles @xmath134 , let @xmath139 be the collection of all projections @xmath134 , and then @xmath140 where we use the riemann approximations @xmath141 and @xmath142 of the integral of @xmath5 : @xmath143 therefore , @xmath144 since @xmath5 is riemann - integrable , this tends to @xmath145 as @xmath146 .
we call @xmath5 a monotone density on @xmath12^{d}$ ] if it decreases along at least one of the dimensions , i.e. , there exists @xmath147 such that for all vectors , , , then . as
before , let @xmath130 be the number of cells at level @xmath128 that are visited by @xmath5 . then ,
@xmath148 because the domain of @xmath5 is divided into @xmath138 cells and the @xmath136 cells along the @xmath19 dimension give rise to a walk .
this walk along the @xmath19 is at most of length @xmath149 as illustrated on figure [ fig_mono_curve ] .
we have @xmath150 and thus , @xmath151 in other words , for monotone densities , the inner loop of the algorithm has a uniform performance guarantee . for a complete analysis of algorithm @xmath152
, we need to consider the total number @xmath153 of trials before deciding .
the number of uses of the oracle is given by @xmath154 where @xmath155 is the number of iterations in the @xmath55-th trial
these @xmath155 s are i.i.d .
the number of random bits used is @xmath156 during the first phase ( the decision phase ) , and is bounded by @xmath157 in the second phase ( the bisection phase ) . since @xmath158^{d}}{f(x)}$ ] , we see that the expected number of uses of the oracle is @xmath159 and that the expected number of random bits required is bounded from above by @xmath160 as noted earlier , for any coordinate - wise monotone density on @xmath12^{d}$ ] , @xmath161 however , for general riemann - integrable densities we can not insure that @xmath162 converges .
we will address that point below .
[ thm_compact_case ] let @xmath5 be a riemann - integrable density on @xmath12^{d}$ ] , with . 1 .
if @xmath5 is monotone in at least one coordinate , then the expected number of uses of the oracle is not more than @xmath163 and the expected number of bits needed to generate an @xmath0-approximation @xmath29 is not more than @xmath164 2 .
if @xmath165 and @xmath166 are the riemann approximation of @xmath167 for regular grids of size @xmath138 i.e. , each coordinate is split into @xmath136 equal intervals , then the expected number of uses of the oracle is not more than @xmath168 where @xmath169 and the expected number of random bits used by algorithm [ algo_reject_compact ] is not more than @xmath170 just recall the estimates of @xmath162 obtained above and recall the upper bound @xmath171 in terms of @xmath172 .
part ( 2 ) of the theorem [ thm_compact_case ] is only useful if @xmath173 . for most densities , @xmath173 , as can be seen from this simple sufficient condition .
let the modulus of continuity be @xmath174 we have @xmath173 if @xmath175 because @xmath176 it suffices that as @xmath177 , @xmath178 or @xmath179 for some @xmath180 .
the number of bits used by our algorithm behaves as @xmath181 as @xmath182 , and thus matches the lower bound mentioned earlier .
in general , we use von neumann s rejection method when we know a density @xmath183 for which random variate generation is `` easy '' , and can verify that @xmath184 where @xmath9 is a known constant : generate @xmath2 with density @xmath183 generate @xmath11 uniformly on @xmath12 $ ] @xmath2 the expected number of iterations of algorithm [ gen_algo ] is @xmath9 .
we offer a generalization of algorithm [ gen_algo ] for this situation under a certain number of assumptions .
assume for now that @xmath185 , and that we can compute both @xmath35 and @xmath38 , where @xmath35 is the c.d.f . for @xmath183 .
assume furthermore that we have an oracle for @xmath186 for all intervals @xmath95 of @xmath187 .
one may be able to work things out with oracles for @xmath188 , @xmath189 , @xmath190 , and @xmath191 as well but we opt to take the more convenient approach .
define @xmath192 , which is known thanks to our oracle .
the goal is to decompose @xmath193 as before , into regions for which random variate generation is `` easy '' . for a bounded density on @xmath12
$ ] , we are content with the rectangles .
this can be mimicked by transforming the @xmath194-axis with @xmath195 since @xmath35 is monotone and continuous . using this transformation
, we note that if @xmath2 has density @xmath183 , then @xmath196 is uniform on @xmath12 $ ] .
furthermore , note that if @xmath197 , then @xmath198 where @xmath199 is a density on @xmath12 $ ] on which we can use our @xmath200-@xmath201 oracle .
since @xmath202 , we can use the quadtree method of algorithm [ algo_reject_compact ] to select a rectangle @xmath203 with probability @xmath204 in the decomposition @xmath205\big\}=\bigcup_{i\in\mathbb{n}}{\big\{r_{i}:\phantom{1}\text{$r_{i}$ is an accepting rectangle}\big\}}.\ ] ] this decomposition is valid , and the procedure halts with probability one , if @xmath199 is riemann - integrable . put differently , it works if @xmath206 is riemann - integrable .
the expected number of bits required in the decision phase of the algorithm ( selecting a leaf of the quadtree ) is bounded as in theorem [ thm_compact_case ] , when applied to @xmath199 .
it is bounded by @xmath207 where @xmath165 and @xmath166 are the riemann approximation of @xmath208 for a grid of @xmath136 equal intervals that partition @xmath12 $ ] .
we only need to analyze the second phase the method to generate an @xmath0-approximation once a rectangle @xmath95 has been selected , i.e. , the method that starts by accepting rectangle as illustrated by figure [ fig_bisect ] and outputs @xmath29 such that @xmath209 where @xmath2 , @xmath29 are coupled and @xmath210 is uniform in @xmath95 .
this can be achieved by bisection , noting that @xmath211 has distribution function @xmath35 restricted to @xmath212 $ ] .
note that with @xmath213 and @xmath214 , we have @xmath215 also if @xmath216 is such that @xmath217 , then the corresponding @xmath218 point is such that @xmath219 and similarly for @xmath220 .
the inversion algorithm [ algo : inv_bisecto ] of devroye and gravel @xcite , which is extension of a similar method for the discrete case first proposed by han and hoshi @xcite is summarized as follows : @xmath221 , @xmath222 such that @xmath223 @xmath224 @xmath225 @xmath226 @xmath29 @xmath227 @xmath228 @xmath229 @xmath225 @xmath230 @xmath224 this algorithm picks a uniform subinterval and , if permitted to run forever , would produce a random variable with distribution function @xmath35 restricted to as is illustrated in figure [ fig_bisect ] .
so , for random variate generation , we only replace line ( [ bisecto_sub_routine_line_reject_algo ] ) of algorithm [ algo_reject_compact ] by algorithm [ algo : inv_bisecto ] where @xmath231=r^{*}$ ] , and note that elsewhere in algorithm [ algo_reject_compact ] , @xmath5 must be replaced by @xmath199 .
[ thm_gen_case_i ] the expected number of bits used by algorithm [ algo : inv_bisecto ] is not more than @xmath232 where @xmath233 , and @xmath234 . in particular
, if that sum is finite for @xmath235 and if @xmath236 , then , as @xmath182 , the expected number of bits does not exceed @xmath237 theorem [ thm_gen_case_i ] establishes that algorithm [ algo : inv_bisecto ] is optimal to within an additive constant .
in particular , its main term , @xmath238 , and second term , the differential entropy @xmath239 , match the lower bound . the expected number of bits required in the decision phase of the algorithm , @xmath240 , is finite under smoothness conditions on @xmath199 .
it depends also on @xmath9 , but clearly not on @xmath0 .
let us denote an accepting rectangle @xmath203 and its projection by @xmath241 .
so , if @xmath242 $ ] , then @xmath243\times [ \alpha_i,\alpha_i+cq_i]$ ] , where @xmath244 , @xmath245 $ ] .
the probability mass of @xmath203 is @xmath246 . by the mapping @xmath38 ,
@xmath203 gets mapped to a contiguous region @xmath247 , of projection , with @xmath248 and thus , @xmath249 here we use the notation @xmath250)=g(b_i)-g(a_i)$ ] .
we also note that for all @xmath194 , @xmath251 define a regular @xmath48-grid on @xmath187 with intervals @xmath252 for all @xmath253 .
if we exit with rectangle @xmath203 , then the bisection phase of the algorithm takes an expected number of bits bounded by @xmath254 where @xmath255 is a probability vector in @xmath256 .
this result is due to the observation that the bisection method is equivalent to the algorithm analyzed by devroye and gravel @xcite and gravel @xcite in the context of the discrete distribution algorithm by han and hoshi @xcite .
thus , the expected number of bits , averaged over all @xmath203 , is not more than @xmath257 by the concavity of @xmath258 in @xmath63 , we have by jensen s inequality that ( [ ikuiku_no1 ] ) is not more than @xmath259 but @xmath260 where @xmath261 is the distribution function of @xmath5 , and @xmath262 .
thus the expected number of bits in the bisection phase does not exceed @xmath263 in the last term , we recognize the entropy defined by the probability vector @xmath264 . a theorem due to csiszr @xcite established that if @xmath264 has a finite entropy for some @xmath24 , and @xmath265 , then as @xmath182 , @xmath266 the `` 5 '' can be replaced by `` 3 '' if in addition @xmath5 is bounded and decreasing on its support , @xmath267 ( see gravel and devroye @xcite ) .
the extension of our results to dimensions greater than one for densities with unbounded support should pose no big problems . with the oracles introduced in our modification of von neumann s method
, we believe that it is impossible to design a rejection algorithm for densities that are not riemann - integrable , so the question of the design of a universally valid rejection algorithm under the random bit model remains open .
the authors would like to thank all three referees for their feedback . they are invited to luc s house for drinks if they wish to come out of the closet .
luc devroye s research was supported by an nserc discovery grant .
claude gravel thanks gilles brassard
. 10 , _ elements of information theory _ , wiley , new - york , 1991 . , _ some remarks on the dimension and entropy of random variables _ , acta mathematica academiae scientiarum hungarica , 12 ( 1961 ) , pp
. 399408 . , _ non - uniform random variate generation _ , springer , new york , 1986 . , _ sampling with arbitrary precision _
http://arxiv.org/abs/1502.02539 . ,
_ chantillonnage de distributions non uniformes en prcision arbitraire et protocoles dchantillonnage distribu des distributions discrtes quantiques _ , phd thesis , universit de montral , 2015
. https://papyrus.bib.umontreal.ca/xmlui/handle/1866/12337 . ,
_ exact classical simulation of the quantum - mechanical ghz distribution _ , ieee transactions on information theory , 62 ( 2016 ) , pp
. 876890 . , _ interval algorithm for random number generation _
, ieee transactions on information theory , 43 ( 1997 ) , pp . 599611 . , _ sampling exactly from the normal distribution _ ,
acm trans .
, 42 ( 2016 ) , pp .
114 . ,
_ the complexity of nonuniform random number generation _ , in algorithms and complexity : new directions and recent results .
, j. f. traub , ed .
, new york , 1976 , carnegie - mellon university , computer science department , academic press , pp .
reprinted in knuth s _ selected papers on analysis of algorithms _
( csli , 2000 ) . , _ various techniques used in connection with random digits .
onte arlo ethods _ , national bureau of standards , 12 ( 1951 ) , pp
. 3638 .
, _ mass transportation problems : volume 1 : theory _ , springer ( probability and its applications ) , 1998 . , _ on the dimension and entropy of probability distributions _ , acta mathematica academiae scientiarum hungarica , 10 ( 1959 ) , pp
. 193215 . ,
_ foundations of multidimensional and metric data structures _ , morgan kaufmann , elsevier / morgan kaufmann , san mateo , 2006 .
in this section , we contruct a family of densities that are not riemann - integrable for which the oracle is useless .
let @xmath268 be a parameter , and let @xmath269 $ ] be a cantor - like set constructed below .
setting @xmath270 where @xmath271 is the lebesgue measure , we have ( see below ) @xmath272 .
the lebesgue measure of the set of discontinuities is @xmath273 and is zero only if @xmath274 .
the case of @xmath274 corresponds to the usual uniform density on @xmath12 $ ] that is riemann - integrable .
all cases of @xmath275 are lebesgue integrable but not riemann - integrable because the set of discontinuities is non - zero and yet possess a cumulative distribution function .
for every @xmath24 and every @xmath276 we have @xmath277}{f(x)}=0,\\ \text{and}&\\ & \sup_{[x-\epsilon , x+\epsilon]}{f(x)}=\frac{1 - 2\delta}{1 - 3\delta}.\end{aligned}\ ] ] since these boundaries are invariant under changes of @xmath0 , algorithm [ algo_reject_compact ] , when used for rejection , say , from a uniform density , has an infinite loop with positive probability .
it is , therefore , essential that riemann - integrable densities are considered as that the supremum and infimum returned by the oracle over given small intervals converge to each other as intervals shrink . for the construction of @xmath278 , we recursively remove middle _
open _ subintervals of geometrically decreasing sizes .
let @xmath279 $ ] for @xmath280 and @xmath281 be the @xmath282 _ closed _ subintervals that are _ left _ once the middle parts are removed from the previous subintervals @xmath283 with @xmath284 corresponding to the leftmost subinterval and so on .
initially , @xmath285 $ ] . for all @xmath280 and @xmath286 , the length of a removed middle part is @xmath287 . for all @xmath280 , the total length _ not _ removed at the @xmath256-th step
is @xmath288 because the subintervals are of the same length .
let @xmath278 be the limiting subset of @xmath12 $ ] that is left , i.e. , @xmath289 we compute @xmath290 as follows : @xmath291 and therefore @xmath292 so that @xmath293
a trivial , but incorrect , modification to algorithm [ orig_vnabc ] would be : by bisection , generate @xmath29 on @xmath12^{d}$ ] such that @xmath29 is an @xmath0-approximation of @xmath2 .
generate @xmath11 uniformly on @xmath12 $ ] .
decide `` @xmath294 '' by using two bits in expected value.[doomed_algo_line1 ] if the condition from the previous line is satisfied , then return @xmath29 .
this attempt leads to failure .
let @xmath152 be the support of @xmath29 , which is necessarily contained in a fixed countable subset of @xmath12^{d}$ ] . then given any riemann - integrable density @xmath183 ,
take a finite subset @xmath295 of @xmath152 , and modify @xmath183 on @xmath295 by setting @xmath296 we have that @xmath5 is still a riemann - integrable density bounded by @xmath9 ( and its set of discontinuities is of measure @xmath145 because @xmath295 is countable ) but since @xmath297 if @xmath298 , we accept all @xmath298 , regardless of the density @xmath183 we started with .
since @xmath299 , we make an error with positive probability . | in 1952 , von neumann introduced the rejection method for random variate generation .
we revisit this algorithm when we have a source of perfect bits at our disposal . in this random bit model ,
there are universal lower bounds for generating a random variate with a given density to within an accuracy @xmath0 derived by knuth and yao , and refined by the authors . in general
, von neumann s method fails in this model .
we propose a modification that insures proper behavior for all riemann - integrable densities on compact sets , and show that the expected number of random bits needed behaves optimally with respect to universal lower bounds . in particular , we introduce the notion of an oracle that evaluates the supremum and infimum of a function on any rectangle of @xmath1 , and develop a quadtree - style extension of the classical rejection method .
* keywords : * random number generation , random bit model , von neumann sampling algorithm , tree - based algorithms , random sampling , entropy * ams subject classifications : * 65c10 random number generation , 68q25 analysis of algorithms and problem complexity , 68q30 algorithmic information theory , 68q87 probability in computer science ( algorithm analysis , random structures , phase transitions , etc . ) , 68w20 randomized algorithms , 68w40 analysis of algorithms |
classification of knots , here and below we consider knots , links and braids in @xmath0 , is one of the basic problems of the theory of knots , starting with works of tait @xcite . to classify knots means somehow or other to order a set of all knots , for example , to divide a set of all knots on equivalence classes to any reference indication , property of knots and to build all knots ( or to specify really working algorithm of build - up of all knots ) for the given equivalence class .
it seems natural as such reference indication to take the least number of crosses of a knot ( at an identification of a knot with its diagram a projection to some plane ) . by virtue of alexander s
known theorem ( see , for example , @xcite ) any knot can be presented as closure of some braid .
braids are organized much easier knots .
therefore , it is natural to use braids for classification of knots . before describing of proposed algorithm of construction of knots ( links ) with the given number of crosses on the diagram of a knot ( link )
, we shall make some estimates .
in computer program , implementing proposed algorithm , it is supposed to use partially `` br9z.p '' program in the language pascal , written by short and morton in 1985 .
`` br9z.p '' program is taken from the site of the liverpool mathematical group . in this program , authors used connection of group of knots @xmath1 with permutation group @xmath2 .
the permutation group of a set with @xmath3 elements consists of @xmath4 elements .
therefore , in `` br9z.p '' program for braids coding the two - dimensional array @xmath5 by a size @xmath6 of numbers from 0 up to @xmath3 was used . as
already @xmath7 is very major number , short and morton have been forced to be restricted to a maximum number of threads of braids @xmath8 ( in `` jones12.p '' program , written in 1994 , morton increased number of threads up to 12 by usage of a special representation of the braid group @xmath1 on @xmath9 dimensional space ) . using results of vogel @xcite , we shall prove , that in the problem of classification of knots ( links ) it is possible to substitute @xmath10 for @xmath4 . for comparison , we may mark , that @xmath11 is much less @xmath7 .
there is a representation of a knot ( link ) as closure of braid with @xmath3 threads and length of this braid ( i.e. an amount of generators in braid representation as product of standard generators @xmath12 of the group @xmath1 ) does not exceed @xmath10 , where @xmath3 number of crosses for the given diagram of a knot ( link ) .
the diagram of ( oriented ) knot , link as a result of the _ operation of erasure of crosses _ disintegrates on ( oriented ) closed curves , named _
seifert circles , _ - see fig . [ operation ] ( fig .
[ operation ] is taken from @xcite ) .
each cross can break up no more than to 2 seifert circles .
therefore , after an erasure of all diagram crosses we shall have @xmath13 seifert circles , where @xmath3 number of crosses .
this estimate is strict in the sense , that there are links , for which @xmath14 : for example , @xmath3-component link , consisting of @xmath3 eight curves , represented on the top or on the bottom of fig . [ link ] , after an erasure of all crosses , obviously , disintegrates on @xmath15 seifert circles .
-component link . ] in @xcite , vogel s article is cited from the book @xcite , vogel proposed the algorithm of representation of link as closure of braid , and proved the estimate : obtained braid has @xmath3 threads and its length ( i.e. an amount of generators in braid representation as product of standard generators @xmath12 of the group @xmath1 ) does not exceed @xmath16 , where @xmath3 number of crosses on the diagram of knot ( link ) .
assuming in vogel s estimate @xmath13 , we can assert , that there is a representation of a knot ( link ) as closure of braid with @xmath3 threads and braid length does not exceed @xmath17 , where @xmath3 number of crosses of the given diagram of knot ( link ) . incidentally , it is possible to mark , that proposed by vogel algorithm of representation of a link as closure of braid , from our point of view , is difficult for realization in practice for two reasons : \1 ) a starting point of algorithm is coding of knot ( link ) very intricate problem , actually substituting classification of knots ; \2 ) the _ operation of replacement of infinity _ ( see , for example , @xcite ) , used in the algorithm , is difficult for programming .
if the problem is restricted to classification only knots , then coding of representing braids becomes greatly simpler , as it is possible to be restricted to braids , which correspond to cyclical permutation of the order @xmath3 see above .
the algorithm of construction of all knots ( links ) with the given number @xmath3 of crosses on the diagram of a knot ( link ) can be formulated as follows .
\1 . in the braids group @xmath1 , @xmath18 with standard generators
@xmath12 we select words with @xmath10 characters .
if the problem is restricted to classification only knots , we select words with @xmath10 characters , which corresponds cyclical permutation of the order @xmath3 .
\2 . using duhornoy algorithm of reduction @xcite
, we reject superfluous braids
we select only one representative from each equivalence class of braids ( it is necessary to pay attention , that the algorithm of reduction can increase a word length ) .
we use markov relations ( see , for example , @xcite ) for final selection of braids , giving at closure different knots ( links ) . item 2 , apparently , without major detriment , can be omitted . leaving in items 2 - 3 braids with minimum word length ( minimum amount of crosses ) , as a result of closure of selected braids we shall receive diagrams of knots ( links ) with minimum number of crosses . as a result of work of suggested algorithm we shall build all knots and links , which diagrams have @xmath19 crosses . therefore ,
if we are interested only in knots ( links ) with the least number of crosses on the diagram @xmath20 , we should use algorithm twice : first to find a set of knots ( links ) @xmath21 with the least number of crosses on the diagram @xmath22 , then to find a set of knots ( links ) @xmath23 with the least number of crosses on the diagram @xmath19 , then the set of knots ( links ) @xmath24 with the least number of crosses on the diagram @xmath20 is equal to set difference @xmath25 .
in conclusion , it would be desirable to mark , that at preparation this article the report on third khariton school readings @xcite has been used . with necessary for understanding of problems discussed above theoretical questions on the theory of knots it is possible to familiarize , for example , in chapters 1 - 3 of the book @xcite . | algorithm of construction of all knots , links with given number of crosses on diagram of knot , link is offered .
this algorithm is based on simple proposition , that there is a representation of knot ( link ) as closure of braid with n threads and length of this braid does not exceed n(4n-5)+2 . |
there are currently more than fiffty extrasolar planets with published secondary eclipse measurements ( e.g. , knutson et al . 2010 ) , with many more observations taken but not yet published .
these data come from both space and ground - based facilities , and span wavelengths ranging from the visible to the mid - infrared .
the combination of data from multiple telescopes spanning a broad range of wavelengths offers an invaluable tool for constraining the pressure - temperature profiles and compositions of exoplanetary atmospheres . with an increasingly larger sample of spectra , we can begin to understand the underlying physical and chemical processes that control the atmospheric abundances through comparative exoplanetology .
there are several interesting questions that we might address through a comparative study of exoplanet atmospheres .
one is the frequency of planets with super - solar c to o ratios .
the c to o ratio can potentially provide constraints on the region of the disk where the planet formed ( berg , murray - clay , & bergin 2011 ; madhusudhan et al .
these studies propose that planets that accrete their gas envelopes outside of the water snow line will have modestly super - stellar c to o ratios , which increase to even higher values for planets that form beyond the co@xmath0 ice line ( berg , murray - clay , & bergin 2011 ) . a uniform analysis of hot jupiter atmospheres
can also be used to investigate the origin of the temperature inversions detected in a subset of these planets .
it has been suggested ( hubeny et al .
2003 ; fortney et al . 2008 ) that gas - phase tio and vo , which are effective absorbers at optical wavelengths , could lead to the formation of temperature inversions in these atmospheres .
tio and vo thermochemically exist in the gas phase in hotter planets ( @xmath2 k ) .
therefore , one would expect thermal inversions to be limited to planets that are hot enough to have gas phase tio in their upper atmospheres .
however , in a more recent study spiegel et al .
( 2009 ) suggested that vigorous vertical mixing is required to keep both gas phase and condensed phase tio and vo aloft in upper , inversion forming regions of the atmosphere .
showman et al .
( 2009 ) and parmentier , showman & lian ( 2013 ) demonstrated that tio could be severely depleted due to cold traps in the deep atmosphere and on the planet s night side .
furthermore , should tio and vo persist despite the aforementioned reasons , knutson et al .
( 2010 ) speculated that high amounts of uv flux from active stars might dissociate tio and vo .
recently madhusudhan ( 2012 ) proposed that the abundances of tio and vo could be significantly depleted if the atmosphere has a super - solar c to o ratio . in this paper
we use a uniform retrieval analysis to determine if there is a correlation between the presence or absence of a temperature inversion , the c to o ratio of the planet s atmosphere , and the activity level of its host star . in paper i ( line et al .
2013 ) we developed a suite of inverse modeling algorithms , called chimera , to determine the ranges of temperatures and compositions that were consistent with a given data set .
chimera uses three bayesian retrieval approaches including optimal estimation , bootstrap monte carlo , and differential evolution markov chain monte carlo to determine the allowable ranges of temperatures and abundances for a given planet . in this investigation
we apply chimera to nine planets in order to undertake the first uniform ( meaning we use the same model , model assumptions , and retrieval approaches for each object ) retrieval analysis of a set of exoplanet spectra .
our uniform analysis allows us to make robust comparisons between planets , as we utilize the same model parameters and retrieval techniques for each target . comparing results derived from different retrieval approaches or different models
can be complicated due to the differing model assumptions .
such an analysis can also provide a useful set of statistical atmospheric properties from which detailed physical models such as general circulation models , photochemical models , and radiative equilibrium models can be compared .
our goal in this study is to provide a catalogue of temperatures and abundances for nine well - observed planets and to address some of the outstanding questions regarding c to o ratios and the possible causes of thermal inversions .
our sample includes the following planets : hd189733b , hd209458b , gj436b , hd149026b , wasp-12b , wasp-19b , wasp-43b , tres-2b , and tres-3b .
these planets were selected because they span a wide range of physical parameters with a reasonably high number of spectral data points per planet . in [ sec : methods ] we will describe the retrieval approaches and forward model assumptions .
the details of our approach can be found in paper i. in [ sec : analysis ] we present our temperature and abundance retrieval results for each of the nine planets and compare them to previously published analyses .
we then use the derived abundances to asses the allowed range of c to o ratios and comment on the implications of these results for current hypotheses for the origin of temperature inversions on these planets . finally , in
[ sec : conclusions ] we will discuss the big picture view from our retrieval results , which show little evidence for c to o ratios larger than one .
we summarize our retrieval methods and forward models here and refer the reader to paper i for more detailed descriptions .
the goal of a retrieval , given some forward model , is to characterize the posterior probability distribution of the parameters of interest , in this case , temperatures and abundances via a radiative transfer model .
this posterior is determined from a combination of prior information and the data .
paper i describes three approaches that are commonly used to characterize posterior distributions .
these include : optimal estimation ( oe , e.g. , rodgers 2000 ) , bootstrap monte carlo ( bmc , press 1992 ; ford 2005 ) , and differential evolution markov chain monte carlo ( demc , ter braak 2006 ) .
optimal estimation minimizes a quadratic cost function using the levenberg - marquardt scheme and approximates the posterior as multivariate normal .
bootstrap monte carlo uses a data resampling method based on a best - fit model to derive the parameter distributions .
differential evolution markov chain monte carlo is a type of markov chain monte carlo approach ( mcmc ) which uses a genetic algorithm to efficiently explore highly correlated parameter spaces .
demc and oe both evaluate the following log - likliehood function : @xmath3 where @xmath4 is the set of @xmath5 data points , @xmath6 is the @xmath7-dimensional parameter vector , @xmath8 is the forward model , and @xmath9 is the @xmath10 data error matrix .
@xmath11 is the _ a priori _ state vector and @xmath12 is the @xmath13 _ a priori _ covariance matrix .
the first term in equation [ eq : cost_func ] is simply the standard chi - squared " and the second term represents the prior knowledge of the parameter distribution before we make the observations .
in paper i we determined that the three approaches agree for high signal - to - noise , high - resolution data , but diverge for the data sets currently available for typical hot jupiters .
we also found that oe and demc tended to agree better than the bmc , and that our implementation of bmc was only able to characterize the probability distribution in the region very close to the nominal solution .
therefore , in this investigation we utilize the oe and demc approaches to estimate the posterior probability distributions of the temperatures and abundances .
we use the results from oe to initialize the demc as described in paper i. by using two distinct approaches , we can determine whether or not our results are sensitive to our choice of fitting method .
we present results from the demc approach in our final abundance catalogue , as this is the more widely accepted , robust approach for current exoplanet data ( e.g , benneke & seager 2012 ; line et al .
2013 ) .
the physical parameters we are most interested in are the temperature structure and the mole fractions of various gases .
we therefore choose to retrieve vertically uniform mixing ratios of h@xmath0o , ch@xmath1 , co , and co@xmath0 . these are generally the most thermochemically abundant ( with the exception of co@xmath0 ) and infrared active species over the observational bandpasses for a variety of metallicities and c / o ratios .
furthermore these species tend to have the most complete line lists either via experimentation or _
ab initio _ modeling ( see part i for description of our line lists ) .
we have assumed vertically uniform because the data do not provide useful constraints on the vertical abundance profile ( lee et al .
also , the species of interest in this case , when abundant , thermochemically have near vertically uniform profiles to begin with , and vertical mixing tends to quench minor species resulting in vertically uniform profiles ( moses et al .
2011 ; line et al .
we note that we have not included other potentially important absorbers such as nh@xmath14 cross - section data base .
we do not expect the lack of nh@xmath14 to have a significant impact on our retrievals because strong absorption is mainly limited to the 10 @xmath15 m region and hence should not introduce strong correlations with the other absorbers .
] , hcn , h@xmath0s , c@xmath0h@xmath0 , as accurate line - lists at high temperatures do not yet exist ( to the best of our knowledge ) .
we plan to expand the scope of our future retrievals as more reliable sepctroscopic databases become available , and as improvements in the observations justify the inclusion of additional gases .
we also assume that these planets are hydrogen dominated and hence fix the h@xmath0 and he mixing ratios to thermochemically appropriate abundances of 0.85 and 0.15 , respectively , which hold over a wide range of temperatures and metallicities .
there could also be other optically inactive species such as n@xmath0 , o@xmath0 , noble gases other than he , etc . , but these molecules would have a minimal impact on the shape of our retrieved emission spectra and we therefore do not include them in our fits .
the retrievable species are assumed to have mole fractions much much less than that of h@xmath0 , therefore we do not include mean molecular weight effects on the spectrum .
we neglect to consider clouds in our models for the same reasons discussed madhusudhan & seager ( 2009 ) .
furthermore , fortney et al .
( 2005 ) suggested that many cloud species ( minor condensates and photochemical hazes ) will have small normal optical depths resulting in minimal impact on emission spectra .
this idea is supported by recent findings that , though hd189733b ( pont et al . ( ) ) and gj436b ( knutson et al .
( 2014 ) ) show strong evidence for high altitude absorbers in transmission , these planets still show strong signs of absorption in the emission spectra
. however , given the mounting evidence for clouds in primary eclipse , it is perhaps worth investigating the impacts of clouds on emission spectra in the future . in paper
i we described two temperature retrieval approaches : the level - by - level approach and the parametrized approach . for this analysis
we choose to use the parametrized temperature profile approach .
the limited number of data points per planet limits the practicality of a full level - by - level retrieval .
the parameterization we use is based on analytic radiative equilibrium profiles ( see guillot 2010 ; heng et al .
2012 ; robinson & catling 2012 ) controlled by 5 free parameters .
these parameters are the infrared opacity , two visible opacities , partitioning of the two opacity streams , and a catch - all factor for the albedo , emissivity , and redistribution .
this last parameter effectively accounts for energy balance at the top of the atmosphere .
we do not try to self - consistently relate the temperatures back to the composition for reasons discussed in paper i. once we have determine the uncertainty distributions for each of the five temperature profile parameters , we can then reconstruct the ensemble of temperature profiles .
our priors are the same as those described in paper i. when using the optimal estimation formalism , by construction , the priors are gaussian .
we choose very broad ( 12 orders - of - magnitude spanning the 68% confidence interval ) gas priors .
this mitigates the effect of the prior on the gas abundance retrievals .
we also choose broad gaussian priors on the 5 parameters governing the parametrized temperature profile . when using demc we choose flat gas priors , but use gaussian priors on the temperature parameters .
these produce a spread in the reconstructed temperature profiles that is more consistent with numerical radiative equilibrium models than a flat prior would . for any mcmc search when using flat priors one must impose limits in order to prevent the random walk process from venturing too far from useful phase space .
we choose a lower limit mixing ratio of 10@xmath16 and an upper limit of 0.1 .
although somewhat arbitrary , we would expect the molecular abundances of these four species to fall well within these limits .
we obtain the planet and system parameters ( stellar radius , planet radius , stellar temperature , semimajor axis , planet gravity ) from the published literature ( see table [ tab : table0 ] ) .
these parameters are used in generating the parameterized temperature profile and when dividing by the stellar grid models .
we use interpolated ( logg , fe / h , and teff ) phoenix stellar grid models ( allard et al . 2000 ) to compute the contrast spectrum .
we also should note that the retrieval results presented here must be taken in the context of our particular model . though not likely to change significantly , the retrieval abundances for the molecules presented here may change if we included a larger or different set of molecules .
it is worth investigating the impact of different molecules on our retrieval results through a nested model comparison in a future investigation , similar to the one presented in swain et al .
in this section we provide a catalogue of abundances and temperatures for nine planets observed in secondary eclipse .
the sources of the secondary eclipse data are shown in table [ tab : table1 ] .
we provide detailed descriptions of the results for each planet in our sample and compare these results to those of previous investigations .
graphical results of the retrievals are summarized in figures [ fig : figure1]-[fig : figure3 ] .
figure [ fig : figure1 ] shows the secondary eclipse observations , the best fit spectra , and a statistical summary of all of the fits from the demc retrieval .
these fits are summarized with 68@xmath17 and 95@xmath17 confidence bounds along with a median spectrum .
figure [ fig : figure2 ] shows the temperature profiles summarized with 68@xmath17 and 95@xmath17 confidence bounds , a median profile , and a best fit .
we also show the averaged thermal emission contribution function to get a sense for which pressure levels the observations probe . for comparative curiosity
we show , in some cases , self - consistent first - principle temperature profiles from the literature .
finally , figure [ fig : figure3 ] shows the marginalized gas abundance posteriors for each planet along with the imposed priors ( flat for demc , gaussian for oe ) .
we use the resulting retrieval results to derive the c to o ratio probability distributions in figure [ fig : figure4 ] .
these distributions generally have a double - peaked structure , which is the manifestation of the uniform abundance priors in the c to o distribution ( see paper i for a detailed discussion of this issue ) . rather than showing the two peaked c to o distribution resulting from the prior and posterior together ,
we normalize the posterior derived c / o distribution by the double - peaked prior c / o distribution to give us a sense for how the data contributed to our knowledge of the c to o ratio .
this has no formal statistical meaning , but it provides a clear visual representation of the information provided by the observations independent of the assumed priors .
we also include a table ( table [ tab : table2 ] ) comparing our numerical results to those of previous studies for easy reference .
we provide our 68@xmath17 confidence intervals for the molecular mixing ratios along with the nominal best - fit values derived from demc .
we also report the reduced cost function value from equation [ eq : cost_func ] , @xmath18 for the best fit , where n is the number of data points .
a @xmath18of one suggests that the model on average fits the data within the 1-sigma error bars ( just as in madhusudhan & seager 2009 ) .
since we place limits on the flat gas priors used in the demc retrieval , we must interpret the retrieved abundance range in the context of those limits .
given those limits , the 68@xmath17 confidence interval from the flat gas prior would result in abundances that span @xmath19 , or @xmath208 orders of magnitude .
anything smaller than this suggests that the data was informative within the context of our model .
furthermore , we quote the 68@xmath17 confidence interval for the c / o ratios derived from our abundance retrievals . these values are computed directly from the posterior c to o distributions , not the normalized - by - prior distribution shown in figure [ fig : figure4 ] .
given the imposed abundance limits , the 68@xmath17 confidence interval in the c / o distribution resulting from the uniform gas priors is @xmath21 , or @xmath20 3 orders of magnitude .
together with hd209458b , hd189733b is one of the best - studied hot jupiters to date .
this is because it orbits a bright , nearby k star with a favorable planet - star radius ratio .
secondary eclipse observations have been made with a variety of instruments including hst - nicmos ( swain et al . 2009 ) , spitzer irac ( deming et al .
2006 ; charbonneau et al .
2008 ; knutson et al . 2009 ; 2012 ; agol et al .
2012 ) , mips ( knutson et al . 2009 ; charbonneau et al .
2008 ) , and irs ( grillmair et al . 2007 ; 2008 ) spanning @xmath201.5 @xmath15 m to @xmath2025 @xmath15 m .
several previous retrieval investigations have constrained the range of allowable temperature structures and compositions for this atmosphere .
the first complete study via a systematic parametric grid search was published by madhusudhan & seager ( 2009 ) .
the composition they derived was high ( relative to a solar composition atmosphere ) in co@xmath0 and co followed by low abundances of ch@xmath1 and a moderate abundance of h@xmath0o .
lee et al .
( 2012 ) and line et al .
( 2012 ) came to similar conclusions using the optimal estimation retrieval approach .
the large abundance of co@xmath0 relative to the other species remains chemically perplexing ( line et al .
2010 ; moses et al . 2011 ; 2013a ) .
there is also no evidence for a thermal inversion .
our abundances are generally consistent with the previous results to within an order of magnitude
. however , not all of the previous investigations are consistent .
for instance , all of the results for hd189733b in table [ tab : table2 ] , including ours , require an anomalously high abundance of co@xmath0 aside from the analysis presented in swain et al .
( 2009 ) ( see shabram et al .
( 2011 ) for a more detailed discussion of this discrepancy ) . also , the upper limit on ch@xmath1 quoted by line et al .
( 2012 ) is much higher than the limits from previous studies .
this is likely due to the problems associated with using optimal estimation in data regimes in which the resulting posterior probability distributions are non - gaussian .
we note that the data used in the retrievals were not the same in all cases .
we simultaneously retrieved the abundances and temperatures using a combination of data including the nicmos , irs , and spitzer photometry .
madhusudhan & seager ( 2009 ) considered these data sets separately .
swain et al .
( 2009 ) and line et al .
( 2012 ) only considered the nicmos data set .
lee et al .
( 2012 ) considered all available data sets .
it is worth considering whether individual data sets give consistent results , as data taken at different epochs could vary as a result of stellar variability , differences in detector systematics , etc . despite these differences the results of all of these studies
are in generally good agreement , suggesting that our conclusions for this planet are reasonably robust . in our new study
we find an enhanced best fit c / o ratio of 0.85 ( 0.47 - 0.90 ) , well within the range quoted by previous investigations .
this is a fairly robust result relative to the prior ( see figure [ fig : figure4 ] ) .
the large number of spitzer irs data points that have thermal emission weighting functions near the 10 mbar level combined with our requirement of radiative equilibrium via the analytic parameterization , provide a well constrained temperature profile above the 0.1 bar level .
below this level the temperature profile begins to diverge because very little emission is able to escape from these deeper regions of the atmosphere .
our derived spread in profiles in the well - constrained regime between 10 - 100 mbar is in agreement with the profile derived in lee et al .
we also show a temperature profile from moses et al .
( 2011 ) , which is generally cooler than our derived spread .
hd209458b is also a favorable target for secondary eclipse observations due to its relatively deep eclipse depths and bright host star .
multiple groups have obtained secondary eclipse observations using broadband photometry ( cowan et al . 2007 ; knutson et al . 2008 ; crossfield et al .
2012a ) , nicmos spectroscopy
( swain et al .
2009b ) , and spitzer irs spectroscopy ( richardson et al . 2007 ; swain et al .
the consensus from these observations is that this planet possesses a strong thermal inversion ( burrows et al .
2007 ; madhusudhan & seager 2009 ; swain et al . 2009 ) and poor day - night heat redistribution ( cowan et al . 2007 ) .
madhusudhan & seager ( 2009 ) used spitzer photometry to constrain the range of allowable dayside temperature profiles and compositions .
their ensemble of fits suggest a strong thermal inversion beginning near the @xmath20500 mbar level and a high c / o ratio . however , they point out that simple 1-dimensional dayside averaged profiles may result in over - interpretation of the data . combining nicmos and irs spectroscopy with spitzer broadband photometry , swain et al .
( 2009 ) salso found that a thermal inversion is needed to explain the combined observations .
they also placed some limited constraints on the molecular abundances .
our independent analysis focuses on the broadband data only .
we are unable to simultaneously fit the nicmos / irs spectroscopy and the spitzer photometry with a single model that provides a good match to all of the observations .
our best fits including both data sets generally result in reduced @xmath22 values greater than 5 , and hence such solutions may not be statistically meaningful . our broadband retrieval confirms the existence of a strong ( median inversion depth - to - error at the temperature minimum of 7.5 , or 7.5@xmath23 ) thermal inversion with the temperature minimum occurring within the 1bar -100 mbar region .
this tropopause location is consistent with tropopause locations in our own solar system and with what is predicted by robinson & catling ( 2013 ) , but is somewhat deeper than what is shown in moses et al .
we also find c to o ratios near unity .
the inversion is needed to explain the high 4.5 and 5.7 @xmath15 m fluxes relative to the other points .
this is also the location of the strong co band .
large abundances of co ( @xmath24 ) are required to push the weighting functions at these wavelengths high up in the atmosphere where the inversion is strongest .
this high abundance of co dominates the c to o ratio forcing c / o s of unity , again consistent with the results of madhusudhan & seager ( 2009 ) .
the temperatures at pressure levels higher than the @xmath2050 mbar level exceed those of the tio condensation curve supporting the hypothesis that tio is the cause for inversions .
however , these temperatures exceed the tio condensation curve _ because _ of the inversion .
if the inversion were not present , the temperatures at these levels would be cooler than the tio condensation curve , begging the question of whether or not tio is actually the cause of the inversion or if it is some other absorber .
we suppose though , that if the planet had a higher internal heat flux when it first formed , it may have reached this state by cooling from a hotter profile where tio and vo were in gas phase everywhere future observations with the hst wfc3 may should provide additional confirmation of the existence of a temperature inversion .
the co feature near 1.6 @xmath15 m probes altitude regions near the 100 mbar level where the temperature is increasing from the minimum whereas the bluer wavelengths probe the few bars levels . hence these wavelengths probe regions of the atmosphere that bracket the temperature minimum , as determined by the broadband data .
the existence of this feature will confirm the presence of a temperature inversion as well as the high co abundance .
gj436b is a warm ( @xmath20700 - 900 k ) neptune - mass planet with a seemingly unusual atmosphere chemistry that has generated considerable attention in the modeling community .
the first set of spitzer observations ( stevenson et al . 2010 ) indicated that this planet is rich in co and depleted in ch@xmath1 . this conclusion is primarily driven by the high 3.6 to 4.5 @xmath15 m flux ratio .
this composition is at odds with its temperature given the assumption of thermochemical equilibrium and solar elemental abundances .
one would expect such a planet to be rich in methane and low in co. these authors suggested that photochemistry might be responsible for the apparent depletion of methane .
however , photochemical depletion of methane does not appear to be significant in hydrogen dominated atmospheres in this temperature range even under a wide variety of assumed vertical mixing strengths and uv fluxes ( line et al .
2011 ; miller - ricci et al . 2012 ; moses et al .
it has recently been suggested that this planet has an extremely metal - rich ( 300 - 2000@xmath25 solar ) atmosphere , which provides a simple explanation for the measured enhancement of co over ch@xmath1 without the need to invoke extreme or exotic chemistry ( moses et al .
2013b ) .
table [ tab : table2 ] shows how our results compare to those of stevenson et al .
( 2011 ) and madhusudhan & seager ( 2011 ) .
we also conclude that the atmosphere is rich in co and depleted in ch@xmath1 relative to what one would expect for a solar composition atmosphere at these temperatures .
we point out that although we quote a confidence interval , on all four gases , our fits only provide upper limits on most species with the exception of co. our lower limit for these unconstrained gases comes from our prior , which artificially places a hard limit at @xmath26 .
if we examine the histogram for methane in figure [ fig : figure3 ] we see that an appropriate upper limit might be placed at the half max location of @xmath27 , which is within an order of magnitude of the previous results .
we also find upper limits on the co@xmath0 and h@xmath0o abundances , but again no strong lower limit .
the co@xmath0 marginalized posterior shows a multimodal behavior , with a strong mode occurring at @xmath27 and a weaker mode near @xmath28 .
our estimate of the c to o ratio is similar to that quoted by madhusudhan & seager ( 2011 ) as the most probable values seem to fall between solar and unity .
figure [ fig : figure4 ] indicates that a c to o ratio greater than one is highly improbable .
our temperature profile dispersion is nearly identical to those found in madhusudhan & seager ( 2011 ) between @xmath200.1 - 1 bars and is generally hotter than the lewis et al .
( 2010 ) derived general circulation model profile . to date
there has been no detailed atmospheric retrieval analysis for this planet .
it is a saturn - mass planet with a very large core orbiting a metal - rich star ( sato et al .
2005 ) , and therefore is a good candidate for a metal - rich atmosphere .
furthermore , its likely high atmospheric metallicity and temperature make it a prime candidate for an inversion caused by gas - phase tio and vo ( fortney et al .
stevenson et al . ( 2012 ) obtained spitzer photometry and interpreted the observations using the self - consistent models from fortney et al .
( 2005 ; 2006 ; 2008 ) .
these data - model comparisons suggest that the planet s emission spectrum is well - described by a relatively high ( 30@xmath25 solar ) metallicity atmosphere with correspondingly enhanced co and co@xmath0 features , with no evidence for a temperature inversion .
our retrieval results indicate that the planet has more co and co@xmath0 than methane , consistent with the stevenson et al .
( 2012 ) results .
the marginalized gas posteriors ( figure [ fig : figure3 ] ) show an upper limit on methane , a strong peak in the probability distribution for water near @xmath29 with an unconstrained tail towards low abundances , and a preference for large ( @xmath30 ) abundances of co. these results are consistent with the planet s relatively high atmospheric temperature ( @xmath201700 k ) , which tends to favor co over ch@xmath1 at near - solar abundances .
there is a slight preference for high abundances of co@xmath0 ( relative to solar ) consistent with the stevenson et al .
( 2012 ) results .
we find a c to o ratio that is consistent with solar and can rule out ratios greater than unity .
our temperature profiles are most consistent with the solar metallicity model from stevenson et al .
( 2012 ) without an inversion .
our range of temperature profiles are also similar to those in fortney et al .
( 2006 ) over the ir photosphere .
we are in good agreement with their general conclusions that this planet lacks a strong dayside temperature inversion . the atmosphere of hot jupiter
, wasp-12b has generated some excitement as the first candidate for a planet with a super - solar c to o ratio .
madhusudhan et al .
( 2011a ) carried out a retrieval analysis and found that their fits preferred a high c / o ratio atmosphere based on seven broadband photometry points ranging from j - band through the spitzer irac 8 micron point .
this conclusion stems primarily from the apparent lack of water absorption in the near - infrared . due to the difficulty in removing water via disequilibrium processes , madhusudhan et al .
( 2011a ) argue that the most plausible explanation for the low water abundance is a high c to o ratio .
berg et al .
( 2011 ) have suggested that gas giant planets could form with different c to o ratios depending on their location in the protoplanetary disk , suggesting that c to o ratios for hot jupiters could provide constraints on their formation locations . since the publication of the madhusudhan et al .
( 2011a ) analysis there have also been some corrections to the estimated secondary eclipse depths for this planet .
bergfors et al .
( 2013 ) identified a companion star to wasp-12 responsible for contaminating the measured photometry .
crossfield et al .
( 2012 ) reanalyzed the spitzer data to account for contamination from the binary m star companion ( bechter et al .
2013 ) , which dilutes the light from the primary star and decreases the measured eclipse depths .
the same paper also presents new k band measurements along with the hst wfc3 data from swain et al .
their updated data - model comparison indicates that a nearly isothermal atmosphere without the presence of significant absorbers could explain the data .
swain et al .
( 2013 ) observed wasp-12b in both emission and transmission using the hst wfc3 .
like crossfield et al .
( 2012 ) , they also find that the data do not require a high c to o ratio , particularly if oxygen bearing species such as tio and vo are included to explain the transmission spectrum .
in our analysis we use the most current data that could be found in the literature .
this includes the subaru narrowband photometric 2.315 @xmath15 m point from crossfield et al .
( 2012b ) , an updated k@xmath31 band photometric point from zhao et al .
( 2012 ) ( larger error bar than the previous k@xmath31 band data from croll et al .
2012 ) and we account for the null " and ellipsoidal " 4.5 @xmath15 m analysis in cowan et al .
all of the data have been updated to account for the dilution correction factor discussed and implemented in crossfield et al .
( 2012 ) and swain et al .
( 2013 ) .
we applied several different analyses .
first , we retrieve the temperatures and abundances under the assumption of the cowan et al .
( 2012 ) null " hypothesis for the 4.5 @xmath15 m flux .
crossfield et al .
( 2012b ) suggests ( via private comm . from n. cowan ) that the flux resulting from this analysis is likely more reliable than the flux resulting from the ellipsoidal " analysis due to the more realistic assumptions about the planetary shape .
hence , we choose the retrieval results that derive from the set of observations that include the 4.5 @xmath15 m null " data point as our nominal result .
we find a nearly isothermal atmosphere with the potential for a weak inversion ( @xmath202.5@xmath23 as defined in [ sec : hd209 ] ) , though the small temperature gradients near the temperature minimum are enough to invoke absorption and emission features ( figure [ fig : figure1 ] ) . the self - consistent temperature profile from crossfield et al .
( 2012b ) is in very good agreement with our retrieved spread . the preferred solutions from the demc favor an atmosphere that is very abundant in co@xmath0 ( figure [ fig : figure3 ] , blue histogram ) .
this high abundance of co@xmath0 drives the c to o ratio closer to solar than to unity . if we repeat the results using the 4.5@xmath15 m ellipsoidal " derived data point , the lower flux value drives the co@xmath0 abundance to even higher levels .
the inversion depth strengthens slightly . in both cases we note that the water abundance can span a wide range
this is different than in madhusudhan et al .
( 2011 ) where in general the retrieved water abundances were lower than 10 parts - per - million , which was the primary driver in determining a high c / o .
our gas abundance retrievals do not provide strong evidence for a high c / o atmosphere , though we can not rule out such a possibility ( blue and red curves in figure [ fig : figure4 ] ) .
one might question how realistic solutions with such high abundances of co@xmath0 may be .
such high co@xmath0 abundances are generally not thermochemically permissible in highly reducing hot - jupiter - like atmospheres . in order to explore this
we have placed an upper limit on the co@xmath0 abundance at a mixing ratio of 10@xmath32 and repeated the retrievals under the null " and ellipsoidal " assumptions .
we show the resulting gas posterior probability distributions as the gray curves in figure [ fig : figure3 ] .
the upper limit prevents the demc from finding the high co@xmath0 abundance mode .
the other molecules then compensate to fit the data .
the best fit is not as good ( @xmath332 , not shown in table [ tab : table2 ] ) as if we permit the higher co@xmath0 abundances , though not by a statistically significant amount . imposing this limit has virtually no effect on the ch4 and co ( which is completely unconstrained as indicated by its nearly flat posterior in figure [ fig : figure3 ] ) abundances .
the imposed upper limit does , however , drive up the water abundance .
again , we are unable to find strong evidence for c to o ratios larger than unity ( figure [ fig : figure4 ] , gray curves ) .
the imposed upper limit on the co@xmath0 abundance has little effect on the temperature profile no strong evidence for an inversion .
the lack of a strong inversion ( @xmath202.5@xmath23 ) suggests that perhaps tio and vo are not strongly absorbing , though a nearly isothermal atmosphere suggests that perhaps there is some high altitude opacity . in a high c / o atmosphere , thermochemically , tio and vo
should be depleted as suggested by madhusudhan ( 2012 ) .
however , we do not find strong evidence for a high c / o atmosphere .
it is difficult to disentangle the reason for the discrepancy between our results and those of madhusudhan et al .
potential differences could be due to the use of different data sets ( dilution corrected fluxes vs. non corrected ) and different model assumptions such as the assumed gases , gas opacity databases , or the temperature profile parameterization .
a detailed model comparison over a wide range of parameter space would be needed to identify potential differences .
future observations and modeling of this intriguing planet are certainly needed in order to draw firm conclusions .
given its high equilibrium temperature ( 2400 k ) and the assumption of solar elemental abundances , wasp-19b is expected to have moderate quantities of tio and vo , hence causing a thermal inversion ( fortney et al .
anderson et al . (
2013 ) and madhusudhan ( 2012 ) found no evidence for an inversion , leading madhusudhan ( 2012 ) to suggest a high c to o ratio in order to deplete tio and vo .
wasp-19 is one of the most active stars to host a hot jupiter , therefore uv destruction of an absorbing molecule could also provide an alternative explanation ( e.g. knutson et al . 2010 ) .
madhusudhan ( 2012 ) explored both carbon - rich and oxygen - rich models for this planet and concluded that the observations are unable to constrain the c to o ratio , but may slightly favor a c to o ratio greater than one .
figure [ fig : figure3 ] suggests that co and methane are almost completely unconstrained , as their marginalized posteriors closely track the prior .
the data do appear to place an upper limit on water ( @xmath28 ) and a weak upper limit on co@xmath0 ( @xmath34 ) .
we concur with madhusudhan ( 2012 ) , who conclude that the current data for this planet do not provide useful constraints on the c to o ratio ( figure [ fig : figure4 ] ) .
both the oxygen rich and carbon rich derived abundances of madhusudhan ( 2012 ) are consistent with our 68@xmath17 confidence intervals as well as the temperature profiles
. future observations of wasp-19b with wfc3 would greatly reduce the uncertainties in the gas abundances and temperature profile .
figure [ fig : figure1 ] shows a large divergence in the spectra at wavelengths less than 2 @xmath15 m .
wfc3 observations with a signal to noise similar to those obtained for wasp-12b would narrow this dispersion down and ultimately reduce the uncertainties in the abundance of h@xmath0o .
a strong lower bound on the water abundance would provide improved constraints on the c to o ratio .
wasp-43b has a very favorable planet - star radius ratio and correspondingly deep secondary eclipses , making it a prime target for both ground and space - based observations .
it is currently one of the coolest ( @xmath201500 k ) planets accessible via ground based observations .
blecic et al .
( 2013 ) explored the composition of wasp-43b using the temperature and abundance retrieval approach of madhusudhan et al .
( 2011a ) .
they found that the composition is only weakly constrained , but rule out a temperature inversion based on the relative fluxes between the spitzer irac channels and the ground based j and k band photometry .
we obtain stronger constraints from our fits that allow us to confidently rule out c to o ratios larger than one ( figure [ fig : figure4 ] ) .
this is primarily due to the upper limit on methane near @xmath29 ( figure [ fig : figure3 ] ) .
we also find that the temperature profile lacks a temperature inversion , consistent with the conclusions of blecic et al .
( 2013 ) .
tres-2b is a highly irradiated hot - jupiter that is theoretically predicted to possess a temperature inversion ( a pm " class planet , fortney et al .
croll et al .
( 2010a ) compared the spectral energy distribution of photometric data points at 2.14 , 3.6 , 4.5 , 5.6 , and 7.8 @xmath15 m to forward models of fortney et al .
( 2005 ; 2006 ; 2008 ) which assume local thermal equilibrium and solar metallicity . depending on the details of the temperature profile , tio could be present in gas phase , leading to the formation of a temperature inversion in tres-2b s atmopshere .
they found that cooler temperature profiles fit the data better , naturally explaining the absence of an inversion as tio and vo should condense out of the planet s dayside atmosphere .
we find that in our fits the data provide minimal constraints on the relative molecular abundances .
the marginalized posterior distributions show very little preference for a specific combination of abundances ( i.e. , there is no strong mode , figure [ fig : figure3 ] ) .
although we quote a 68@xmath17 confidence interval for each species , we note that these confidence intervals are nearly the same as those given by the prior , suggesting relatively weak constraints from the data aside from a slight preference for higher abundances of methane . as a result of these weak constraints
, we can not make strong statements about the c to o ratio .
our retrievals rule out a thermal inversion , in agreement with the results from croll et al .
( 2010a ) .
we find that the dispersion in the temperature profiles straddles the tio condensation boundary , with temperatures between 300 mbars and 0.7 mbars dipping below the condensation temperatures .
this suggests that tio and vo would likely be lost to cold traps in these cooler regions even if there are local regions in the atmopshere that are warm enough for them to exist in gas phase ( e.g. , spiegel et al .
2009 ; showman et al . 2009 ; parmentier , showman & lian 2013 ) . like tres-2b ,
tres-3b is an interesting target for identifying thermal inversions that may be due to tio ( fortney et al .
croll et al .
( 2010b ) examined h and k band photometry combined with spitzer photometry in order to determine the temperature structure .
they found that the atmosphere could be explained with an isothermal temperature structure .
our retrievals allow us to expand on these initial conclusions , which were based on a comparison to the same class of forward models by fortney et al .
cited in the previous discussion on tres-2b .
we find that h@xmath0o is fairly well constrained with abundances near @xmath35 .
this is reasonable for a planet at these temperatures with solar composition .
methane has a well defined upper limit of @xmath36 , again consistent with solar composition .
co remains unconstrained due to the large uncertainty on the irac 4.5 @xmath15 m data point and the lack of constraints from data at other wavelengths ( figure [ fig : figure3 ] ) .
co@xmath0 has a weak upper limit near @xmath35 .
this upper limit arises from the wings of the 2.1 @xmath15 m co@xmath0 band probed by the k band photometry .
these abundances are generally consistent with a solar composition atmosphere .
we can also confidently rule out c / o ratios larger than one due to the well constrained water abundance and low upper limit to the methane abundance .
the temperature profile dispersion for tres-3b is largely cooler than those of tres-2b over the range of pressures probed by these infrared observations , and lies well below the condensation curve for tio .
the lack of inversions in tres-2 and -3b therefore appears to be consistent with the hypothesis that tio and vo could be the inversion - causing opacity sources .
secondary eclipse spectra simultaneously tell us about the dayside temperature structures and compositions of the planets examined in our study .
we have performed a comprehensive , uniform retrieval analysis of nine exoplanets observed in secondary eclipse and provide a catalogue of the resulting temperature and molecular abundance estimates .
such analyses provide a useful set of atmospheric information that can be used to test various hypotheses about the origin of the observed properties of hot jupiter atmospheres .
our results are generally consistent with those of previous investigations despite differences in model assumptions and retrieval approaches . however , unlike previous investigations
, we find no statistically compelling evidence in support of c to o ratios larger than solar as shown in figure [ fig : figure4 ] ( though hd209458b appears to have a c / o ratio near unity due to the retrieved co abundance ) .
solar c / o falls within the 68% confidence interval for eight of the nine planets in our sample .
we also find no strong evidence for thermal inversions eight of the nine planets we investigated , with the exception being hd209458b .
wasp12b has a nearly isothermal atmosphere , but there is not strong evidence for an actual inversion . out of our sample of nine planets ,
three ( hd209458b , wasp-19b , wasp-12b ) are hot enough to contain tio in the gas phase , but our retrievals do not strongly detect inversions in two of their atmospheres .
madhusudhan ( 2012 ) have suggested that high c to o ratios could reduce the gas phase tio and vo abundances thus eliminating the absorbers that cause inversions .
however , we do not find strong evidence in support of high c to o ratios for the two planets hot enough to posses gas phase tio and vo , although we can not rule out the possibility high c to o ratios in wasp-19b and wasp-12b .
we do however find an inversion in the atmosphere of hd209458b while also having a high c to o ratio contradictory to the idea proposed by madhusudhan ( 2012 ) , if tio and vo are indeed the responsible absorbers .
this would seem to provide another line of evidence against tio and vo as the absorber responsible for the inversions , although it does not rule out the more general hypothesis that variations in elemental abundances are responsible for the lack of temperature inversions in a subset of hot jupiters . in the future ,
spectroscopic observations of these planets with the hst wfc3 instrument offer a promising avenue for further constraining their c to o ratios and refining our knowledge of their atmospheres . as can be seen in figure [ fig : figure1 ] , hd209458b , wasp-19b , wasp-43b , tres-2b , and tres-3b all
have widely divergent spectra between 1 and 2 @xmath15 m . comparing these spectral spreads to that for wasp-12b
, we find that the wfc3 data reduces the overall uncertainty in the spectra between 1.1 and 1.8 @xmath15 m . looking at figure [ fig : figure2 ] , we can see that the wfc3 data can constrain temperatures in the deep atmosphere to a higher precision than those planets without the wfc3 data . if comparable signal - to - noise could be obtained for the other planets using wfc3 , we could expect similar retrieved temperature precisions in the 1 - 10 bar regions of the atmosphere .
wfc3 measurements of hd189733b during secondary eclipse would also help to confirm or refute the high co@xmath0 abundance required to explain the nicmos data . in the longer term ,
higher resolution , high signal - to - noise instruments such as those planned for the james webb space telescope or a future dedicated exoplanet atmosphere characterization mission will enable definitive determinations of the atmospheric properties of hot jupiters . [ cols="^,^,^,^,^,^,^",options="header " , ] c p8 cm p8 cm planet & spectroscopic data sources & broadband data sources + hd189733b & hst nicmos ( swain et al . 2009 ) , spitzer irs ( grillmair et al . 2007 ) & spitzer irac 3.6 , 4.5 ( knutson et al . 2012 ) , 5.7 , 7.8 ( agol et al . 2010 ) irs 16,mips 24 ( charbonneau et al . 2008 ) + + hd209458b & - & spitzer irac 3.6 , 4.5 , 5.7 ( knutson et al . 2008 ) , 7.8 ( h. knutson et al .
private com . ) , irs 16 ( madhusudhan & seager 2009 via d. deming private com),mips 24 ( crossfield et al .
2012a ) + + gj436b & - & spitzer irac 3.6 , 4.5 , 5.7 , 7.8 , irs 16,mips 24 ( stevenson et al .
2010 ) + + hd149026b & - & spitzer irac 3.6 , 4.5 , 5.7 , 7.8 , irs 16 ( stevenson et al . 2012 ) + + wasp-12b & hst wfc3 ( swain et al . 2013 ) & cfht wic 1.66 ( croll et al . 2011 ) , 2.15 ( zhao et al .
2012 ) , subaru nb 2.315 ( crossfield et al .
2012 ) , spitzer irac 3.6 , 4.5 , 5.7 , 7.8 ( crossfield et al . 2012 ) + + wasp-19b & - & vlt hawk - i1.620 ( anderson et al . 2010 ) , 2.095 ( gibson et al .
2010 ) , spitzer irac 3.6 , 4.5 , 5.7 , 7.8 ( anderson et al . 2013 ) + + was43b & - & vlt hawk - i 1.19 ( gillon et al . 2012 ) , 1.620 ( wang et al .
2013 ) , 2.1 ( gillon et al .
2012 ) , spitzer irac 3.6 , 4.5 ( blecic et al . 2013 ) + + tres-2b & - & cfht wic 2.15 ( croll et al .
2010a ) , spitzer irac 3.6 , 4.5 , 5.7 , 7.8 ( odonovan et al . 2010 ) + + tres-3b & - & cfht wic 2.15 ( croll et al .
2010b ) , spitzer irac 3.6 , 4.5 , 5.7 , 7.8 ( fressin et al . 2010 ) + + @xmath37hubble space telescope near infrared camera and multi - object spectrometer + @xmath38infrared spectrometer + @xmath39nfrared array camera + @xmath40multiband imaging photometer for spitzer + @xmath41wide field camera 3 + @xmath42we use both the null and ellipsoidal cases from cowan et al .
2012 as corrected by crossfield et al .
2012 + @xmath43very large telescope high acuity wide field k - band imager + @xmath44canada - france - hawaii telescope wide - field infrared camera + @xmath37we present the best fit mixing ratios and the resulting c / o ratio along with their 68@xmath17 confidence intervals .
the best fit is defined as the fit that produces the minimum cost function value , @xmath45 from equation [ eq : cost_func ] .
we quote the reduced cost - fucntion value as the cost function value divided by the number of data points , @xmath46 .
note that the best - fit values for some of the parameters can fall outside of the 68% confidence interval .
+ @xmath38since we place upper and lower limits on the demc prior , these results must be interpreted in the context of the prior .
the 68@xmath17 confidence interval resulting from the flat priors alone would give values that range from @xmath19 , or @xmath208 orders of magnitude .
one should use caution when interpreting uncertainties that approach these levels , which is indicative of little information gain from the data .
+ @xmath39the c / o ratios resulting from the prior would span @xmath47 , or @xmath20 3 orders of magnitude .
+ @xmath40we show both the oxygen rich planet ( orp ) values and the carbon rich planet ( crp ) values quoted in his investigation [ orp / crp ] +
we thank jonathan fortney , ian crossfield , xi zhang , nikku madahusudhan , caroline morely , and jonathan frain for insightful conversations .
we also thank the anonymous referee for their insightful comments that made this paper stronger .
this research was supported in part by an nai virtual planetary laboratory grant from the university of washington to the jet propulsion laboratory and california institute of technology .
allard , f. , hauschildt , p. h. , & schweitzer , a. 2000 , , 539 , 366 anderson , d. r. , smith , a. m. s. , madhusudhan , n. , et al .
2013 , , 430 , 3422 bechter , e. b. , crepp , j. r. , ngo , h. , et al . 2013 , arxiv:1307.6857 benneke , b. , & seager , s. 2012 , , 753 , 100 bergfors , c. , brandner , w. , daemgen , s. , et al . 2013 , , 428 , 182 blecic , j. , harrington , j. , madhusudhan , n. , et al . 2013 , arxiv:1302.7003 charbonneau , d. , knutson , h. a. , barman , t. , et al . 2008 , , 686 , 1341 cowan , n. b. , machalek , p. , croll , b. , et al
2012 , , 747 , 82 crossfield , i. j. m. , knutson , h. , fortney , j. , et al . 2012a , , 752 , 81 crossfield , i. j. m. , barman , t. , hansen , b. m. s. , tanaka , i. , & kodama , t. 2012b , , 760 , 140 croll , b. , albert , l. , lafreniere , d. , jayawardhana , r. , & fortney , j. j. 2010a , , 717 , 1084 croll , b. , jayawardhana , r. , fortney , j. j. , lafrenire , d. , & albert , l. 2010b , , 718 , 920 deming , d. , harrington , j. , seager , s. , & richardson , l. j. 2006 , , 644 , 560 ford , e. b. 2005 , , 129 , 1706 fortney , j. j. , marley , m. s. , lodders , k. , saumon , d. , & freedman , r. 2005 , , 627 , l69 fortney , j. j. , saumon , d. , marley , m. s. , lodders , k. , & freedman , r. s. 2006 , , 642 , 495 fortney , j. j. , lodders , k. , marley , m. s. , & freedman , r. s. 2008 , , 678 , 1419 grillmair , c. j. , charbonneau , d. , burrows , a. , et al .
2007 , , 658 , l115 grillmair , c. j. , burrows , a. , charbonneau , d. , et al .
2008 , , 456 , 767 guillot , t. 2010 , , 520 , a27 hebb , l. , collier - cameron , a. , loeillet , b. , et al .
2009 , , 693 , 1920 hellier , c. , anderson , d. r. , collier - cameron , a. , et al . 201a , , 730 , l31 hellier , c. , anderson , d. r. , collier cameron , a. , et al .
2011b , , 535 , l7 heng , k. , hayek , w. , pont , f. , & sing , d. k. 2012 , , 420 , 20 hubeny , i. , burrows , a. , & sudarsky , d. 2003 , , 594 , 1011 knutson , h. a. , et al .
2009 , , 690 , 822 knutson , h. a. , howard , a. w. , & isaacson , h. 2010 , , 720 , 1569 knutson , h. a. , benneke , b. , deming , d. , & homeier , d. 2014 , , 505 , 66 lee , j .- m . ,
fletcher , l. n. , & irwin , p. g. j. 2012 , , 420 , 170 lewis , n. k. , showman , a. p. , fortney , j. j. , et al .
2010 , , 720 , 344 line , m. r. , liang , m. c. , & yung , y. l. 2010 , , 717 , 496 line , m. r. , vasisht , g. , chen , p. , angerhausen , d. , & yung , y. l. 2011a , , 738 , 32 line , m. r. , zhang , x. , vasisht , g. , et al .
2012 , , 749 , 93 line , m. r. , wolf , a. s. , zhang , x. , et al .
2013 , , 775 , 137 madhusudhan , n. , & seager , s. 2009 , , 707 , 24 madhusudhan , n. , & seager , s. 2009 , , 707 , 24 madhusudhan , n. , harrington , j. , stevenson , k. b. , et al . 2011a , , 469 , 64 madhusudhan , n. , & seager , s. 2011 , , 729 , 41 madhusudhan , n. , mousis , o. , johnson , t. v. , & lunine , j. i. 2011b , , 743 , 191 madhusudhan , n. 2012 , , 758 , 36 miller - ricci kempton , e. , zahnle , k. , & fortney , j. j. 2012 , , 745 , 3 moses , j. i. , visscher , c. , fortney , j. j. , et al .
2011 , , 737 , 15 moses , j. i. , madhusudhan , n. , visscher , c. , & freedman , r. s. 2013a , , 763 , 25 moses , j. i. , line , m. r. , visscher , c. , et al . 2013b , arxiv:1306.5178 berg , k. i. , murray - clay , r. , & bergin , e. a. 2011 , , 743 , l16 parmentier , v. , showman , a. p. , & lian , y. 2013 , arxiv:1301.4522 pont , f. , sing , d. k. , gibson , n. p. , et al .
2013 , , 432 , 2917 press , w.h . ,
teukolksy , s. , vetterling , w.t . ,
fannery , b. 1995 , numerical recipies : the artof scientific computing , sec . ed .
press robinson , t. d. , & catling , d. c. 2012 , , 757 , 104 robinson , t. d. , & catling , d. c. 2013 , arxiv:1312.6859 rodgers , c. d. 1976 , reviews of geophysics and space physics , 14 , 609 rodgers , c.d . , inverse methods for atmospheric sounding , theory and practice , 2000 sato , b. , fischer , d. a. , henry , g. w. , et al .
2005 , , 633 , 465 shabram , m. , fortney , j. j. , greene , t. p. , & freedman , r. s. 2011 , , 727 , 65 showman , a. p. , fortney , j. j. , lian , y. , et al .
2009 , , 699 , 564 sozzetti , a. , torres , g. , charbonneau , d. , et al .
2009 , , 691 , 1145 spiegel , d. s. , silverio , k. , & burrows , a. 2009 , , 699 , 1487 stevenson , k. b. , et al .
2010 , , 464 , 1161 stevenson , k. b. , harrington , j. , fortney , j. j. , et al .
2012 , , 754 , 136 swain , m. r. , vasisht , g. , tinetti , g. , et al . 2009a , , 690 , l114 swain , m. , deroo , p. , tinetti , g. , et al .
2013 , icarus , 225 , 432 ter braak , c. j. , & vrugt , j. a. 2008 , statistics and computing , 18 , 435 - 446 | secondary eclipse spectroscopy provides invaluable insights into the temperatures and compositions of exoplanetary atmospheres .
we carry out a systematic temperature and abundance retrieval analysis of nine exoplanets ( hd189733b , hd209458b , hd149026b , gj436b , wasp-12b , wasp-19b , wasp-43b , tres-2b , and tres-3b ) observed in secondary eclipse using a combination of space- and ground - based facilities .
our goal with this analysis is to provide a consistent set of temperatures and compositions from which self - consistent models can be compared and to probe the underlying processes that shape these atmospheres .
this paper is the second in a three part series of papers exploring the retrievability of temperatures and abundances from secondary eclipse spectra and the implications of these results for the chemistry of exoplanet atmospheres . in this investigation
we present a catalogue of temperatures and abundances for h@xmath0o , ch@xmath1 , co , and co@xmath0 .
we find that our temperatures and abundances are generally consistent with those of previous studies , although we do not find any statistically convincing evidence for super - solar c to o ratios ( e.g. , solar c / o falls in the 1-sigma confidence intervals in eight of the nine planets in our sample ) .
furthermore , within our sample we find little evidence for thermal inversions over a wide range of effective temperatures ( with the exception of hd209458b ) , consistent with previous investigations .
the lack of evidence for inversions for most planets in our sample over such a wide range of effective temperatures provides additional support for the hypothesis that tio is unlikely to be the absorber responsible for the formation of these inversions . |
the main goal of the present paper is to give a rough survey of some _ combinatorial _ , often _ set - theoretical _ issues raised by recent work on representation problems in _
universal algebra_. it is an expanded version of the author s talk at the blast 2010 conference in boulder , colorado .
let us first quickly describe the universal algebraic background .
a _ signature _ ( or _ similarity type _ ) consists of a set @xmath8 , together with a map @xmath9 ( the _ arity map _ ) .
an _ algebra _ is a structure @xmath10 , where @xmath11 is a nonempty set and for each @xmath12 , the `` fundamental operation '' @xmath13 is a map from @xmath14 to @xmath11 , identified with the unique constant in its range in case @xmath15 . a _ congruence _ of @xmath16 is an equivalence relation on @xmath11 which is compatible with all the fundamental operations of @xmath16 .
the set @xmath17 of all congruences of @xmath16 forms an _ algebraic lattice _ under set inclusion , and we denote by @xmath18 its semilattice of compact elements .
hence @xmath18 is the @xmath19-semilattice of all finitely generated congruences of @xmath16 .
it is a central question in universal algebra to determine which algebraic lattices can be represented as @xmath17 , for an algebra @xmath16 in a given class of algebras . as it turns out that infinite _
@xmath19-semilattices _ are often more conveniently dealt with than infinite algebraic _ lattices _ , the class of questions above can be rephrased as which @xmath19-semilattices can be represented as @xmath18 , for an algebra @xmath16 in a given class of algebras
variety of algebras _ is the class of all the algebras satisfying a given set of identities ( in a given signature ) .
many combinatorial features of the work under discussion in the present paper are contained in the notion of _ critical point _ between varieties of algebras .
the following definition was introduced in the survey paper tma and wehrung @xcite , then slightly amended in gillibert @xcite . for a class @xmath20 of algebras , we set @xmath21 the _ compact congruence class _ of @xmath20 .
[ d : critpt ] the _ critical point _
@xmath22 , between classes @xmath0 and @xmath1 of algebras , is the least possible cardinality of a member of @xmath23 if @xmath24 , and @xmath25 if @xmath26 ) .
thus the critical point measures the `` containment defect '' of @xmath27 into @xmath28 .
it turns out that estimating the critical point between varieties of algebras , in particular varieties of lattices , is a highly nontrivial task , of which we shall give a rough outline in section [ s : critpt ] . proving that a critical point is _ large _ requires tools called _ ladders _
; we present a foretaste of the methods used in sections [ s : expleua ] ( for @xmath29-ladders ) and [ s : al1 ] ( for @xmath30-ladders ) .
although no universal - algebraic use of @xmath31-ladders has been found so far ( in my optimistic moments i do hope that there will be some ) , these objects raise fascinating set - theoretical questions .
furthermore , the failure of three - dimensional amalgamation is , in some sense , at the basis of nontrivial counterexamples ( cf .
rika , tma , and wehrung @xcite or wehrung @xcite ) , and one can not completely discard the possibility of future uses of that property , see section [ s : con3 ] . on the other hand , proving that a critical point is _ small _ involves quite different combinatorial tools , articulated around the notion of a _ @xmath4-lifter _
( section [ s : cll ] ) and the more amenable relation @xmath5 , for infinite cardinals @xmath6 and @xmath4 and a poset @xmath7 ( cf .
section [ s : kurat ] ) .
it turns out that the relation @xmath5 , when specialized to _ truncated cubes _
@xmath7 , gives a relation known to infinite combinatorists , namely the @xmath3 relation ( cf .
notation [ not : arrrel ] and proposition [ p : rightarr2leadsto ] ) . the more algebraic viewpoint brought by the @xmath5 framework made it possible , in gillibert and wehrung @xcite , to establish new relations of the form @xmath3 , of which we give an overview in section [ s : largefree ] . in the middle of `` large critical point results '' ( involving ladders ) or `` small critical point results '' ( involving the relation @xmath5 ) ,
there is a work of category theory ( cf .
gillibert and wehrung @xcite ) , introducing the categorical notion of a _ larder _ and the combinatorial notion of a _ lifter _
section [ s : cll ] ) .
these tools are mainly designed to introduce , and use , the _ condensate lifting lemma _ cll , and its main precursor the _ armature lemma_. these results make it possible to turn _ diagram counterexamples _ to _ object counterexamples_. while there are surprisingly many statements that can be proved that way ( beginning with those involving critical points ) , we present in section [ s : bowtie ] a quite isolated diagram counterexample , due to tma and the author , for which there is still no known associated object counterexample .
we shall use basic set - theoretical notation and terminology about ordinals and cardinals .
we denote by @xmath32 the image of a set @xmath33 under a function @xmath34 .
cardinals are initial ordinals .
we denote by @xmath35 the cofinality of an ordinal @xmath36 .
we denote by @xmath37 the first limit ordinal , mostly denoted by @xmath38 in case it is viewed as a cardinal .
we identify every non - negative integer @xmath39 with the finite set @xmath40 ( so @xmath41 ) .
we denote by @xmath42 the successor cardinal of a cardinal @xmath6 , and we define @xmath43 , for a non - negative integer @xmath39 , by @xmath44 and @xmath45 .
we denote by @xmath46 the powerset of a set @xmath33 , and we set @xmath47^\kappa&:={{\{y\in{\mathfrak{p}}(x)\mid\operatorname{card}y=\kappa\}}}\,,\\ [ x]^{<\kappa}&:={{\{y\in{\mathfrak{p}}(x)\mid\operatorname{card}y<\kappa\}}}\,,\\ [ x]^{{{\leqslant}}\kappa}&:={{\{y\in{\mathfrak{p}}(x)\mid\operatorname{card}y\leq\kappa\}}}\ , , \end{aligned}\ ] ] for every cardinal @xmath6 . by `` countable ''
we shall always mean `` at most countable '' .
all our posets will be nonempty . for posets @xmath7 and @xmath48 ,
a map @xmath49 is _ isotone _ if @xmath50 implies that @xmath51 , for all @xmath52 .
we denote by @xmath53 the least element of @xmath7 if it exists .
an element @xmath54 in a poset @xmath7 is _ join - irreducible _ if @xmath55 implies that @xmath56 , for every ( possibly empty ) finite subset @xmath33 of @xmath7 ; we denote by @xmath57 the set of all join - irreducible elements of @xmath7 , endowed with the induced partial ordering .
we set @xmath58 for all subsets @xmath48 and @xmath33 of @xmath7 ; in case @xmath59 is a singleton , then we shall write @xmath60 instead of @xmath61 , and so on .
we say that @xmath7 is _ lower finite _ if @xmath62 is finite for each @xmath63 .
a subset @xmath48 of @xmath7 is a _ lower subset of @xmath7 _ if @xmath64 .
we say that @xmath7 is a _ tree _
if @xmath7 has a smallest element and @xmath62 is a chain for each @xmath63 .
ideal _ of a poset @xmath7 is a nonempty , upward directed , lower subset of @xmath7 .
we recall the definitions of a few classical numerical invariants associated to a finite poset @xmath7 : * the _ width _ of @xmath7 , denoted by @xmath65 , is the maximal cardinality of a pairwise incomparable subset of @xmath7 .
* the _ breadth _ of @xmath7 , denoted by @xmath66 , is the least natural number @xmath67 such that for all @xmath68 , for @xmath69 , if @xmath70 for all @xmath71 , then there exists @xmath72 such that @xmath73 . *
the _ order - dimension _ of @xmath7 , denoted by @xmath74 , is the least possible natural number @xmath67 such that @xmath7 embeds , as a poset , into a product of @xmath67 chains .
a @xmath19-semilattice @xmath75 is _ distributive _ if for all @xmath76 , if @xmath77 , then there are @xmath78 and @xmath79 in @xmath80 such that @xmath81 .
equivalently , the ideal lattice @xmath82 of @xmath75 is a distributive lattice ( cf .
grtzer ( * ? ? ?
* section ii.5 ) ) .
let us start with an example .
we are given a class @xmath20 of algebras , and we assume that every _ finite boolean _ lattice is isomorphic to @xmath18 for some @xmath83 . can we extend this representation result to _ countable _ distributive @xmath19-semilattices ? without additional assumptions , this can not be done .
for example , let @xmath84 be the variety of all _ distributive lattices_. then every finite boolean lattice @xmath85 belongs to @xmath86 ( for @xmath87 ) , but for any @xmath88 , the @xmath19-semilattice @xmath89 is generalized boolean , thus it is a lattice ; and many distributive @xmath19-semilattices ( even countable ones ) are not lattices .
can we find reasonable assumptions , on the class @xmath20 , that make it possible to represent any _ countable _ distributive @xmath19-semilattice , starting with a _
representability result ?
here is the required finite representability result .
one - dimensional amalgamation property for @xmath90 the statement above involves a yet undefined notation , namely @xmath91 , where @xmath34 is no longer an _ object _ , but a _ morphism_. nor surprisingly , @xmath91 is defined as the map from @xmath18 to @xmath92 that to every compact congruence @xmath36 of @xmath16 associates the congruence of @xmath85 generated by @xmath93 . the map @xmath91 is a @xmath19-homomorphism from @xmath18 to @xmath92 .
we obtain the missing piece of information that is necessary in order to understand the `` amalgamation '' statement above , with a plus about _ direct limits _ ( called , in category theory , _ directed colimits _ ) .
the one - dimensional amalgamation property is illustrated in figure [ fig : con1 ] .
@xmath94^{{\varepsilon}}_{\cong } & & { { \boldsymbol{s}}}\\ \operatorname{con_c}{{\boldsymbol{a}}}\ar[u]^{\varphi } \save+<0ex,-5ex>\drop{\text{starting from this\dots}}\restore & & & { { \boldsymbol{a}}}\ar[u]^f & \save+<0ex,-5ex>\drop{\text{\dots getting to that}}\restore & \operatorname{con_c}{{\boldsymbol{a}}}\ar[ul]^{\operatorname{con_c}f}\ar[ur]_{\varphi } & } \ ] ] [ l : concfunct ] the assignment @xmath95 , @xmath96 defines a _ functor _ from the class of all algebras with a given signature , with its homomorphisms , to the category of all @xmath19-semilattices with @xmath19-homomorphisms .
this functor preserves all direct limits .
now suppose that the one - dimensional amalgamation property holds for the functor @xmath90 on the class @xmath20 .
let @xmath75 be a countable @xmath19-semilattice . using a result from bulman - fleming and mcdowell @xcite
( see also goodearl and wehrung @xcite for a self - contained proof ) , there is a representation of the form @xmath97 where all the @xmath98 are _ finite boolean _ @xmath19-semilattices and all the transition maps of the direct system in are @xmath19-homomorphisms .
suppose that @xmath99 , with an isomorphism @xmath100 , and denote by @xmath101 the transition map associated with the direct system in .
we get a @xmath19-homomorphism @xmath102 , thus , by the one - dimensional amalgamation assumption , there are @xmath103 , a homomorphism @xmath104 , and an isomorphism @xmath105 such that @xmath106 . defining @xmath16 as the direct limit of the system @xmath107^{f_0 } & { { \boldsymbol{a}}}_1\ar[r]^{f_1 } & { { \boldsymbol{a}}}_2\ar[r]^{f_2 } & \cdots\cdots\ , , } \ ] ] we obtain , as the functor @xmath90 preserves direct limits , that @xmath108 .
so we have obtained the following result .
[ p : con1toctble ] let @xmath20 be a class of algebras of the same signature . if the one - dimensional amalgamation property holds for @xmath90 on @xmath20 , then every countable distributive @xmath19-semilattice is isomorphic to @xmath18 for the direct limit @xmath16 of some countable sequence in @xmath20 .
this approach is used in rika , tma , and wehrung @xcite to establish the following statement .
[ c : con1lgrps ] every distributive algebraic lattice with countably many compact elements is isomorphic to the congruence lattice of some lattice - ordered group .
the corresponding statement with `` countable '' replaced by `` with at most @xmath109 elements '' is still open .
yet there are also known positive results in cardinality @xmath109 , for which we shall set the framework in the next section .
it will turn out that none of the corresponding representation results holds at cardinality @xmath110 .
in the context of section [ s : expleua ] , one may try to represent larger distributive @xmath19-semilattices than just the countable ones .
thus the first step consists of trying semilattices of cardinality @xmath109 .
the nave approach , consisting of keeping the one - dimensional amalgamation property and replacing the chain @xmath111 of all natural numbers by the larger chain @xmath112 of all countable ordinals , obviously fails : if @xmath113 , then there is no way to represent @xmath114 with all @xmath115 finite .
the relevant strengthening of the assumption on @xmath20 is the following .
two - dimensional amalgamation property for @xmath90 the two - dimensional amalgamation property can be illustrated by figure [ fig : con2 ] .
on that picture , the plain vanilla arrows are those that are assumed to exist from the start , while the dotted ones are those whose existence follows from the property .
@xmath116^(.4){{\varepsilon } } \\ { { \boldsymbol{a}}}_1\ar@{-->}[ru]^{g_1 } & & { { \boldsymbol{a}}}_2\ar@{-->}[lu]_{g_2 } & \operatorname{con_c}{{\boldsymbol{a}}}_1\ar@{-->}[ur]_{\operatorname{con_c}g_1}\ar[uur]^{\psi_1 } & & \operatorname{con_c}{{\boldsymbol{a}}}_2\ar@{-->}[ul]^{\operatorname{con_c}g_2}\ar[uul]_{\psi_2}\\ & { { \boldsymbol{a}}}_0\ar[ul]^{f_1}\ar[ur]_{f_2 } & & & \operatorname{con_c}{{\boldsymbol{a}}}_0\ar[ul]^{\operatorname{con_c}f_1}\ar[ur]_{\operatorname{con_c}f_2 } } \ ] ] assuming two - dimensional congruence amalgamation , we still need to replace the chain @xmath111 by a poset of cardinality @xmath109 that makes it possible to use two - dimensional congruence amalgamation .
we remind the reader that lower finite posets are defined in section [ su : posets ] .
[ d : kladd ] let @xmath2 be a positive integer .
a lower finite lattice @xmath117 is a _
@xmath2-ladder _ if every element of @xmath118 has at most @xmath2 lower covers in @xmath117 . note that every @xmath2-ladder has breadth at most @xmath2 .
the @xmath29-ladders are exactly the finite chains and the chain @xmath111 of all natural numbers .
note that @xmath2-ladders are called _
@xmath2-frames _ in h. dobbertin @xcite . due to a terminology conflict with another lattice - theoretical concept , the _
von neumann frames _ , we chose in grtzer , lakser , and wehrung @xcite to call these objects _ ladders_. the concept originates in s.z .
ditor @xcite , where @xmath2-ladders are simply called _
@xmath2-lattices_. ditor proves in @xcite that every @xmath2-ladder has at most @xmath119 elements .
( this is also a straightforward consequence of kuratowski s free set theorem , see kuratowski @xcite . ) more relevant to our present congruence representation problem , ditor proves the following .
[ p:2laddal1 ] there exists a @xmath30-ladder of cardinality @xmath109 .
now suppose that two - dimensional congruence amalgamation holds .
we prove that every distributive @xmath19-semilattice @xmath75 with at most @xmath109 elements is isomorphic to @xmath18 , for some direct limit @xmath16 of members of @xmath20 .
write @xmath120 , for a directed poset @xmath7 with at most @xmath109 elements and a direct system @xmath121 of finite boolean @xmath19-semilattices and @xmath19-homomorphisms .
for example , @xmath7 can be assumed to be the set of all finite subsets of @xmath122 , partially ordered under set inclusion .
let @xmath123 be a @xmath30-ladder of cardinality @xmath109 .
as @xmath123 is lower finite , it is easy to construct , by induction , an isotone cofinal map @xmath124 ; this reduces the situation to the case where @xmath125 .
then , using two - dimensional amalgamation and the definition of a @xmath30-ladder , one constructs , by induction within the ladder , an @xmath123-indexed direct system @xmath126 in @xmath20 such that @xmath127 .
now take @xmath128 . as the functor @xmath90 preserves direct limits , @xmath108 .
we have thus outlined the proof of the following .
[ p : con2toal1 ] let @xmath20 be a class of algebras of the same signature .
if the two - dimensional amalgamation property holds for @xmath90 on @xmath20 , then every distributive @xmath19-semilattice with at most @xmath109 elements is isomorphic to @xmath18 for the direct limit @xmath16 of a diagram , indexed by a @xmath30-ladder of cardinality @xmath109 , of members of @xmath20 .
this argument has turned out to be very useful in proving various representation results , in size @xmath109 , with respect to the @xmath90 functor , starting with a.p .
huhn @xcite ; see , for example , grtzer , lakser , and wehrung @xcite , or rika , tma , and wehrung @xcite .
see also wehrung @xcite for a representation result with respect to another functor than @xmath90 , namely the compact ideal semilattice functor on a regular ring .
for higher cardinalities than @xmath109 , the situation gets somehow romantic .
there are , to this date , no known congruence representation results specifically stated in cardinality @xmath110 .
yet there are theorems , now of set - theoretical nature , which are an order of magnitude harder than those about representation in cardinality below @xmath109 .
these results go together with an array of still unsolved problems .
we first deal with the easy problem of @xmath31-dimensional congruence amalgamation , for a class @xmath20 of algebras with the same signature . set @xmath129 ( recall that @xmath130 ) , partially ordered under set inclusion , and set @xmath131 .
the statement of three - dimensional amalgamation is quite discouraging at first sight : three - dimensional amalgamation property for @xmath90 this can be explained as follows .
we are given a truncated cube ( here , a @xmath132-indexed diagram ) @xmath126 of members of @xmath20 , and we form the image @xmath133 of this diagram under the @xmath90 functor . at the top of this diagram , we put the finite boolean semilattice @xmath75 , thus getting a cube @xmath134 of @xmath19-semilattices . three - dimensional amalgamation states that there is a way to adjoin a top vertex to @xmath126 , thus getting a cube @xmath135 , such that @xmath136 , in a way that preserves the already existing relations .
this is illustrated in figure [ fig : con3 ] , with @xmath137 for each @xmath138 .
the unlabeled arrows should be obvious : for example , the arrow from @xmath139 to @xmath140 is @xmath141 ; and so on .
@xmath142_(.4){{\varepsilon}}^(.4){\cong } & \\ { { \boldsymbol{a}}}_{{\{0,1\}}}\ar@{-->}[ur]^{\alpha_{{\{0,1\ } } } } & { { \boldsymbol{a}}}_{{\{0,2\}}}\ar@{-->}[u]^(.3){\alpha_{{\{0,2\ } } } } & { { \boldsymbol{a}}}_{{\{1,2\}}}\ar@{-->}[ul]_{\alpha_{{\{1,2\ } } } } & \operatorname{con_c}{{\boldsymbol{a}}}_{{\{0,1\}}}\ar@{-->}[ur]_(.42){{\overline{\alpha}}_{{\{0,1\ } } } } \ar@/^1pc/[uur]^{\sigma_{{\{0,1\ } } } } & \operatorname{con_c}{{\boldsymbol{a}}}_{{\{0,2\}}}\ar@{-->}[u]_(.3){{\overline{\alpha}}_{{\{0,2\ } } } } \ar@/^1.8pc/[uu]^(.5){\sigma_{{\{0,2\}}}\!\ ! } & \operatorname{con_c}{{\boldsymbol{a}}}_{{\{1,2\}}}\ar@{-->}[ul]^(.4){{\overline{\alpha}}_{{\{1,2\}}}\!\ ! } \ar[uul]_{\sigma_{{\{1,2\}}}}\\ { { \boldsymbol{a}}}_{{\{0\}}}\ar[u]\ar[ur ] & { { \boldsymbol{a}}}_{{\{1\}}}\ar[ul]\ar[ur ] & { { \boldsymbol{a}}}_{{\{2\}}}\ar[ul]\ar[u ] & \operatorname{con_c}{{\boldsymbol{a}}}_{{\{0\}}}\ar[u]\ar[ur ] & \operatorname{con_c}{{\boldsymbol{a}}}_{{\{1\}}}\ar[ul]\ar[ur ] & \operatorname{con_c}{{\boldsymbol{a}}}_{{\{2\}}}\ar[ul]\ar[u]\\ & { { \boldsymbol{a}}}_{{\varnothing}}\ar[ul]\ar[u]\ar[ur ] & & & \operatorname{con_c}{{\boldsymbol{a}}}_{{\varnothing}}\ar[ul]\ar[u]\ar[ur ] & } \ ] ] unfortunately , the following result shows that three - dimensional amalgamation occurs only in exceptional cases .
[ p:3amalg ] suppose that a class @xmath20 of algebras of the same signature satisfies three - dimensional amalgamation .
let @xmath143 , @xmath144 , @xmath145 be members of @xmath20 with one - to - one homomorphisms @xmath146 .
we suppose that @xmath143 is a substructure of @xmath144 . if @xmath147 , then @xmath148 .
assume that @xmath147 .
set @xmath149 , and for each @xmath150 , denote by @xmath151 the unique @xmath19-homomorphism that sends every nonzero element to @xmath29 .
we apply three - dimensional amalgamation to the diagrams represented in figure [ fig : con31 ] , minus @xmath16 , @xmath18 , and the dotted arrows .
note that the one - to - oneness of both @xmath34 and @xmath152 ensures that both maps @xmath153 and @xmath154 separate zero , thus the right hand side diagram commutes .
three - dimensional amalgamation ensures the existence of @xmath16 together with homomorphisms @xmath155 and @xmath156 with an isomorphism @xmath157 such that both diagrams of figure [ fig : con31 ] ( dotted arrows included ) commute .
again , we are using the notational convention @xmath158 , and so on .
the unlabeled arrows should be obvious , for example the arrow from @xmath143 to @xmath144 is the inclusion map .
@xmath159_(.4){{\varepsilon}}^(.4){\cong } & \\ { { \boldsymbol{a}}}_1\ar@{-->}[ur]^u & { { \boldsymbol{a}}}_2\ar@{-->}[u]^(.4)w & { { \boldsymbol{a}}}_1\ar@{-->}[ul]_v & \operatorname{con_c}{{\boldsymbol{a}}}_1\ar@{-->}[ur]_(.42){{\overline{u } } } \ar@/^1pc/[uur]^{\sigma_1 } & \operatorname{con_c}{{\boldsymbol{a}}}_2\ar@{-->}[u]_(.3){{\overline{w } } } \ar@/^1.8pc/[uu]^(.5){\sigma_2\ ! } & \operatorname{con_c}{{\boldsymbol{a}}}_1\ar@{-->}[ul]^(.42){{\overline{v } } } \ar[uul]_{\sigma_1}\\ { { \boldsymbol{a}}}_1\ar@{=}[u]\ar@{^(->}[ur]_(.35){\!\!\!f } & { { \boldsymbol{a}}}_1\ar@{=}[ul]\ar@{=}[ur ] & { { \boldsymbol{a}}}_1\ar@{_(->}[ul]^(.35){g\!\!}\ar@{=}[u ] & \operatorname{con_c}{{\boldsymbol{a}}}_1\ar@{=}[u]\ar[ur]_(.4){\!\!\!{\overline{f } } } & \operatorname{con_c}{{\boldsymbol{a}}}_1\ar@{=}[ul]\ar@{=}[ur ] & \operatorname{con_c}{{\boldsymbol{a}}}_1\ar[ul]^(.4){{\overline{g}}\!\!}\ar@{=}[u]\\ & { { \boldsymbol{a}}}_0\ar@{_(->}[ul]\ar@{_(->}[u]\ar@{^(->}[ur ] & & & \operatorname{con_c}{{\boldsymbol{a}}}_0\ar[ul]\ar[u]\ar[ur ] & } \ ] ] from the commutativity of the right hand side diagram it follows that @xmath160 . as @xmath161 separates zero and
@xmath162 is an isomorphism , @xmath163 separates zero , which means that @xmath164 is one - to - one . from the commutativity of the left hand side diagram it follows that @xmath165 , @xmath166 , and @xmath167 ; so @xmath168 .
as @xmath164 is one - to - one , we obtain that @xmath148 .
proposition [ p:3amalg ] suggests at first sight that it is quite hopeless to achieve any representation proof in cardinality @xmath110 _ via _ three - dimensional amalgamation .
although some hope remains to apply three - dimensional amalgamation on special configurations , such as _ strong amalgamation _
, no such application has been found so far . even in case , some day , someone manages to work out a usable version of three - dimensional amalgamation , another catch is the _ existence of ladders_. indeed , in order to be able to extend the proof of proposition [ p : con2toal1 ] to classes satisfying some version of three - dimensional amalgamation , we need the existence of a @xmath31-ladder of cardinality @xmath110 . this problem was first raised in ditor @xcite .
ditor s problem does there exist a @xmath31-ladder of cardinality @xmath110 ? the following result , by wehrung @xcite , shows that a positive solution of ditor s problem is at least consistent with @xmath169 .
[ t : ex3ladd ] suppose that either the axiom @xmath170 holds or there exists a gap-@xmath29 morass . then there exists a @xmath31-ladder of cardinality @xmath110 .
in particular , as every @xmath171 $ ] , for @xmath172 , has a gap-@xmath29 morass ( cf .
devlin @xcite ) , it follows that _ if there is no @xmath31-ladder of cardinality @xmath110 , then @xmath173 is inaccessible in the constructible universe @xmath174_. the axiom @xmath170 , which is martin s axiom restricted to collections of @xmath109 dense subsets in posets of precaliber @xmath109 ( cf .
weiss ( * ? ? ?
* section 3 ) ) , holds in a generic extension of the universe by a suitable notion of forcing satisfying the _ countable chain condition _
jech ( * ? ? ?
* section 16 ) ) . on the other hand ,
a gap-@xmath29 morass exists in a generic extension by a suitable _ countably closed _ notion of forcing ( cf .
velleman ( * ? ? ?
* section 3.3 ) ; see also brooke - taylor and friedman ( * ? ? ? * proposition 25 ) ) . as countably closed posets with the countable chain condition
are , from the forcing viewpoint , trivial , the two notions of forcing involved in those generic extensions can not be the same : they must be , in some sense , `` orthogonal '' !
this suggests a positive solution to ditor s problem in @xmath169 , nevertheless no such solution has been found so far .
our proof of theorem [ t : ex3ladd ] uses the equivalent form of morasses called _ simplified morasses _ introduced in velleman @xcite .
it seems reasonable to expect that higher morasses , or higher versions of martin s axiom , would imply the existence of @xmath2-ladders of cardinality @xmath119 for @xmath175 . in an ideal world , ditor s problem would first be settled in @xmath169 alone .
however , we are not living in an ideal world
propositions [ p : con1toctble ] and [ p : con2toal1 ] are typical illustrations of a `` finite '' representation result on the functor @xmath90 making it possible to prove an `` infinite '' representation result .
now there are situations where the problematic is just opposite : namely , going from an infinite representation result to a finite one .
finite would appear as a limit case of infinite ! a paradigm for such situations ( _ critical points _ ) will shortly be given in section [ s : critpt ] .
the arguments underlying propositions [ p : con1toctble ] and [ p : con2toal1 ] are best set in a _ category theory _ context .
basically , we are given _ categories _ @xmath0 , @xmath1 , @xmath176 together with _
functors _ @xmath177 and @xmath178 .
we are trying to find an assignment @xmath179 such that @xmath180 , _ naturally in @xmath11 _ , on a `` large '' subcategory of @xmath0 .
we can paraphrase this further by saying that we are trying to make @xmath181 `` as functorial as possible '' .
hence we need an assumption of the form `` for many @xmath182 , there exists @xmath183 such that @xmath184 '' .
the situation is illustrated in figure [ fig : getfctgam ] .
@xmath185^{\phi } & \save+<0ex,-5ex>\drop{\text{starting from this\dots}}\restore & { \mathcal{b}}\ar[ul]_{\psi } & & { \mathcal{a}}\ar[ur]^{\phi}\ar[rr]^{\gamma } & \save+<0ex,-5ex>\drop{\text{\dots getting to that}}\restore & { \mathcal{b}}\ar[ul]_{\psi } } \ ] ] this problem is tackled , in gillibert and wehrung @xcite , by the introduction of a whole theory , involving a categorical concept called a _
larder_. a larder consists of categories @xmath0 , @xmath1 , @xmath176 together with functors @xmath177 and @xmath178 , plus a few add - ons , namely full subcategories @xmath186 and @xmath187 together with a subcategory @xmath188 . these add - ons are required to satisfy a few obvious - looking preservation properties ( e.g. , directed colimits ) , plus a less obvious - looking `` lwenheim - skolem type '' property with respect to the functor @xmath189 .
an important additional attribute of a larder , reminiscent of the first isomorphism theorem in group theory , is called _
projectability_. most applications of our theory involve a result called the `` condensate lifting lemma '' , or cll , which we paraphrase below .
condensate lifting lemma suppose that @xmath0 , @xmath1 , @xmath176 , @xmath190 , and @xmath189 are part of a projectable larder and let @xmath7 be a `` nice '' poset .
if @xmath191 then @xmath192 in this statement , @xmath193 denotes the category whose objects are all the direct systems @xmath194 , with @xmath195 an object of @xmath0 and @xmath196 , with the natural transformations as morphisms .
we shall go back to the meaning of `` nice poset '' in a moment . to paraphrase cll further , if @xmath189 lifts many _ objects
_ , then it lifts many ( @xmath7-indexed ) _
diagrams_. yet another way to see this is that cll turns _
diagram counterexamples _ to _ object counterexamples_. applications of cll , or its main precursor the _ armature lemma _ , include the following . * let @xmath197 be a nondistributive variety of lattices . then the free lattice ( resp .
, free bounded lattice ) in @xmath197 on @xmath109 generators has no congruence - permutable , congruence - preserving extension ( cf . gillibert and wehrung @xcite ) .
* there exists a non - coordinatizable sectionally complemented modular lattice ( without unit ) , of cardinality @xmath109 , with a large @xmath198-frame ( cf .
wehrung @xcite ) .
* there exists a lattice of cardinality @xmath109 , in the variety generated by the five - element modular non - distributive lattice @xmath199 , without any congruence @xmath39-permutable , congruence - preserving extension for any positive integer @xmath39 ( cf .
gillibert @xcite ) .
* let @xmath0 and @xmath1 be varieties of lattices such that @xmath0 is contained neither in @xmath1 nor its dual , and every simple member of @xmath1 has a prime interval .
then the critical point @xmath22 ( cf .
definition [ d : critpt ] ) lies below @xmath110 ( cf .
gillibert @xcite ) .
* there is a unital exchange ring @xmath200 , of cardinality @xmath201 , such that the monoid @xmath202 of all isomorphism types of finitely generated right @xmath200-modules is not isomorphic to @xmath203 for any ring @xmath204 that is either von neumann regular or a c*-algebra of real rank zero ( cf .
wehrung @xcite ) .
* let @xmath0 and @xmath1 be locally finite varieties of algebras such that every finite @xmath19-semilattice has , up to isomorphism , only finitely many liftings , with respect to the functor @xmath90 , in @xmath1 , and every such lifting is finite .
( _ the latter condition holds , in particular , if @xmath1 omits the tame congruence theory types @xmath205 and @xmath206 ; thus , in particular , if @xmath1 satisfies a nontrivial congruence lattice identity , see hobby and mckenzie @xcite_. ) if @xmath27 is not contained in @xmath28 , then @xmath207 ( cf .
gillibert @xcite ) .
there is still one completely undefined term in the statement of cll above , namely `` nice '' , as an attribute of a poset . not
every poset qualifies as `` nice '' .
the definition of `` niceness '' , formally _ existence of a @xmath4-lifter _
( @xmath4 is a certain infinite cardinal depending of the data ) , is relatively complex and thus we shall relegate it to section [ s : app ] ( cf . definition [ d : lifter ] ) .
nonetheless , it is still possible to describe with relatively little effort which posets are dealt with by cll , and which are not .
observe that these definitions have , at first sight , little to do with category theory .
[ not : sor ] let @xmath33 be a subset in a poset @xmath7 .
we denote by @xmath208 the set of all minimal elements of @xmath209 .
[ d : pjs ] we say that a subset @xmath33 in a poset @xmath7 is _ @xmath210-closed _ if @xmath211 for any finite @xmath212 .
the _ @xmath210-closure _ of a subset @xmath33 of a poset @xmath7 is the least @xmath210-closed subset of @xmath7 containing @xmath33 .
we say that @xmath7 is * a _ pseudo join - semilattice _ if for each finite @xmath213 there exists a finite @xmath214 such that @xmath215 ( cf .
section [ su : posets ] ) ; * _ supported _ if @xmath7 is a pseudo join - semilattice and the @xmath210-closure of every finite subset of @xmath7 is finite ; * an _ almost join - semilattice _
if @xmath7 is a pseudo join - semilattice and @xmath62 is a join - semilattice for each @xmath63 . _
every @xmath19-semilattice is an almost join - semilattice_. furthermore , every almost join - semilattice is supported and every supported poset is a pseudo join - semilattice .
both converse containments fail , as illustrated in figure [ fig : threeposets ] ; the leftmost poset is not a pseudo join - semilattice .
as mentioned above , `` nice '' in the statement of cll means having a @xmath4-lifter , where , roughly speaking , @xmath4 is an upper bound on the sizes of the categorical data .
this brings a severe restriction on the shape of @xmath7 , involving a connection between the objects of definition [ d : pjs ] and the definition of a lifter ( definition [ d : lifter ] ) : namely , _ if @xmath7 has a @xmath4-lifter , then it is an almost join - semilattice_. more precisely , the following result is proved in gillibert and wehrung @xcite .
it relates existence of lifters , almost join - semilattices , and the more familiar - looking ( at least for those familiar with large free sets problems ) infinite combinatorial statement in ( ii ) .
[ t : shapeliftable ] let @xmath4 be an infinite cardinal and let @xmath216 be a @xmath4-lifter of a poset @xmath7 . put @xmath217 .
then the following statements hold : 1 .
@xmath7 is a disjoint union of finitely many almost join - semilattices with zero . 2 . for every
_ isotone _ map @xmath218^{<\operatorname{cf}(\lambda)}\to[\kappa]^{<\lambda}$ ] , there exists a _
one - to - one _ map @xmath219 such that @xmath220 in particular , as none of the posets represented on figure [ fig : threeposets ] is an almost join - semilattice , none of them has a lifter .
as to the statement of cll , these posets are not `` nice '' .
item ( ii ) of theorem [ t : shapeliftable ] bears an intriguing relationship with the problem of _ well - foundedness _ of liftable posets .
the following result is proved in gillibert and wehrung @xcite .
[ t : wfstrcard ] let @xmath4 be an infinite cardinal such that @xmath221 if a poset has a @xmath4-lifter , then it is well - founded . in particular , if a poset has a @xmath222-lifter , then it is well - founded .
hence , the poset @xmath223 , with the total ordering defined by @xmath224 , has no @xmath222-lifter .
it is also proved in gillibert and wehrung @xcite that if an ill - founded ( i.e. , non well - founded ) poset has a @xmath4-lifter , then so does @xmath225 . by using theorem [ t : shapeliftable ]
, we obtain the following . [ c : w+1lif ] if there exists an ill - founded poset with an @xmath109-lifter , then there exists an infinite cardinal @xmath6 such that for each _ isotone _ @xmath218^{{\leqslant}\aleph_0}\to[\kappa]^{{\leqslant}\aleph_0}$ ] , there exists a sequence @xmath226 from @xmath6 such that @xmath227 for each @xmath228 .
we do not know whether the final statement of corollary [ c : w+1lif ] can occur at all ( cf .
problem [ pb : supererd ] ) . actually , we do not know whether @xmath225 can have a @xmath4-lifter , for any infinite cardinal @xmath4 . by theorem [ t : wfstrcard ]
, @xmath229 is the first possible candidate .
a good illustration of the various combinatorial principles involved in section [ s : cll ] is provided by the theory of _ critical points _ ( cf
. definition [ d : critpt ] ) .
an upper bound for many critical points is provided by the following deep and difficult result , proved in gillibert @xcite .
[ t : critptdich ] let @xmath0 and @xmath1 be varieties of _ algebras _ , with @xmath0 locally finite and @xmath1 finitely generated congruence - distributive . if @xmath24 , then @xmath230 .
this result is extended to quasivarieties ( and relative congruence lattices ) in gillibert and wehrung @xcite . moreover , by using the results of tame congruence theory tct ( cf .
hobby and mckenzie @xcite ) , it is shown in gillibert and wehrung @xcite that theorem [ t : critptdich ] can be extended to the case where @xmath1 has finite signature , is finitely generated , and omits both tct types @xmath205 and @xmath206 . in particular , the latter condition holds in case @xmath1 satisfies a nontrivial congruence lattice identity .
the strongest refinement of theorem [ t : critptdich ] to date , obtained in gillibert @xcite , is for @xmath0 and @xmath1 both locally finite with @xmath1 `` strongly congruence - proper '' ( which holds in case @xmath1 omits both tct types @xmath205 and @xmath206 ) ; furthermore , the conclusion is strengthened there to @xmath207 .
the bound @xmath110 is sharp . for lattice varieties ,
theorem [ t : critptdich ] is refined further in gillibert @xcite .
the proof of theorem [ t : critlatvar ] is extremely involved , and it uses a precursor of cll ( cf .
section [ s : cll ] ) called the _
armature lemma_. [ t : critlatvar ] let @xmath0 and @xmath1 be varieties of _ lattices _ such that every simple member of @xmath1 has a prime interval . if @xmath0 is contained neither in @xmath1 nor in the dual variety of @xmath1 , then @xmath24 and @xmath207 .
the bound @xmath110 is sharp .
it is not known whether the assumption about simple members of @xmath1 can be removed from the assumptions of theorem [ t : critlatvar ] . in theorem [ t :
critlatvar ] , the optimality statement on the bound @xmath110 follows from earlier results of ploica @xcite and gillibert @xcite : ploica obtains examples with @xmath231 and other examples with @xmath232 , while gillibert obtains an example with @xmath233 . in all those examples , @xmath0 and @xmath1 can be taken _
finitely generated modular _ lattice varieties and @xmath234 .
estimates for further critical points , between more complicated modular lattice varieties , are given in gillibert @xcite .
establishing an estimate of the form @xmath235 requires an analogue of proposition [ p : con1toctble ] , in particular a direct limit along the chain @xmath111 of all natural numbers .
likewise , establishing an estimate of the form @xmath236 requires an analogue of proposition [ p : con2toal1 ] , in particular a direct limit along a @xmath30-ladder of cardinality @xmath109 . on the other hand ,
proving an estimate of the form @xmath237 requires different techniques , the combinatorial aspects of which we shall outline in section [ s : kurat ] . in many proofs of such estimates ,
one first finds a finite diagram @xmath126 in @xmath0 , indexed by a finite lattice @xmath7 , such that @xmath133 is never isomorphic to @xmath238 for any @xmath7-indexed diagram @xmath135 in @xmath1 ; then , using cll , this diagram counterexample is turned to an object counterexample in @xmath0 . before going back , in section [ s : kurat ] , to the combinatorial aspects of such arguments , we shall see in section [ s : bowtie ] that certain problems may have diagram counterexamples but no known object counterexamples .
the congruence lattice problem ( clp ) , raised by r.p .
dilworth in the forties , asked whether every distributive algebraic lattice is isomorphic to the congruence lattice of a lattice ; equivalently , whether every distributive @xmath19-semilattice is isomorphic to @xmath239 for a lattice @xmath117 .
this problem got finally settled in the negative in wehrung @xcite ; the sharp bound for the size of the counterexample ( viz .
@xmath110 ) was obtained in rika @xcite . for an account of the problem ,
see grtzer @xcite .
before clp got settled , pudlk @xcite asked whether clp could have a positive , _ functorial _ solution , that is , a functor @xmath181 , from distributive @xmath19-semilattices and @xmath19-homomorphisms to lattices and lattice homomorphisms , such that @xmath240 naturally in @xmath75 for any distributive @xmath19-semilattice @xmath75 .
this problem got solved in the negative ( before clp got settled ) in tma and wehrung @xcite .
while the first version of that negative answer was obtained for congruence lattices of lattices , it got soon extended to a much wider class of structures by using deep results in commutator theory by kearnes and szendrei @xcite .
the result can be stated as follows .
[ t : bowtie ] there exists a diagram @xmath241 , indexed by a finite poset , of finite boolean semilattices and @xmath242-embeddings , which can not be lifted , with respect to the @xmath90 functor , in any variety satisfying a nontrivial congruence lattice identity . in particular , the diagram @xmath241 can not be lifted , with respect to the @xmath90 functor , by _ lattices _ , _ majority algebras _ , _ groups _ , _ loops _ , _ modules _ , and so on . on the other hand , lampe @xcite proved that every @xmath242-semilattice ( distributive or not ) is isomorphic to @xmath243 for some _ groupoid _
@xmath244 ( a _ groupoid _ in universal algebra is just a nonempty set with a binary operation ) . by gillibert
@xcite , this result can be extended to any diagram of @xmath242-semilattices and @xmath242-homomorphisms indexed by a finite poset , and in particular to @xmath241 : that is , _
@xmath241 can be lifted by a diagram of groupoids_. we shall now describe this diagram . consider the @xmath242-homomorphisms @xmath245 , @xmath246 , and @xmath247 ( for @xmath248 ) , determined by their values on the atoms of their respective domains : @xmath249 the diagram @xmath241 is represented on figure [ fig : bowtie4 ] . @xmath250^{{\boldsymbol{u}}_0 } & { \mathfrak{p}}(3)\ar[u]|-(.45){\strut{\boldsymbol{u}}_1 } & { \mathfrak{p}}(3)\ar[lu]_{{\boldsymbol{u}}_2}\\ & & \\ { \mathfrak{p}}(2)\ar[uu]^{{\boldsymbol{f}}_0}\ar[ruu]^(.35){{\boldsymbol{f}}_0}\ar[rruu]|-(.15){{\boldsymbol{f}}_0 } & { \mathfrak{p}}(2)\ar[luu]|-(.7){\strut{\boldsymbol{f}}_1}\ar[uu]|-(.7){\strut{\boldsymbol{f}}_1 } \ar[ruu]|-(.7){\strut{\boldsymbol{f}}_1 } & { \mathfrak{p}}(2)\ar[lluu]|-(.15){{\boldsymbol{f}}_2}\ar[luu]_(.35){{\boldsymbol{f}}_2}\ar[uu]_{{\boldsymbol{f}}_2}\\ & { \mathfrak{p}}(1)\ar[lu]^{{\boldsymbol{e}}}\ar[u]|-(.45){\strut{\boldsymbol{e}}}\ar[ru]_{{\boldsymbol{e } } } & } } \ ] ] ] the underlying poset of @xmath241 is represented on figure [ fig : k33pos ] .
it does not have any lifter ( because it is _ not an almost join - semilattice _ ) .
thus cll is not sufficient to turn it into an object counterexample . in particular , it is still unknown whether every distributive @xmath19-semilattice is isomorphic to @xmath251 for some _ majority algebra
_ @xmath252 .
( a _ majority algebra _ is a nonempty set @xmath253 , endowed with a ternary operation @xmath67 , such that @xmath254 for all @xmath255 .
it is well - known that the congruence lattice of a majority algebra is distributive . )
therefore , the representation problem of distributive @xmath19-semilattices as compact congruence semilattices of majority algebras has a diagram counterexample ( namely @xmath241 ) but no known object counterexample .
it is very difficult to prove that the containment of the congruence class of all lattices into the one of all majority algebras is _ proper _ ; this is done in ploica @xcite .
the complexity of the definition of a lifter ( definition [ d : lifter ] ) makes its verification quite unpractical , even for easily described finite posets .
the following much more user - friendly variant is introduced in gillibert and wehrung @xcite .
[ not : leadsto ] for infinite cardinals @xmath6 , @xmath4 and a poset @xmath7 , let @xmath5 hold if for every mapping @xmath256^{<\lambda}$ ] there exists a one - to - one map @xmath219 such that @xmath257 as proved in gillibert and wehrung @xcite , if @xmath7 is _ lower finite _ , then it is sufficient to replace @xmath258 by @xmath259 in the statement above .
this makes the verification of the statement @xmath5 much more convenient on given finite posets .
the following result from gillibert and wehrung @xcite tells exactly when a small enough poset has a lifter ( cf . definition [ d : lifter ] ) , and gives an estimate , in terms of the @xmath5 notation , of the size of such a lifter .
[ t : charlift ] let @xmath4 and @xmath6 be infinite cardinals and let @xmath7 be a lower finite poset in which every element has less than @xmath260 upper covers
. then the following are equivalent : 1 .
@xmath7 has a @xmath261-lifter @xmath216 such that @xmath262 consists of all principal ideals of @xmath33 and @xmath263 , while @xmath33 is a lower finite almost join - semilattice .
2 . @xmath7 has a @xmath4-lifter @xmath216 such that @xmath264 .
3 . @xmath7 is a finite disjoint union of almost join - semilattices with zero , and @xmath265 .
the following definition and result can be found in gillibert and wehrung @xcite . [
d : kurind ] the _ kuratowski index _ of a finite poset @xmath7 , denoted by @xmath266 , is defined as @xmath267 if @xmath7 is pairwise incomparable , and , otherwise , as the least @xmath268 if it exists such that @xmath269 for each infinite cardinal @xmath4 .
[ p : kur(p)exists ] the kuratowski index of a finite poset @xmath7 always exists , and @xmath270 .
furthermore , if @xmath7 is a join - semilattice , then @xmath271 .
the following corollary can be found in gillibert and wehrung @xcite .
[ c : estimkur(p ) ] for every infinite cardinal @xmath4 , every nontrivial finite almost join - semilattice @xmath7 with zero has a @xmath4-lifter @xmath216 with @xmath272 .
there are finite posets for which the kuratowski index is easy to calculate , see gillibert and wehrung @xcite .
[ p : easykurp ] the following statements hold , for any finite posets @xmath7 and @xmath48 . 1 .
@xmath273 whenever @xmath7 is a nontrivial finite tree .
@xmath274 implies that @xmath275 .
3 . @xmath276 .
4 . let @xmath39 be a natural number . if @xmath7 is a product of @xmath39 nontrivial trees , then @xmath277 .
in particular , it follows from proposition [ p : easykurp](iv ) that @xmath278 for every natural number @xmath39 ; this is part of kuratowski s free set theorem ( cf .
kuratowski @xcite ) .
however , there are also easily described finite lattices for which the kuratowski index is unknown .
for example , consider the finite lattices @xmath7 and @xmath48 represented in figure [ fig : lattices ] .
both @xmath7 and @xmath48 have breadth two and order - dimension three .
as @xmath7 embeds , as a poset , into @xmath48 , we obtain the inequalities @xmath279 we do not know which of those inequalities are actually equalities . a simple infinite combinatorial statement equivalent to @xmath280 ( resp . , @xmath281 )
is given in gillibert and wehrung ( * ? ? ?
* section 6 ) .
establishing upper bounds for various critical points , due either to gillibert or to ploica , makes a heavy use of _ large free sets_. we recall the following notation , see erds _
et al . _
@xcite .
[ not : arrrel ] for cardinals @xmath6 , @xmath4 , @xmath282 , and @xmath283 , let @xmath284 hold if for every @xmath218^{\rho}\to[\kappa]^{<\lambda}$ ] , there exists @xmath285^{\mu}$ ] such that @xmath286 for each @xmath287^\rho$ ] .
for example , kuratowski s free set theorem ( cf .
kuratowski @xcite ) states that @xmath288 holds iff @xmath289 , for all infinite cardinals @xmath6 , @xmath4 and every natural number @xmath39 . for @xmath290 ,
we define the _ truncated @xmath67-dimensional cube _ @xmath291 endowed with containment .
the symbols @xmath284 and @xmath5 are related , in gillibert and wehrung @xcite , as follows .
[ p : rightarr2leadsto ] let @xmath292 and @xmath67 be integers with @xmath293 and let @xmath6 and @xmath4 be infinite cardinals
. then @xmath294 iff @xmath3 .
much is already known about the relation @xmath3 .
[ t : rightarrrem ] let @xmath4 be an infinite cardinal . 1 . if @xmath295 holds , then @xmath296 for every integer @xmath297 .
2 . @xmath298 for every integer @xmath299 .
hence @xmath300 for all @xmath301 .
3 . @xmath302 for every integer @xmath299 .
hence @xmath303 for all @xmath304 .
theorem [ t : rightarrrem](i ) is proved in erds _
et al . _
* theorem 45.5 ) .
theorem [ t : rightarrrem](ii ) is proved in hajnal and mt @xcite , see also erds _ et al .
_ ( * ? ?
* theorem 46.2 ) .
the first use of that property for estimating critical points is due to ploica @xcite .
theorem [ t : rightarrrem](iii ) is attributed , in erds _
et al . _
* theorem 46.2 ) , to hajnal .
the situation for higher truncated cubes is more strange .
set @xmath305 , @xmath306 , and , for each @xmath268 , let @xmath307 .
the latter symbol means that for every @xmath308 ^ 5\to{\{0,1\}}$ ] , either there exists @xmath309^{t_n}$ ] such that @xmath310 ^ 5={\{0\}}$ ] or there exists @xmath309 ^ 7 $ ] such that @xmath310 ^ 5={\{1\}}$ ] . the sequence @xmath311 exists , due to ramsey s theorem .
the following result is established in komjth and shelah @xcite .
[ t : kosh00 ] for every @xmath268 , there exists a generic extension of the universe in which @xmath312 .
in particular , there exists a generic extension of the universe with @xmath313 .
hence @xmath314 in any universe with @xmath295 , while @xmath315 in some generic extension .
in particular , _ the assignment @xmath316 , for a finite lattice @xmath7 , is not absolute _
( in the set - theoretical sense ) . by using propositions [ p : kur(p)exists ] , [ p : easykurp ] , and [ p : rightarr2leadsto ]
, it is possible to get new estimates for large free sets results .
the first such estimate is the following .
for any positive integer @xmath39 , it is not hard to verify that @xmath317 . by proposition [ p : kur(p)exists ]
, it follows that @xmath318 .
we obtain the following result of gillibert @xcite , see also gillibert and wehrung @xcite for a self - contained proof .
[ t : gilln+2 ] the relation @xmath319 holds for each infinite cardinal @xmath4 and each positive integer @xmath39 . in particular , @xmath320 , which answers a question raised in erds _
et al . _
the previously known bound , namely @xmath321 , follows from kuratowski s free set theorem .
the next open question is , naturally , whether @xmath322 .
this can not go on forever , as we know , from komjth and shelah s result , that @xmath323 in some generic extension .
the large free set relation @xmath320 is obtained by finding an upper bound ( in fact , here , an exact evaluation ) of the order - dimension of @xmath324 ( namely , @xmath325 ) .
now the problem of estimating the order - dimensions of truncated cubes @xmath326 is a whole topic , started by dushnik @xcite and going on in various works such as brightwell _
et al . _
@xcite , fredi and kahn @xcite , kierstead @xcite , spencer @xcite . using the various estimates available , we obtain in gillibert and wehrung @xcite new large free sets relations , of which a sample is the following . [ p : freesetsample ] 1 .
2 . @xmath328 .
3 . @xmath329 .
4 . @xmath330 .
( _ although it looks quite strange that the size of the free set is one more than a power of @xmath30 in while it is a power of two in , this seems to be due to the formulas involved having nothing to do with each other_. ) proposition [ p : freesetsample](i , ii ) uses dushnik s exact estimate from @xcite , while ( iii ) uses fredi and kahn s estimate from @xcite , and ( iv ) uses the estimate attributed in spencer @xcite to hajnal . among those estimates ,
the best asymptotic behavior is enjoyed by hajnal and spencer s one .
this estimate makes it possible to prove that @xmath331 with @xmath332 ( where @xmath333 denotes the base @xmath30 logarithm while @xmath334 denotes the natural logarithm ) .
in the present section we shall give formal definitions for some of the objects involved in section [ s : cll ] .
we restrict ourselves to infinite combinatorial objects ( _ norm - coverings _ , _ lifters _ ) and omit the more complicated categorical objects ( _ larders _ ) , for which formal definitions can be found in gillibert and wehrung @xcite .
we remind the reader that pseudo join - semilattices and supported posets are defined in section [ s : cll ] .
we start with an auxiliary definition .
[ d : normcov ] let @xmath7 be a poset . *
a _ norm - covering _ of @xmath7 consists of a pseudo join - semilattice @xmath33 together with an isotone map @xmath335 .
* an ideal @xmath336 of @xmath33 is _ sharp _ if @xmath337 has a largest element , then denoted by @xmath338 .
* then we set @xmath339 , for every set @xmath262 of sharp ideals of @xmath33 .
now we can define lifters .
[ d : lifter ] for an infinite cardinal @xmath4 , a _
@xmath4-lifter _ of a poset @xmath7 is a pair @xmath216 , where @xmath33 is a norm - covering of @xmath7 , @xmath262 is a set of sharp ideals of @xmath33 , and 1 .
every principal ideal of @xmath340 has cardinality @xmath341 .
2 . for every map @xmath342^{<\lambda}$ ]
there exists an isotone section @xmath343 of @xmath344 such that @xmath345 for all @xmath346 in @xmath7 .
3 . if @xmath347 , then @xmath33 is supported .
so far , we know that a positive solution to problem [ pb : ditorladd ] is _ consistent with @xmath169 _ , cf . theorem [ t : ex3ladd ] .
even more , every universe of @xmath169 has a ccc generic extension and a countably closed generic extension both satisfying the existence of a @xmath31-ladder of cardinality @xmath110 .
in fact , ditor asks in @xcite whether there exists a @xmath2-ladder of cardinality @xmath119 ( which is the upper bound given by kuratowski s free set theorem ) , for @xmath348 .
because of theorem [ t : ex3ladd ] , we can also ask the following weaker variant .
we ask in gillibert and wehrung @xcite whether , for any infinite cardinal @xmath4 , any poset with a @xmath4-lifter is well - founded .
equivalently , we ask whether @xmath225 has no @xmath4-lifter , for any infinite cardinal @xmath4 . as explained at the end of section [ s : cll ] , this is related to the following question .
[ pb : supererd ] does there exist an infinite cardinal @xmath6 such that for any _ isotone _ mapping @xmath218^{{\leqslant}\aleph_0}\to[\kappa]^{{\leqslant}\aleph_0}$ ] , there exists a sequence @xmath226 from @xmath6 such that @xmath227 for each @xmath228 ? recall that the isotonicity assumption simply means that @xmath349 implies that @xmath350 for any countable subsets @xmath351 and @xmath352 of @xmath6 .
it can be easily seen that if we remove this isotonicity assumption , then the corresponding problem has no solution ( cf .
the end of the proof of proposition 3 - 6.3 in gillibert and wehrung @xcite ) . fred galvin and pierre gillibert observed independently that every cardinal @xmath6 solving positively problem [ pb : supererd ] should be greater than or equal to @xmath353 . [ pb : abscrpt ] are there finitely generated varieties @xmath0 and @xmath1 of algebras , distinct natural numbers @xmath67 and @xmath39 , and models @xmath354 and @xmath355 of @xmath169 set theory , such that @xmath356 in @xmath354 while @xmath357 in @xmath355 ?
a related question is raised in gillibert and wehrung @xcite , namely whether critical points can be calculated by a recursive algorithm .
this would of course imply a negative answer to problem [ pb : abscrpt ] .
it is plausible that the construction of an example with @xmath358 would involve @xmath31-ladders ( cf .
problem [ pb : ditorladd ] ) together with some form of three - dimensional amalgamation ( see section [ s : con3 ] ) .
we remind the reader that a challenging question , raised in gillibert and wehrung ( * ? ? ?
* section 6 ) , is to determine the kuratowski indexes of the two finite lattices @xmath7 and @xmath48 represented in figure [ fig : lattices ] .
another puzzling question is the following ( cf .
notation [ not : arrrel ] ) .
an even far more ambitious question is to ask whether one can determine the largest integer @xmath67 such that @xmath169 proves the relation @xmath359 .
all we know so far is that @xmath360 ( cf .
section [ s : largefree ] ) .
this leaves plenty of room .
p. erds , a. hajnal , a. mt , and r. rado , `` combinatorial set theory : partition relations for cardinals '' .
studies in logic and the foundations of mathematics * 106*. north - holland publishing co. , amsterdam , 1984 .
isbn : 0 - 444 - 86157 - 2 g. grtzer , `` general lattice theory .
second edition '' , new appendices by the author with b.a .
davey , r. freese , b. ganter , m. greferath , p. jipsen , h.a .
priestley , h. rose , e.t .
schmidt , s.e .
schmidt , f. wehrung , and r. wille .
birkhuser verlag , basel , 1998 .
xx+663 p. isbn : 3 - 7643 - 5239 - 6 ( basel ) , 0 - 8176 - 5239 - 6 ( boston ) .
a. hajnal and a. mt , _ set mappings , partitions , and chromatic numbers_. logic colloquium 73 ( bristol , 1973 ) , p. 347379 .
studies in logic and the foundations of mathematics , vol . *
80 * , north - holland , amsterdam , 1975 .
f. wehrung , _ a non - coordinatizable sectionally complemented modular lattice with a large jnsson four - frame _ , adv .
, to appear .
available online at ` http://hal.archives-ouvertes.fr/hal-00462951 ` . | the _ critical point _ between varieties @xmath0 and @xmath1 of algebras is defined as the least cardinality of the semilattice of compact congruences of a member of @xmath0 but of no member of @xmath1 , if it exists .
the study of critical points gives rise to a whole array of problems , often involving lifting problems of either diagrams or objects , with respect to functors .
these , in turn , involve problems that belong to infinite combinatorics .
we survey some of the combinatorial problems and results thus encountered .
the corresponding problematic is articulated around the notion of a _ @xmath2-ladder _ ( for proving that a critical point is _ large _ ) , _ large free set theorems _ and the classical notation @xmath3 ( for proving that a critical point is _ small _ ) . in the middle , we find _
@xmath4-lifters _ of posets and the relation @xmath5 , for infinite cardinals @xmath6 and @xmath4 and a poset @xmath7 . |
in @xcite we introduced a new model to describe biological invasions in the plane when a strong diffusion takes place on a straight line . in this model
, we consider a coordinate system on @xmath0 with the @xmath1-axis coinciding with the line , referred to as `` the road '' .
the rest of the plane is called `` the field '' . for given time @xmath2
, we let @xmath3 denote the density of the population at the point @xmath4 of the field and @xmath5 denote the density at the point @xmath6 of the road . owing to the symmetry of the problem
, one can restrict the field to the upper half - plane @xmath7 . there
, the dynamics is assumed to be given by a standard fisher - kpp equation with diffusivity @xmath8 , whereas , on the road , there is no reproduction nor mortality and the diffusivity is given by another constant @xmath9 .
we are especially interested in the case where @xmath9 is much larger than @xmath8 . on the vicinity of the road
there is a constant exchange between the densities @xmath10 , and the one in the field adjacent to the road , @xmath11 , given by two rates @xmath12 respectively .
that is , a proportion @xmath13 of @xmath10 jumps off the road into the field while a proportion @xmath14 of @xmath11 goes onto the road .
this model gives rise to the following system : _ t u - d _
xx u= .v|_y=0-u , & x , t>0 + _ t v - dv = f(v ) , & ( x , y) , t>0 + -d_y.v|_y=0=u- .v|_y=0 , & x , t>0 , where @xmath15 are positive constants and @xmath16 satisfies the usual kpp type assumptions : @xmath17 these hypotheses will always be understood in the following without further mention .
we complete the system with initial conditions : @xmath18 where @xmath19 , @xmath20 are always assumed to be nonnegative , bounded and continuous .
the existence of a classical solution for this cauchy problem has been derived in @xcite , together with the weak and strong comparison principles .
let @xmath21 denote the kpp spreading velocity ( or invasion speed ) @xcite in the field : @xmath22 this is the asymptotic speed at which the population would spread in any direction in the open space - i.e. , when the road is not present ( see @xcite , @xcite ) .
+ the question that we treat in this paper is the following . in @xcite ( c.f .
also theorem 1.1 in @xcite ) we proved that there exists @xmath23 such that , if @xmath24 is the solution of emerging from @xmath25 , there holds that @xmath26 moreover , @xmath27 if and only if @xmath28 .
in other words , the solution spreads at velocity @xmath29 in the direction of the road .
clearly , the convergence of @xmath30 to @xmath31 in the second limit can not hold uniformly in @xmath32 .
the purpose of this paper is precisely to understand the asymptotic limits in various directions , and this turns out to be a rather delicate issue . here
is one of our main results .
[ t1.1 ] there exists @xmath33)$ ] such that @xmath34 @xmath35 locally uniformly in @xmath36 and uniformly in @xmath37 $ ] such that @xmath38 , for any given @xmath39 .
moreover , @xmath40 and , if @xmath41 , there is @xmath42 such that @xmath43 if and only if @xmath44 .
in other words , this theorem provides the spreading velocity in every direction @xmath45 , and reveals a critical angle phenomenon : the road influences the propagation on the field much further than just in the horizontal direction . in section [ sec : main ] , we state a slightly more general result , .
the paper is organised as follows . in section [ sec : main ] we state the main results and discuss them . in section [ sec : exp ] we compute the planar waves of system linearised around @xmath46 . in section
[ sec : subsolutions ] , we construct compactly supported subsolutions to , based on the already computed planar waves .
this is perhaps the most technical part of the paper , but which yields a lot of of information about the system .
the main result , that is , the asymptotic spreading velocity in every direction , is proved in section [ sec : spreading ] .
section [ sec : w * ] is devoted to further properties of the asymptotic speed in therms of the angle of the spreading directions with the road .
finally , section [ sec : t ] describes the modifications that should be made when further effects , namely transport and mortality on the road , are included . a comparison result between generalised sub and supersolutions
is given in the appendix .
we say that admits the _ asymptotic expansion shape _
@xmath47 if any solution @xmath24 emerging from a compactly supported initial datum @xmath25 satisfies > 0,_t+ _ v(x , y , t)=0 , > 0,_t+_|v(x , y , t)-1|=0 . roughly speaking
, this means that the upper level sets of @xmath30 look approximately like @xmath48 for @xmath49 large enough .
let us emphasise that the shape @xmath47 does not depend on the particular initial datum which is a strong property .
in order for conditions , in this definition to genuinely make sense ( and not be vacuously satisfied
think of the set @xmath50 ) , we further require that the asymptotic expansion shape coincides with the closure of its interior .
this condition automatically implies that the asymptotic expansion shape is unique when it exists . in the sequel
, we will sometimes consider the polar coordinate system with the angle taken with respect to the vertical axis .
namely , we will write points in the form @xmath51 .
we now state the main result of this paper .
[ thm : main ] assume the above conditions on @xmath52 .
a. _ ( spreading ) .
_ problem admits an asymptotic expansion shape @xmath47 .
b. _ ( shape of @xmath53 ) .
_ the set @xmath53 is convex and it is of the form @xmath54 here , @xmath33)$ ] , is even and there is @xmath42 such that @xmath55,\qquad w_*'>0\ \text { in } ( \vt_0,\pi/2].\ ] ] moreover , @xmath47 contains the set @xmath56\times\{0\}\big),\ ] ] and the inclusion is strict if @xmath28 . c. _ ( directions with enhanced speed ) .
_ if @xmath57 then @xmath58 . otherwise , if @xmath28 , @xmath59 .
furthermore , as functions of @xmath9 , @xmath60 is strictly decreasing for @xmath28 and @xmath61 is strictly increasing if @xmath62 .
if @xmath57 then @xmath63 , that is , the road has no effect on the asymptotic speed of spreading , in any direction , which means that the asymptotic speed coincides with the fisher -kpp invasion speed @xmath21 . on the contrary , in the case @xmath28 ,
the spreading speed is enhanced in all directions outside a cone around the normal to the road .
the closer the direction to the road , the higher the speed . of course , @xmath64 coincides with @xmath29 from .
the opening @xmath65 of this cone is explicitly given by below .
the case @xmath28 is summarized by figure [ fig : w ] .
= .5 cm ( solid line ) and @xmath66 ( dashed line ) in the case @xmath28.,height=188 ] the inclusion @xmath67 yields the following estimates on @xmath53 : @xmath68 consider now @xmath69 and @xmath29 as functions of @xmath9 , with the other parameters frozen . we know from @xcite that @xmath70 as @xmath71 .
hence , the above inequalities yield @xmath72 since @xmath73 , as it is readily seen by comparison with the tangent line @xmath74 , we have the following as functions of @xmath9 , the quantities @xmath60 and @xmath69 satisfy @xmath75,\quad\lim_{d\to\infty}w_*(\vt)= \frac{c_k}{\cos\vt}.\ ] ] that is , as @xmath76 , the set @xmath47 increases to fill up the whole strip @xmath77 .
let us give an extension of theorem [ thm : main ] . in @xcite
, we further investigated the effects of transport and reaction on the road .
this results in the two additional terms @xmath78 and @xmath79 in the first equation of .
we were able to extend the results of @xcite under a concavity assumption on @xmath52 and @xmath80 .
the additional assumption on @xmath52 is not required if @xmath80 is a pure mortality term , i.e. , @xmath81 with @xmath82 .
this is the most relevant case from the point of view of the applications to population dynamics .
the system with transport and pure mortality on the road reads _ t u - d _
u= .v|_y=0-u -u , & x , t>0 + _ t v - dv = f(v ) , & ( x , y) , t>0 + -d_y.v|_y=0=u- .v|_y=0 , & x , t>0 , with @xmath83 and @xmath82 . the first difference with is that @xmath84 is no longer a solution if @xmath85 .
however , we showed in @xcite that admits a unique positive , bounded , stationary solution @xmath86 , with @xmath87 constant and @xmath88 depending only on @xmath32 and such that @xmath89 as @xmath90 .
we then derived the existence of the asymptotic speed of spreading ( to @xmath86 ) in the direction of the line .
this is not symmetric if @xmath91 .
there are indeed two asymptotic speeds of spreading @xmath92 , in the directions @xmath93 respectively .
they satisfy @xmath94 , with strict inequality if and only if d>2+q .
the method developed in the present paper to prove can be adapted to the case of system .
the details on how this is achieved are given in section [ sec : t ] below .
in this framework , the notion of the asymptotic expansion shape is modified by replacing @xmath31 with @xmath95 in .
[ thm : maint ] for system , the following properties hold true : a. _ ( spreading )
. _ there exists an asymptotic expansion shape @xmath47 .
b. _ ( expansion shape ) . _
the set @xmath53 is convex and it is of the form @xmath96 with @xmath33)$ ] such that @xmath97,\qquad w_*'<0\ \text { in } [ -\pi/2,\vt_-),\qquad w_*'>0\ \text { in } ( \vt_+,\pi/2],\ ] ] for some critical angles @xmath98 . c. _ ( directions with enhanced speed ) . _ if does not hold then @xmath99 . otherwise , if holds , @xmath100 .
= .5 cm ) .,height=264 ] let us first comment on how the spreading velocity in the direction @xmath101 is sought for .
it will be the least @xmath102 such that the linearisation of around 0 admits solutions of the form @xmath103 with @xmath104 , @xmath105 .
let us point out that @xmath106 is not exactly a planar wave in the direction @xmath107 , for the simple reason that its level sets are not hyperplanes orthogonal to @xmath107 , but to @xmath108 .
we will find that , when @xmath28 and @xmath109 is larger than a critical angle @xmath60 , the vector @xmath108 associated with the least @xmath110 is not parallel to @xmath107 .
this is the reason why the velocity @xmath61 looks different from the classical freidlin - grtner formula @xcite , that we recall here : for a scalar equation of the form @xmath111 with @xmath112 , @xmath13 and @xmath113 1-periodic , the spreading velocity in the direction @xmath107 is given by @xmath114 where @xmath115 is the least @xmath110 such that the linearisation of around 0 : @xmath116 admits solutions of the form @xmath117 the optimal assumption for @xmath13 is not , by the way , @xmath112 .
a more general assumption is @xmath118 , where @xmath119 denotes the first periodic eigenvalue . in any case , gives the formula @xmath120 we will see in section 6 ( lemma [ lem : tangent ] below ) that a similar , but different , formula holds in our case , namely , @xmath121,\ \t\vt\in[0,\pi/2],\quad w_*(\t\vt)\leq\frac{\cos(\vt-\vp_*(\vt))}{\cos(\t\vt-\vp_*(\vt))}\,w_*(\vt).\ ] ] it will , in fact , be derived as a consequence of the expression of the spreading velocity .
several proofs of the freidlin - grtner formula have been given , besides that of @xcite .
see evans - souganidis @xcite for a viscosity solutions / singular perturbations approach , weinberger @xcite for an abstract monotone system proof ; berestycki - hamel - nadin @xcite for a pde proof .
see also @xcite for equivalent formulae and estimates of the spreading speed in periodic media , as well as @xcite for one - dimensional general media .
many of these results are explained , and developped , in @xcite .
let us now discuss the shape of the set @xmath122 in theorem [ thm : main ] , and how it compares to @xmath66 .
the latter has a very natural interpretation as the reachable set from the origin in time @xmath31 by moving with speed @xmath29 on the road and @xmath21 in the field .
indeed , considering trajectories obtained by moving on the road until time @xmath123 $ ] and then on a straight line in the field for the remaining time @xmath124 , one finds that the reachable set is the convex hull of the union of the segment @xmath125\times\{0\}$ ] and the half - disc @xmath126 , that is , @xmath66 .
another way to obtain the set @xmath66 is the following : consider a set - valued map @xmath127 and impose that the trace of @xmath128 expands at speed @xmath29 on the @xmath1-axis , and that the rest evolves by asking that the normal velocity of its boundary equals @xmath21 . in pde terms ,
@xmath129 , where @xmath130 solves the eikonal equation @xmath131}(x ) & t>0,x\in\r \end{cases}\ ] ] so , the family of sets @xmath132 is simply obtained by applying the huygens principle with the segment @xmath133 $ ] on the road as a source . in other words ,
@xmath134 and it evolves with normal velocity @xmath21 .
notice that imposing that a family of sets @xmath135 evolves with normal velocity @xmath21 forces the curvature of @xmath136 to be either @xmath137 or @xmath138 , i.e. , @xmath136 is locally either a disc of radius @xmath21 or a half - plane .
it would have been tempting to think that @xmath122 coincides with @xmath66 , just as in the singular perturbation approach to front propagation in parabolic equations or systems - see evans - souganidis @xcite , @xcite .
the fact that the asymptotic expansion shape is actually larger than this set is remarkable . and ,
as a matter of fact , we estimate in proposition [ pro : normalspeed ] below the difference between @xmath122 and @xmath66 in terms of the normal velocities of their boundaries when magnified by @xmath49 .
namely , we discover that the normal speed of @xmath139 at a boundary point @xmath140 , @xmath62 , coincides with the normal speed of the level lines of the planar wave for the linearised system which defines @xmath61 - see the next section .
this speed is larger than @xmath21 because the exponential decay rate of @xmath106 in the direction orthogonal to its level lines is less than the critical one : @xmath141 .
we expect this decay to be approximatively satisfied for large time by the solution of emerging from a compactly supported initial datum .
thus , heuristically , the presence of the road would result in an `` unnatural '' decay for solutions of the kpp equation with compactly supported initial data , which , in turns , would be the reason why @xmath47 does not coincide with the set @xmath142 following from huygens principle .
consider the linearisation of system around @xmath143 : _ t u - d _
xx u= .v|_y=0-u & x , t>0 + _ t v - dv = f(0)v & ( x , y) , t>0 + -d_y.v|_y=0=u(x , t)- .v|_y=0 & x , t>0 . take a unit vector @xmath144 , with @xmath145 . by symmetry , we restrict to @xmath146 . as said above , solutions are sought for in the form @xmath147 with @xmath148 , @xmath105 and @xmath104 ( not necessarily positive ) .
this leads to the system @xmath149 the third equation yields @xmath150 and then @xmath151 . setting @xmath152 , the system on @xmath108 reads @xmath153 the first equation in the unknown @xmath154 has the roots @xmath155 which are real if and only if @xmath156 is larger than some value @xmath157 $ ] .
the set of real solutions @xmath158 of the first equation in is then @xmath159 , with @xmath160 this is a smooth curve with leftmost point @xmath161 . rewriting the second equation in as @xmath162 , we see that it has solution if and only if @xmath163 , where @xmath164 is the invasion speed in the field . in the @xmath158 plane ,
it represents the circle @xmath165 of centre @xmath166 and radius @xmath167 given by @xmath168 let @xmath169 denote the closed set bounded from below by @xmath170 and from above by @xmath171 and let @xmath172 denote the closed disc with boundary @xmath165 .
exponential functions of the type are supersolutions of if and only if @xmath173 .
since the centre @xmath166 belongs to the line @xmath174 and the closest point of @xmath165 to the origin , @xmath175 , satisfies @xmath176 we find that @xmath177 on the other hand , @xmath178 is increasing in @xmath110 and concave in @xmath156 , the latter following from the concavity of @xmath179 .
therefore , there exists @xmath40 , depending on @xmath107 , such that @xmath180 with @xmath181 consisting in a singleton , denoted by @xmath182 , see figures [ fig : w*>ck ] and [ fig : w*=ck ] .
moreover , @xmath183 if and only if @xmath184 , namely , if and only if @xmath185 satisfies the first condition in with @xmath186 replaced by @xmath187 : @xmath188 since @xmath189 , this inequality rewrites @xmath190 the function @xmath191 defined by @xmath192 is increasing and satisfies @xmath193 , @xmath194 . as a consequence , @xmath183 if and only if either @xmath57 , or @xmath28 and @xmath195 .
observe that the sets @xmath169 shrink as @xmath9 increases and therefore @xmath69 is a strictly increasing function of @xmath9 when @xmath196 .
we now consider the critical speed @xmath69 as a function of the angle formed by the vector @xmath107 and the vertical axis .
namely , for @xmath197 $ ] , we call @xmath61 the quantity defined above associated with @xmath101
. we further let @xmath198 denote the first contact point @xmath182 . for @xmath199 ,
the above construction reduces exactly to the one of @xcite , thus @xmath200 coincides with the value @xmath29 arising in .
the function @xmath69 is even and continuous , as it is immediate to verify .
we know that if @xmath57 then @xmath201 . otherwise , if @xmath28 , @xmath43 if and only if @xmath62 , where _ 0:=(^-1(0 ) ) .
notice that @xmath60 is a decreasing function of @xmath9 .
we finally define @xmath202 the object of sections [ sec : subsolutions ] and [ sec : spreading ] is to show that @xmath53 is the asymptotic expansion set for .
this section is dedicated to the construction , for all @xmath203 , of compactly supported subsolutions moving in the direction @xmath101 with speed less than , but arbitrarily close to , @xmath61 .
we derive the following [ lem : gensub ] for all @xmath203 and @xmath39 , there exist @xmath204 and a pair @xmath205 of nonnegative functions with the following properties : @xmath206 and @xmath207 are compactly supported , ( x , y) , t0 , v(x+ct , y+ct , t)=v(x , y,0)>0 , and @xmath208 is a generalised subsolution of for all @xmath209 $ ] . by symmetry , it is sufficient to prove the lemma for @xmath210 .
the case @xmath199 was treated in @xcite . if @xmath211 $ ] then @xmath212 and the construction is standard , as we will see in section [ sec : vertical ] . in section [ sec :
horizontal ] we treat the remaining cases by exploiting the analysis of planar waves performed in the previous section .
we will proceed as follows : 1 .
we first give a definition of generalised subsolutions adapted to our context .
2 . for @xmath213
close enough to @xmath61 , we apply rouch s theorem to prove the existence of a complex exponential solution @xmath214 of the linearised system , which moves in the direction @xmath101 with speed @xmath110 .
we actually work on a perturbed system in order to get strict subsolutions of the nonlinear one .
the connected components of the positivity set of @xmath215 are bounded intervals and those of @xmath216 are infinite strips . in order to truncate those strips
, we consider the reflection @xmath217 of @xmath30 with respect to the line @xmath218 .
we then define the pair @xmath205 by setting @xmath219 in a connected component of the positivity sets of @xmath10 and @xmath220 , @xmath221 outside .
the function @xmath222 is automatically a generalised subsolution of the equation in the field .
we show that , choosing @xmath223 large enough , @xmath205 is a generalised subsolution of the equations on the road too . in the sequel , we will need to compare the solution of the cauchy problem with a pair @xmath205 which is a subsolution inside some regions , vanishes on their boundaries , and is truncated to 0 outside . in the case of a single equation ,
such type of functions are _ generalised subsolutions _
, in the sense that they satisfy the comparison principle with supersolutions .
this kind of properties has the flavour of those presented in @xcite . in the case of a system
, this property may not hold because , roughly speaking , one could truncate one component in a region where it is needed for the others to be subsolutions .
this is why we need a different notion of generalised subsolution .
we consider pairs @xmath205 such that @xmath224 is the maximum of subsolutions of the first equation in with @xmath225 , while @xmath222 is the maximum of subsolutions of the second equation and of the last equation with @xmath226 .
more precisely : [ def : gensub ] a pair @xmath205 is a _ generalised subsolution _ of if @xmath224 , @xmath222 are continuous and satisfy the following properties : a. for any @xmath6 , @xmath227 , there is a function @xmath10 such that @xmath228 in a neighbourhood of @xmath229 and , at @xmath229 ( in the classical sense ) , @xmath230 b. for any @xmath231 , @xmath227 , there is a function @xmath30 such that @xmath232 in a neighbourhood of @xmath233 and , at @xmath233 , @xmath234 although this will not be needed in the paper , we may define generalised supersolutions in analogous way , by replacing `` @xmath235 '' with `` @xmath187 '' everywhere in definition [ def : gensub ] .
this notion is stronger than that of viscosity solution ( see , e.g. , @xcite ) .
nevertheless , it recovers : ( i ) classical subsolutions , ( ii ) maxima of classical subsolutions and ( iii ) generalised subsolutions in the sense of @xcite . from now on , generalised sub and supersolutions are understood in the sense of definition [ def : gensub ] .
the comparison principle reads : [ pro : gencom ] let @xmath205 and @xmath236 be respectively a generalised subsolution bounded from above and a generalised supersolution bounded from below of such that @xmath205 is below @xmath236 at time @xmath237
. then @xmath205 is below @xmath236 for all @xmath227 .
the proof is similar to the one of proposition 3.3 in @xcite , even if the notion of sub and supersolution is slightly more general here .
it is included here in appendix [ sec : gencom ] for the sake of completeness .
let @xmath239 and @xmath240 be the principal eigenvalue and eigenfunction of the operator @xmath241 in the two dimensional ball @xmath242 , with dirichlet boundary condition .
this operator can be reduced to a self - adjoint one by multiplying the functions times @xmath243 .
one then finds that @xmath244 is equal to the of @xmath245 in @xmath242 .
whence , for @xmath246 , @xmath247 there is then @xmath248 such that @xmath249 for @xmath250 small enough , and therefore we can normalise the principal eigenfunction @xmath240 in such a way that @xmath251,\quad -d\delta(\kappa\vp)-c(\sin\vt,\cos\vt)\.\nabla(\kappa\vp)\leq f(\kappa\vp)\quad\text{in } b_r.\ ] ] it follows that the pair @xmath205 defined by @xmath252 , @xmath253 satisfies the properties stated in lemma [ lem : gensub ] .
suppose now that @xmath28 and consider @xmath254 .
call @xmath255 and , to ease notation , @xmath256 , @xmath257 , @xmath258 .
we start with the following [ lem : c - subsolution ] for @xmath259 close enough to @xmath69 , admits an exponential solution @xmath214 of the type with @xmath260 satisfying , > 0,0<<<. for @xmath261 , problem does not admit exponential solutions of the type , with @xmath262 .
however , if @xmath263 is small enough , applying the rouch theorem to the distance between @xmath264 and @xmath265 as a function of @xmath156 , one obtains an exponential solution @xmath214 with @xmath266 , depending on @xmath110 , and satisfying @xmath267 @xmath268 see the proof of lemma 6.1 in @xcite for the details . writing separately the real and complex terms of the second equation of the system satisfied by @xmath269 , we get c(_r,_r)-d(_r^2-_i^2+_r^2-_i^2)=f(0 ) + c(_i,_i)-2d(_r_i+_r_i)=0 . the first equation of yields @xmath270 + o(1)=c\xi_1\alpha_*-d\alpha_*^2+o(1),\quad\text{as } c\to w_*.\ ] ] it follows that @xmath271 in particular
, @xmath272 stays away from @xmath273 as @xmath274 . rewriting the second equation of as @xmath275
, we then infer that @xmath276 .
then , since @xmath277 too , considering the real part of , we eventually find that @xmath278 as @xmath274 .
we use again the second equation ofprecisely to derive @xmath279 the latter represents the slope of the tangent line to @xmath280 at the point @xmath182 . from the convexity of @xmath281 we know that this line intersects the @xmath154-axis at some @xmath282 .
it follows in particular that its slope is smaller than the one of the line through @xmath221 and @xmath182 .
this , in turn , is less than the slope of the line through @xmath221 and the centre of @xmath280 , which is parallel to @xmath283 , see figure [ fig : slopes ] ( a ) .
we deduce that @xmath284 this concludes the proof .
consider now the penalised system _ t u - d _
xx u=.v|_y=0-u-(u+v ) & x , t>0 + _ t v - dv=(f(0)-)v & ( x , y) , t>0 + -d_y.v|_y=0=u-.v|_y=0 -(u+v ) & x , t>0
. a small perturbation @xmath285 does not affect the qualitative results of section [ sec : exp ] are replaced by some curves converging locally uniformly to @xmath286 as @xmath287 , together with their derivatives ] nor that of lemma [ lem : c - subsolution ] .
thus , for @xmath285 small enough , there exists @xmath288 such that admits exponential solutions in the form with @xmath262 for @xmath289 , and with @xmath260 satisfying for @xmath290 close enough to @xmath288 .
moreover , @xmath291 as @xmath287 .
we are interested in the complex ones . until the end of section [ sec : subsolutions ] , @xmath214 will denote an exponential solution of , with @xmath39 sufficiently small , @xmath260 satisfying and @xmath290 close to @xmath288 . changing the sign to the imaginary part of both @xmath292 and @xmath106 we still have a solution . hence , by , it is not restrictive to assume that @xmath293 .
we set for short @xmath294 , @xmath295 , @xmath296 , @xmath297 .
since @xmath298 by the last equation of , it follows that @xmath299 . resuming
, we have : _
r,_i,_r,_i>0 , < < , ( ^-1)(0,/2 ) .
the pair @xmath24 defined by @xmath300 } \cos((\alpha_i,\beta_i)\.[(x,0)-ct\xi]),\ ] ] @xmath301 } \cos((\alpha_i,\beta_i)\.[(x , y)-ct\xi]-\arg\gamma),\ ] ] is a real solution of .
consider the following connected components of the positivity sets of @xmath10 , @xmath30 at time @xmath138 : @xmath302 as the time @xmath49 increases , these connected components are shifted , becoming @xmath303 in order to truncate the sets @xmath304 we consider the reflection with respect to the line @xmath305 , with @xmath306 , where , we recall , @xmath307 .
namely @xmath308 we then define @xmath309 the function @xmath220 vanishes on @xmath305 and satisfies the second equation of .
the quotient @xmath310 satisfies @xmath311 } } { e^{-(\alpha_r,\beta_r)\.[(x , y)-ct\xi]}}=e^{-2(\alpha_r,\beta_r)\.\xi^\perp ( l-(x , y)\.\xi^\perp)}.\ ] ] let us call @xmath312 .
it follows from that @xmath313 .
hence , 1(x , y).^l , e^-l(x , y).^l2 .
we deduce that , when restricted to the half - plane @xmath314 , a connected component of the set where @xmath315 is positive at time @xmath49 , denoted by @xmath316 , converges in hausdorff distance to @xmath304 as @xmath317 , uniformly in @xmath2 .
we can now define @xmath318 we claim that @xmath222 is bounded .
the set @xmath316 satisfies @xmath319\leq\pi+\arg\gamma\},\ ] ] as it is seen by noticing that @xmath320 on the boundary of the latter set and @xmath321 if @xmath322 .
thus @xmath323\.(\alpha_i,\beta_i)\geq-\pi+\arg\gamma } } } e^{-(\alpha_r,\beta_r)\.[(x , y)-ct\xi ] } = 2|\gamma|\sup_{{\genfrac{}{}{0pt}{}{(x , y)\.\xi^\perp\leq l}{(x , y)\.(\alpha_i,\beta_i)\geq-\pi+\arg\gamma } } } e^{-(\alpha_r,\beta_r)\.(x , y)}.\ ] ] it follows from geometrical considerations that the latter supremum is finite , c.f .
figure [ fig : slopes ] ( b ) .
analytically , one sees that it is finite if and only if @xmath324 which is equivalent to require that @xmath325 with @xmath326 .
this property holds true by .
we therefore have that @xmath205 is bounded .
furthermore , @xmath222 is a generalized subsolution of the second equation of . since @xmath327 for @xmath250
small enough , we can renormalise @xmath205 in such a way that @xmath328 is a generalized subsolution of the second equation of too , for all @xmath329 $ ] . next , like @xmath30 , @xmath217 satisfies @xmath330 and thus holds .
it only remains to show that @xmath205 is a generalized subsolution of the equations on the road in the sense of definition [ def : gensub ] .
let us write @xmath331 since @xmath299 by , we deduce that @xmath332 we further see that = ( -_ix+(^-1 ) ) .
= ( _ ix ) . for @xmath2 and @xmath333
we see that @xmath334 whence @xmath335 it follows that @xmath336 is contained in @xmath337 for @xmath223 large enough and @xmath2 .
thus , the sets @xmath338 approach @xmath339 as @xmath340 , uniformly with respect to @xmath2 .
we consider separately the two equations on the road .
below , the time @xmath2 is fixed and the expressions depending on the @xmath32-variable are always understood at @xmath341 . _ the third equation of . _
+ the condition involving the third equation of in definition [ def : gensub ] is trivially satisfied if @xmath342 . otherwise , if @xmath343 , then @xmath344 and there holds @xmath345 for some @xmath346 only depending on @xmath347 .
for @xmath223 large enough , @xmath338 is contained in @xmath348 and then yields -d_yv+vu-(u+v)+h|v|e^-l . by
, there exists @xmath349 only depending on @xmath350 and @xmath351 such that , for @xmath352 , < -k|x - b_-(t)|<,>k|x - b_+(t)|<. our aim is to show that , for @xmath353 small and @xmath223 large enough independent of @xmath49 , @xmath205 is a generalised subsolution of the last equation of for @xmath354 $ ] .
thus , up to increasing @xmath223 in such a way that @xmath355 for all @xmath2 , it is a generalised subsolution of that equation everywhere .
we first focus on a neighbourhood of @xmath356 , where @xmath357 . from , using , and recalling that @xmath358
, we obtain , for @xmath359 , @xmath360|u| .
\end{split}\ ] ] choosing then @xmath361 yields @xmath362 we eventually infer that , for @xmath223 large enough independent of @xmath49 , @xmath205 is a generalised subsolution of the last equation of in the @xmath353 neighbourhood of @xmath363 .
consider now points such that @xmath364 , where @xmath365 . by we
get @xmath366|u| , \end{split}\ ] ] provided that @xmath367 . taking @xmath368
we end up with the same inequality as in the case @xmath359 treated above .
it remains the case @xmath369 $ ] .
there we have that @xmath370 , for some @xmath371 only depending on @xmath350 , @xmath353 . consequently , using the fact that @xmath372 , we obtain @xmath373 we get again a subsolution for @xmath223 large enough . _ the second equation of . _
+ the non - trivial case is @xmath374 , where @xmath375 if @xmath376 then @xmath377 , provided that @xmath378 . as before , let @xmath349 be such that holds for @xmath352 . using and the equality
@xmath358 we get , if @xmath359 , @xmath379|u|,\ ] ] which is negative for @xmath353 small , independent of @xmath49 .
consider the remaining case @xmath380 $ ] .
there , from one hand @xmath370 with @xmath381 only depending on @xmath382 , from the other , by , @xmath383 provided that @xmath223 is large enough in such a way that @xmath384 .
hence , @xmath385 we eventually infer that , for @xmath223 large enough independent of @xmath49 , @xmath205 is a generalised subsolution of the second equation of .
this concludes the proof of lemma [ lem : gensub ] .
in this section we show that the set @xmath53 defined in section [ sec : exp ] is indeed the asymptotic expansion shape of the system .
this proves part ( i ) .
moreover , by the definition of the critical angle @xmath60 , part ( iii ) also follows .
we show separately that solutions spread at most and at least with the velocity set @xmath53 , c.f . and respectively .
the upper bound follows by comparison with the planar waves of section [ sec : exp ] .
the proof of is more involved .
it combines the convergence result close to the road given by @xcite with the existence of compactly supported subsolutions provided by lemma [ lem : gensub ] .
then one concludes using a standard liouville - type result for strictly positive solutions . throughout this section ,
@xmath24 denotes a solution of with an initial datum @xmath25 compactly supported .
as already mentioned in the introduction , the well - posedness of the cauchy problem is proved in @xcite .
we prove showing that , for any @xmath39 , there exists @xmath386 such that the following holds : @xmath387,\ c\geq w_*(\vt)+\e,\ t\geq t,\quad v(ct\sin\vt , ct\cos\vt , t)<\e.\ ] ] by symmetry , we can restrict ourselves to @xmath388 $ ] .
let @xmath248 be such that @xmath389,\qquad\supp\ , v_0\subset\ol b_r.\ ] ] for @xmath197 $ ] , let @xmath390 be the planar wave for the linearised system defined by with @xmath101 , @xmath391 , @xmath392 , @xmath393 and @xmath394 .
it is straightforward to check that the functions @xmath395 and @xmath396 are continuous , hence bounded .
since for @xmath388 $ ] it holds that @xmath397 there exists @xmath398 , independent of @xmath109 , such that all the @xmath399 are above @xmath24 at time @xmath138 .
the pairs @xmath399 are still supersolutions of , and then of because , by the kpp hypothesis , @xmath400 .
the comparison principle then yields that , for @xmath388 $ ] and @xmath2 , @xmath401 , whence , in particular , @xmath402 notice now that the functions @xmath395 and @xmath396 are strictly positive , excepted at @xmath138 where @xmath403 , @xmath404 , and at @xmath405 where @xmath406 , @xmath407 if @xmath57 .
it follows that @xmath408 is positive on @xmath409 $ ] , thus it has a positive minimum by continuity .
the result then follows .
we first show that @xmath30 is bounded from below away from @xmath138 in some suitable expanding sets .
this allows us to conclude by means of a standard liouville - type result for entire solutions with positive infimum .
_ for @xmath410 and @xmath197 $ ] , there exist @xmath411 and an open set @xmath136 in the relative topology of @xmath412 such that @xmath413 _ + consider the case @xmath414 .
let @xmath205 be a generalised subsolution given by lemma [ lem : gensub ] , with @xmath415 , and set @xmath416 even if it means multiplying @xmath224 , @xmath222 by a small factor @xmath398 , we can assume that @xmath417 , @xmath418 .
we now make use of the spreading result from @xcite , summarized here by . recalling that the @xmath29 there coincides with @xmath200 , the second limit implies the existence of @xmath419 such that , for @xmath420 $ ] and @xmath421 , the following holds true : @xmath422 then , by comparison
, @xmath423 for @xmath424 and @xmath425 , from which , taking @xmath426 and @xmath427 , where @xmath428 is such that holds , we get @xmath429t,\hat y+[c(1-\lambda)\cos\vt]t , t ) > \ul v(\hat x,\hat y,0)>0.\ ] ] namely , @xmath430 where @xmath136 is the following set : @xmath431 which is open in the relative topology of @xmath412 . by the choice of @xmath353 , restricting to the values @xmath432 and @xmath433 in the expression of @xmath136 we recover the segment @xmath434 . while , restricting to @xmath435 and @xmath436 , we obtain @xmath437\times\{0\}$ ] , which is the sought segment in the case @xmath438 .
the proof of the step 1 is thereby complete , because the minimum of @xmath30 on compact subsets of @xmath439 $ ] is positive by the strong comparison principle with @xmath221 .
_ conclusion . _
+ fix @xmath410 .
let @xmath440 be a sequence in @xmath412 and @xmath441 a sequence in @xmath442 such that @xmath443 by the boundedness of @xmath30 it follows that @xmath444 converges up to subsequences .
in order to prove we need to show that the limits of all converging subsequences are equal to 1 .
let us still call @xmath444 one of such subsequences and set @xmath445 if @xmath446 admits a bounded subsequence @xmath447 then , since @xmath448\right ) + \frac{y_{n_k}}{t_{n_k}},\ ] ] we derive @xmath449 for @xmath450 large enough .
it then follows from that @xmath451 in this case .
consider now the case where @xmath446 diverges .
let us write @xmath452 , with @xmath453 and @xmath454 , and call @xmath109 , @xmath455 the limit of ( a subsequence of ) @xmath456 , @xmath457 respectively .
the continuity of @xmath69 yields @xmath458 . consider the sequence of functions @xmath459 defined by @xmath460 for @xmath461 large enough , the @xmath462 are defined in any given @xmath463 and , by interior parabolic estimates ( see , e.g. , @xcite ) they are uniformly bounded in @xmath464 and @xmath465 with respect to the space and time variables respectively , for some @xmath466 .
hence , @xmath459 converges ( up to subsequences ) locally uniformly to a solution @xmath467 of _ t v_-dv_=f(v_),(x , y)^2 , t. moreover , @xmath468 .
consider the point @xmath469 and the set @xmath136 given by the step 1 , associated with @xmath285 and @xmath109 .
for @xmath4 and @xmath470 , we see that @xmath471 thus , for @xmath461 large enough , since @xmath472 and @xmath136 is open in @xmath412 , we have that @xmath473 , whence @xmath474 , with @xmath475 independent of @xmath233 .
it follows that @xmath476 in all @xmath477 . since @xmath478 in @xmath479 and @xmath480 in @xmath481 , it is straightforward to see by comparison with solutions of the ode @xmath482 in @xmath483 , that the unique bounded solution of which is bounded from below away from @xmath138 is @xmath484 . as a consequence , @xmath485 , which concludes the proof of
we now study the function @xmath486\to\r_+$ ] defined in section [ sec : exp ] .
this will complete the proof of part ( ii ) . since @xmath69 is
even , we restrict ourselves to @xmath409 $ ] . if @xmath57 then @xmath201 .
thus , throughout this section , we assume that @xmath28 .
we recall that @xmath198 is the unique intersection point between the sets @xmath487 and @xmath488 associated with @xmath101 .
we start with the following observation . [
lem : tangent ] the function @xmath69 satisfies @xmath121,\ \t\vt\in[0,\pi/2],\quad w_*(\t\vt)\leq\frac{\cos(\vt-\vp_*(\vt))}{\cos(\t\vt-\vp_*(\vt))}\,w_*(\vt),\ ] ] where @xmath489 .
take @xmath490 as in the statement of the lemma .
the pair @xmath214 defined by , with @xmath101 , @xmath391 , @xmath392 , @xmath393 and @xmath394 , is a solution of .
we call @xmath491 and we rewrite @xmath214 in the following way : @xmath492 thus , by the definition of @xmath493 , we derive @xmath494 the result then follows . [ pro : w * ] the function @xmath69 satisfies @xmath495),\qquad w_*=c_k\ \text { in } [ 0,\vt_0],\qquad w_*'>0\ \text { in } ( \vt_0,\pi/2].\ ] ] the fact that @xmath183 in @xmath496 $ ] is just what defines @xmath60 , see section [ sec : exp ] .
the smoothness of @xmath69 outside the point @xmath60 is an easy consequence of the implicit function theorem .
lemma [ lem : tangent ] implies that , for fixed @xmath254 , the smooth function @xmath497 touches @xmath69 from above at the point @xmath109 , whence we derive @xmath498 in particular , @xmath499 . for @xmath254 , we deduce that @xmath500 if and only if @xmath501 , which is equivalent to @xmath502 .
calling as usual @xmath503 , this inequality reads @xmath504 , which holds true by geometrical considerations , as already seen in the proof of lemma [ lem : c - subsolution ] , see figure [ fig : slopes ] ( a ) . as @xmath505 ,
the disc @xmath488 collapses to the point @xmath506 , whence @xmath507 , @xmath508 and eventually @xmath509 .
this shows that @xmath510 is continuous at @xmath60 too . to conclude the proof of part ( ii ) it remains to show that @xmath53 is convex and that @xmath511\times\{0\}\big),\ ] ] where , we recall , @xmath512 .
proposition [ pro : w * ] implies that @xmath513 is of class @xmath514 , except at the extremal points @xmath515 . the exterior unit normal to @xmath53 at those points is understood as the limit of the normals to points of @xmath516 converging to @xmath515 .
[ pro : w ] the set @xmath53 is strictly convex and , for @xmath517 $ ] , its exterior unit normal at the point @xmath518 is parallel to @xmath519 . in particular , @xmath520 .
fix @xmath521 $ ] . for @xmath522 ,
we write @xmath523 for some @xmath524 $ ] and @xmath525 .
using the inequality given by lemma [ lem : tangent ] in the form , with @xmath101 and @xmath526 , yields @xmath527 and equality holds if and only if @xmath528 .
this shows that @xmath529 is contained in the half - plane @xmath530 , except for the point @xmath518 which belongs to its boundary .
then , clearly , the same property holds for the whole @xmath53 .
this shows the convexity of @xmath47 and the directions of the normal vectors .
let us prove the last statement of the proposition .
proposition [ pro : w * ] implies that @xmath47 contains @xmath531 , whence , being convex , it contains @xmath532 . we prove that @xmath533 by showing that the ( acute ) angle @xmath534 formed by @xmath47 with the @xmath1-axis is strictly larger than the one formed by @xmath532 , which is @xmath535 . we know from the first part of the proposition that @xmath536 , where , for short , @xmath537 and @xmath538 . recall that @xmath182 is the tangent point between the sets @xmath539 and @xmath540 associated with @xmath541 , defined in section [ sec : exp ] .
it then follows from geometrical considerations that @xmath542 , see figure [ fig : phi_star ] . and
@xmath543.,height=283 ] we deduce from proposition [ pro : w ] and figure [ fig : w*>ck ] ( b ) that , for @xmath517 $ ] , the exterior normal at the point @xmath518 is steeper than @xmath544 .
let us finally estimate by how much @xmath47 is larger than @xmath545 .
[ pro : normalspeed ] the family of sets @xmath139 evolves with normal speed @xmath21 in the sector @xmath546 and with normal speed strictly larger than @xmath21 in the sectors @xmath547 . the assertion for the sector @xmath546 trivially holds because @xmath47 coincides with @xmath548 there .
consider @xmath517 $ ] and set @xmath503 . by proposition
[ pro : w ] , the exterior unit normal to @xmath47 at the point @xmath549 is @xmath550 hence , the speed of expansion of the set @xmath48 at the point @xmath551 in the normal direction @xmath552 is @xmath553 this is precisely the normal speed of the level lines of the function @xmath106 defined by with @xmath391 , @xmath392 , @xmath393 and @xmath394 .
indeed , we can rewrite @xmath554}.\ ] ] plugging the above expression in the second equation of satisfied by @xmath106 , we get @xmath555 the function @xmath556 attains its minimum @xmath21 at the unique value @xmath557 .
thus , to prove the proposition we need to show that @xmath558 .
this follows from the geometrical interpretation of the point @xmath559 , see figure [ fig : w*>ck ] ( b ) : the convexity of @xmath169 implies that the angle between the segments @xmath560 and @xmath561 , @xmath562 denoting the origin , is larger than @xmath405 , whence , since these segments have length @xmath563 and @xmath564 respectively , elementary considerations about the triangle @xmath565 show that @xmath566 .
we now describe how to modify the arguments used for problem in order to treat the case of .
this is done section by section , keeping the same notation .
_ section [ sec : exp ] .
_ + we need to consider the values @xmath567 too .
the transport and mortality terms affect through the additional term @xmath568 in the left - hand side of the first equation .
this results in the new functions @xmath569 one can readily check that @xmath178 is still increasing in @xmath110 and concave in @xmath156 . it further satisfies the following property , that will be crucial in the sequel : @xmath570
. we can therefore define @xmath69 as before .
we have that @xmath183 if and only if @xmath184 , which now reads @xmath571 this inequality can be rewritten in terms of @xmath572 as @xmath573 , with @xmath574 explicit computation shows that all the above terms are concave in @xmath575 .
hence , since @xmath576 and @xmath577 , there are two values @xmath578 such that @xmath183 if and only if @xmath579 $ ] .
we have that @xmath580 if and only if @xmath581 , which is precisely condition .
therefore , writing @xmath69 as a function of the angle @xmath109 , we derive the condition for the enhancement of the speed stated in , with @xmath582 if holds , @xmath99 otherwise . for @xmath438
, we recover the asymptotic speeds of spreading @xmath92 in the directions @xmath93 given by theorem 1.1 of @xcite .
_ section [ sec : subsolutions ] .
_ + the only point one has to check is the argument to derive in the proof of lemma [ lem : c - subsolution ] .
that argument is based on the fact that the slope of the tangent line to @xmath280 at the point @xmath182 is less than @xmath583 , which , in turn , is less than @xmath584 .
this properties follow exactly as before , from the fact that @xmath585 is concave in @xmath156 and it is nonnegative at @xmath586 . _ section [ sec : spreading ] . _
+ the proof of the upper bound works exactly as for . in the lower bound , the value @xmath31 is now replaced by the function @xmath95 .
however , since @xmath587 , we can proceed exactly as in section [ sec : lower ] , by use of the compactly supported subsolutions and the convergence result close to the road . the latter is now provided by theorem 1.1 of @xcite . the arguments in section [ sec : w * ] are unaffected by the presence of the additional terms .
following the arguments of the proof of proposition 3.2 in @xcite , we start with reducing @xmath236 to a strict supersolution @xmath588 which is strictly above @xmath205 at time @xmath138 and satisfies _ |x|u(x , t)=+ , _
|(x , y)|v(x , y , t)=+,t0 . to do this , we first multiply @xmath205 and @xmath236 by @xmath589 , where @xmath590 is the lipschitz constant of @xmath52 , and we end up with generalised sub and supersolutions ] ( still denoted @xmath205 and @xmath236 ) of the new system _ t u - d _
xx u+(+l)u=.v|_y=0 , & x , t>0 + _ t v - dv = h(t , v ) , & ( x , y) , t>0 + -d_y.v|_y=0+.v|_y=0=u , & x , t>0 , + with @xmath591 .
in such a way we gain the nonincreasing monotonicity in @xmath30 of the nonlinear term @xmath475 .
next , we introduce a nonnegative smooth function @xmath592 satisfying @xmath593,\qquad \lim_{r\to+\infty}\chi(r)=+\infty , \qquad|\chi''|\leq\delta,\ ] ] where @xmath594 will be chosen later . then , for @xmath39 , we set @xmath595 we claim that @xmath353 can be chosen small enough , independently of @xmath285 , in such a way that @xmath588 is still a generalised supersolution of , in the strict sense for the first two equations
. take @xmath596 and @xmath597 . by the definition of generalised supersolution
, there exists a function @xmath10 satisfying @xmath598 in a neighbourhood of @xmath599 and , at @xmath599 , @xmath600 the function @xmath601 satisfies @xmath602 in a neighbourhood of @xmath599 and , at @xmath599 , @xmath603 then the desired strict inequality holds provided @xmath604 . for the second equation , we start from a `` test function '' @xmath30 at some @xmath605 , @xmath597 and we see that @xmath606 satisfies , at @xmath607 , @xmath608 if @xmath609 , the right hand side is strictly larger than @xmath610 , which , in turn , is larger than @xmath611 by the monotonicity of @xmath475 . the case of the third equation
is straightforward .
the claim is thereby proved .
the pair @xmath588 is strictly above @xmath205 at @xmath237 .
assume by contradiction that @xmath588 is not strictly above @xmath612 for all time and call @xmath613,\ \ul v<\hat v\text { in } \ol\o\times[0,t]\}\,\in[0,+\infty).\ ] ] it follows that @xmath614 in @xmath615 $ ] , @xmath616 in @xmath617 $ ] .
moreover , by and the continuity of the functions we see that @xmath386 and either @xmath618 or @xmath619 vanish somewhere at time @xmath620 .
suppose that @xmath621 for some @xmath6 .
we now use the fact that @xmath205 and @xmath588 are a subsolution and a strict supersolution respectively of , in the generalised sense .
there exist @xmath622 , @xmath623 such that @xmath624 in some cylinder @xmath625 $ ] and , at @xmath626 , @xmath627 and @xmath628 since @xmath626 is a maximum point for @xmath629 in @xmath630 , we have that , there , @xmath631 and @xmath632 .
we then get a contradiction with the above strict inequality .
thus , @xmath633 and there exists @xmath231 such that @xmath634 . using the other two equations of , we find @xmath635 , @xmath636 such that @xmath637 in a cylinder @xmath638 $ ] and , at @xmath639 , @xmath640 and @xmath641 @xmath642 as before , we get a contradiction with the fact that @xmath643 has a maximum in @xmath630 at @xmath639 .
the research leading to these results has received funding from the european research council under the european union s seventh framework programme ( fp/2007 - 2013 ) / erc grant agreement n.321186 - readi -reaction - diffusion equations , propagation and modelling .
part of this work was done while henri berestycki was visiting the university of chicago .
he was also supported by an nsf frg grant dms - 1065979 .
luca rossi was partially supported by the fondazione cariparo project `` nonlinear partial differential equations : models , analysis , and control - theoretic problems '' . | we establish a new property of fisher - kpp type propagation in a plane , in the presence of a line with fast diffusion .
we prove that the line enhances the asymptotic speed of propagation in a cone of directions .
past the critical angle given by this cone , the asymptotic speed of propagation coincides with the classical fisher - kpp invasion speed .
several qualitative properties are further derived , such as the limiting behaviour when the diffusion on the line goes to infinity . |
this analysis is carried out for lhc point 5 , a minimal supergravity ( sugra ) point with @xmath2 , @xmath3 , @xmath4 , @xmath5 , and @xmath6 . for this point , @xmath7 , @xmath8 , @xmath9 , and @xmath10 .
the lhc can trivially discover susy at this point and can use precision measurements of combination of masses to determine the sugra parameters . using only such precision measurements , the estimated errors for @xmath11 of luminosity are@xcite * @xmath12 , * @xmath13 , * @xmath14 , * @xmath6 .
@xmath15 is poorly determined because the weak - scale phenomenology is insensitive to it .
the ultimate lhc precision is considerably better@xcite . given these results , one would presumably choose the energy of the fmc to be @xmath16 , the value assumed here .
susy and standard model events were generated as @xmath17 events with isajet 7.31@xcite ; @xmath18 s and @xmath19 s were interchanged in the analysis .
the toy detector simulation is the same as that used for the lhc studies@xcite except that the @xmath20 range is limited to @xmath21 , approximately equivalent to @xmath22 .
jets were found with a fixed cone algorithm , and leptons were taken from the generator .
slepton candidates are selected by requiring two @xmath19 or @xmath18 leptons with @xmath23 and @xmath24 and no other leptons or jets .
the two leptons are required to satisfy * @xmath24 , @xmath25 to select two identified leptons in the detector , * @xmath26 to reject lepton pair background , * @xmath27 to reject lepton pair and @xmath28 background , * @xmath29 to reject lepton pair and @xmath28 background .
these cuts eliminate the @xmath30 , @xmath31 , and @xmath32 backgrounds .
a @xmath33 mass cut was found to distort the @xmath34 distributions and was replaced by the cut on @xmath35 .
after these cuts , the dominant background comes from leptonic decays of @xmath36 pairs .
since @xmath37 s decay equally into @xmath38 and @xmath39 , the sm background vanishes up to statistical fluctuations in the combinations @xmath40 ( and also for @xmath41 ) .
this distribution is shown in with error bars appropriate for @xmath11 but with larger monte carlo statistics .
the origins of the signal are shown in .
there are contributions from @xmath42 , @xmath43 , and @xmath44 , the last coming from gaugino exchange in the @xmath45-channel .
two - body kinematics implies that for production of @xmath46 , @xmath47 , the maximum and minimum energies are @xmath48\ ] ] there are four distinct ranges for the lepton energy , one each for @xmath49 and @xmath50 events and two for @xmath51 events : @xmath52 if no other cuts had been made , one would obtain a sum of four square distributions with these limits .
all of these limits can be seen as edges in . the small lower limits for some of these distributions may make it difficult to identify and measure the electrons in the presence of the muon decay background .
these limits are an `` accidental '' consequence of the masses at this point ; they decrease slowly as @xmath53 is increased .
the easiest edge to detect in is the one at @xmath54 . statistically
, one could detect this with bins of @xmath55 , ten times smaller , but detector resolution and possible confusion from the two edges at about @xmath56 must be included .
we assume an error @xmath57 .
this edge is associated with the @xmath58 sleptons , and its position is given by @xmath59\ ] ] the other @xmath49 edge is at such low energy , @xmath60 , that it will probably be difficult to measure .
the lhc can measure the endpoint of the @xmath30 mass spectrum and so determine @xmath61 with an estimated error @xmath62 for @xmath11.@xcite while one can estimate the masses independently , the results are not so accurate , so we assume @xmath63 .
then the @xmath64 error ellipse in the @xmath65 plane is given by @xmath66 \delta m_i \delta m_j\ ] ] the resulting error matrix is shown in and represents a significant improvement over the lhc alone .
we next attempt to determine the @xmath67 mass by measuring the edge at @xmath68 in from @xmath69 production .
the error is probably about one bin width , @xmath70 .
unfortunately , the position of this edge , @xmath71\ ] ] turns out to be very insensitive to @xmath72 for these values of the parameters ; the derivative of the @xmath43 endpoint has a zero as a function of @xmath73 that happens to occur very close to @xmath74 for these masses .
numerically , @xmath75 as a result the sensitivity is accidentally very poor . for `` typical '' values the error on @xmath72
would be about three times that on the edge .
the second lhc point at which the fmc could contribute is lhc point 3 , a sugra point with @xmath76 , @xmath77 , @xmath78 , @xmath79 , and @xmath80 .
these parameters were chosen so that every accelerator could find something ; in particular , lep recently announced discovery of the light higgs at @xmath81 .
the lhc can make many precise measurements at this point .
however , since sleptons do not occur in the cascade decays of gluinos and squarks , they are not directly constrained .
shows the lepton energy distribution for point 3 and its sources ; compare with for point 5 .
the sleptons are nearly degenerate at this point : @xmath82 and @xmath83 .
hence all the slepton edges are nearly degenerate ; the @xmath42 contribution dominates because the branching ratio for @xmath84 is nearly 100% .
there is also a contribution to like - flavor dileptons from @xmath85 at this point , the long - dashed curve in . while the various contributions in probably can not be resolved , both the upper and the lower edges should be measurable , allowing one to determine @xmath86 and @xmath87 without additional assumptions.@xcite@xcite the resulting error ellipse is shown assuming a measurement error of @xmath88 on each edge .
clearly this case is much more favorable than that for point 5 , although in part the difference is due to the fact that there is less information on sleptons from the lhc .
while polarization is not essential to detect the signal , it would help to interpret it .
in particular , the dominance of the @xmath42 contribution seen in is due to @xmath0 exchange in the @xmath45-channel . even rather modest beam polarization would show that this contribution was dominant and so provide another test of the susy model .
the degree of polarization required needs study but is probably much smaller than that needed to suppress standard model backgrounds . | signatures for sleptons , which have been extensively studied for the next linear collider , are reexamined taking into account some of the different features of a first muon collider .
= @=11 # 1#1 # 1 16citenum : undefined argument # 1 b@#1 @=12 # 1#2 = # 1 # 1#2#3 = # 1 = # 1 # 1#2 0#1 # 1figure [ # 1 ] # 10= 0=0 1= 1=1 0>1 # 1 / = cmss10 scaled 1 to 0pt -.25 in .25into 8.9 in = 6.5 in brookhaven national laboratory ' '' '' ' '' '' to ' '' '' ' '' '' * sleptons at a first muon collider * .5 in * frank e. paige * physics department brookhaven national laboratory upton , ny 11973 usa .75 in abstract signatures for sleptons , which have been extensively studied for the next linear collider , are reexamined taking into account some of the different features of a first muon collider . to appear in _
workshop on physics at the first muon collider and at the front end of a muon collider _ , ( fermilab , november 6 9 , 1997 ) .
this manuscript has been authored under contract number de - ac02 - 76ch00016 with the u.s .
department of energy .
accordingly , the u.s .
government retains a non - exclusive , royalty - free license to publish or reproduce the published form of this contribution , or allow others to do so , for u.s .
government purposes .
supersymmetry ( susy ) signatures have been extensively studied for the next linear collider ( nlc)@xcite . the basic strategy@xcite@xcite uses the fact that susy particles are produced in pairs and decay into standard model ( sm ) particles plus an invisible lightest susy particle @xmath0 .
hence the maximum and minimum energies of the visible sm particles determine the initial susy and lsp mass .
however , the nlc studies have used properties of the nlc such as easily variable energy and high electron polarization .
this study makes assumptions appropriate for a first muon collider ( fmc ) , namely operation at a single energy with a @xmath1 hole for shielding and no polarization .
it does not , however , take into account backgrounds from muon decays . |
in the study of population dynamics , it turns out to be very useful to classify individual interactions in terms of evolutionary games @xcite . early mathematical theories of strategic interactions were based on the assumption of rational choice @xcite : an agent s optimal action depends on its expectations on the actions of others , and each of the other agents actions depend on their expectations about the focal agent . in evolutionary game theory , successful strategies spread by reproduction or imitation in a population @xcite .
evolutionary game theory not only provides a platform for explaining biological problems of frequency dependent fitness and complex individual interactions such as cooperation and coordination @xcite . in finite populations , it also links the neutral process of evolution @xcite to frequency dependence by introducing an intensity of selection @xcite .
evolutionary game theory can also be used to study cultural dynamics including human strategic behavior and updating @xcite .
one of the most interesting open questions is how do individuals update their strategies based on the knowledge and conception of others and themselves ?
two fundamentally different mechanisms can be used to classify strategy updating and population dynamics based on individuals knowledge about their strategic environment or themselves : imitation of others and self - learning based on one s own aspiration . in imitation dynamics , players update their strategies after a comparison between their own and another individual s success in the evolutionary game @xcite . for aspiration - driven updating
, players switch strategies if an aspiration level is not met , where the level of aspiration is an intrinsic property of the focal individual @xcite . in both dynamics
, novel strategies can not emerge without additional mechanisms such as spontaneous exploration of strategy space ( similar to mutation ) @xcite .
the major difference is that the latter does not require any knowledge about the payoffs of others .
thus aspiration level based dynamics , a form of self - learning , require less information about an individual s strategic environment than imitation dynamics .
aspiration - driven strategy - update dynamics are commonly observed in studies of animal and human behavioral ecology .
for example , fish would ignore social information when they have relevant personal information @xcite , and experienced ants hunt for food based on their own previous chemical trials rather than imitating others @xcite .
furthermore , a form of aspiration - level - driven dynamics play a key role in the individual behaviors in rat populations @xcite .
these examples clearly show that the idea behind aspiration dynamics , i.e. , self - evaluation , is present in the animal world . in behavioral sciences ,
such aspiration - driven strategy adjustments generally operate on the behavioral level .
however , it can be speculated that self - learning processes can have such an effect that it might actually have a downward impact on regulatory , and thus genetic levels of brain and nervous system .
this , in turn , could be seen as a mechanism that alters the rate of genetic change @xcite .
whereas such wide reaching systemic alterations are more speculative , it is clear that aspiration levels play a role in human strategy updating @xcite .
we study the statistical mechanics of a simple case of aspiration - driven self - learning dynamics in well - mixed populations of finite size .
deterministic and stochastic models of imitation dynamics have been well studied in both well - mixed and structured populations @xcite . for aspiration dynamics ,
numerous works have emerged studying population dynamics on graphs , but its impact in well - mixed populations a basic reference case , one would think is far less well understood .
although deterministic aspiration dynamics , i.e. , a kind of win - stay - lose - shift dynamics , in which individuals are perfectly rational have been analyzed @xcite , it is not clear how processes with imperfect rationality influence the evolutionary outcome . here
, we ask whether a strategy favored under pairwise comparison driven imitation dynamics can become disfavored under aspiration - driven self - learning dynamics . to this end , in our analytical analysis , we limit ourselves to the weak selection , or weak rationality approximation , where payoffs via the game play little role in the decision - making @xcite .
as it has been shown that under weak selection , the favored strategy is invariant for a wide class of imitation processes @xcite .
we show that for pairwise games , the aspiration dynamics and the imitation dynamics always share the same favored strategies . for multi - player games ,
however , the weak selection criterion under aspiration dynamics that determines whether a strategy is more abundant than the other differs from the criterion under imitation dynamics .
this paves the way to construct multi - player games , for which aspiration dynamics favor one strategy whereas imitation dynamics favor another .
furthermore , in contrast to deterministic aspiration dynamics , if the favored strategy is determined by a global aspiration level , the average abundance of a strategy in the stochastic aspiration dynamics is invariant with respect to the aspiration level , provided selection is weak .
we also extrapolate our results to stronger selection cases through numerical simulation .
we consider evolutionary game dynamics with two strategies and @xmath0 players . from these , the more widely studied @xmath1 games emerge as a special case @xcite . in individual encounters ,
players obtain their payoffs from simultaneous actions .
a focal player can be of type @xmath2 , or @xmath3 , and encounter a group containing @xmath4 other players of type @xmath2 , to receive the payoff @xmath5 , or @xmath6 .
for example , a @xmath3 player , which encounters @xmath7 individuals of type @xmath2 , obtains payoff @xmath8 .
an @xmath2 player in a group of one other @xmath2 player and thus @xmath9 @xmath3 players obtains payoff @xmath10 .
all possible payoffs of a focal individual are uniquely defined by the number of @xmath2 in the group , such that the payoff matrix reads @xmath11 for any group engaging in a one - shot game , we can obtain each member s payoff according to this matrix . in a finite well - mixed population of size @xmath12 , groups of size @xmath0 are assembled randomly , such that the probability of choosing a group that consists of another @xmath4 players of type @xmath2 , and of @xmath13 players of type @xmath3 , is given by a hypergeometric distribution @xcite .
for example , the probability that an @xmath2 player is in a group of @xmath4 other @xmath2s is given by @xmath14 , where @xmath15 ( @xmath16 ) is the number of @xmath2 players in the population , and @xmath17 is the binomial coefficient . the expected payoffs for any @xmath2 or @xmath3 in a population of size @xmath12 , with @xmath15 players of type @xmath2 and @xmath18 players of type @xmath3 , are given by @xmath19 in summary , we define a @xmath0-player stage game @xcite , shown in eq .
( [ payoff_matrix ] ) , from which the evolutionary game emerges such that each individual obtains an expected payoff based on the current composition of the well - mixed population . in the following , we introduce an update rule based on a global level of aspiration . this allows us to define a markov chain describing the inherently stochastic dynamics in a finite population : probabilistic change of the composition of the population is driven by the fact that each individual compares its actual payoff to an imaginary value that it aspires .
note here that we are only interested in the simplest way to model such a complex problem and do not address any learning process that may adjust such an aspiration level as the system evolves . for a sketch of the aspiration - driven evolutionary game , see fig . [ model ] .
players is chosen randomly from the finite population to play the game .
according to this , game players calculate and obtain their actual payoffs .
they are more likely to stochastically switch strategies if the payoffs they aspire are not met . on the other hand ,
the higher the payoffs compared to the aspiration level @xmath20 are , the less likely they switch their strategies . besides , strategy switching
is also determined by a selection intensity @xmath21 . for vanishing selection intensity ,
switching is entirely random irrespective of payoffs and the aspiration level . for increasing selection intensity
, the self - learning process becomes increasingly more `` optimal '' in the sense that for high @xmath21 , individuals tend to always switch when they are dissatisfied , and never switch when they are tolerant .
we examine the simplest possible setup , where the level of aspired payoff @xmath20 is a global parameter that does not change with the dynamics .
we show that , however , statements about the average abundance of a strategy do not depend on @xmath20 under weak selection ( @xmath22 ) .
, scaledwidth=95.0% ] in addition to the inherent stochasticity in finite populations , there is randomness in the process of individual assessments of one s own payoff as compared to a random sample of the rest of the population ; even if an individual knew exactly what to do , he might still fail to switch to an optimal strategy , e.g. , due to a trembling hand @xcite . here
we examine the simplest case of an entire population having a certain level of aspiration .
players need nt see any particular payoffs but their own , which they compare to an aspired value .
this level of aspiration , @xmath20 , is a variable that influences the stochastic strategy updating .
the probability of switching strategy is random when individuals payoffs are close to their level of aspiration , reflecting the basic degree of uncertainty in the population .
when payoffs exceed the aspiration , strategy switching is unlikely . at high values of aspiration
compared to payoffs , switching probabilities are high .
the level of aspiration provides a global benchmark of tolerance or dissatisfaction in the population .
in addition , when modeling human strategy updating , one typically introduces another global parameter that provides a measure for how important individuals deem the impact of the actual game played on their update , the intensity of selection , @xmath21 .
irrespective of the aspiration level and the frequency dependent payoff distribution , vanishing values of @xmath21 refer to nearly random strategy updating . for large values of @xmath21 ,
individuals deviations from their aspiration level have a strong impact on the dynamics .
note that although the level of aspiration is a global variable and does not differ individually , due to payoff inhomogeneity there can always be a part of the population that seeks to switch more often due to dissatisfaction with the payoff distribution . in our microscopic update process
, we randomly choose an individual , @xmath23 , from the population , and assume that the payoff of the focal individual is @xmath24 . to model stochastic self - learning of aspiration - driven switching , we can use the following probability function @xmath25 which is similar to the fermi - rule @xcite , but replaces a randomly drawn opponent s payoff by one s own aspiration .
the wider the positive gap between aspiration and payoff , the higher the switching probability .
reversely , if payoffs exceed the level of aspiration individuals become less active with increasing payoffs . the aspiration level , @xmath20 ,
provides the benchmark used to evaluate how `` greedy '' an individual is .
higher aspiration levels mean that individuals aspire to higher payoffs .
if payoffs meet aspiration , individuals remain random in their updates .
if payoffs are below aspiration , switching occurs with probability larger than random ; if they are above aspiration , switching occurs with probability lower than random .
the selection intensity governs how strict individuals are in this respect . for @xmath26 , strategy switching
is entirely random ( neutral ) .
low values of @xmath21 lead to switching only slightly different from random but follow the impact of @xmath20 . for increasing @xmath21 ,
the impact of the difference between payoffs and the aspiration becomes more important . in the case of @xmath27 , individuals are strict in the sense that they either switch strategies with probability one if they are not satisfied , or stay with their current strategy if their aspiration level is met or overshot .
the spread of successful strategies is modeled by a birth and death process in discrete time . in one time step ,
three events are possible : the abundance of @xmath2 , @xmath15 , can increase by one with probability @xmath28 , decrease by one with probability @xmath29 , or stay the same with probability @xmath30 .
all other transitions occur with probability zero .
the transition probabilities are given by @xmath31 } } , \label{tc_i+1 } \\
t_{i}^{-}&=\frac{i}{n}\ , \frac{1}{1+e^{-\omega\,[\,\alpha-\pi_{a}(i)\ , ] } } , \label{tc_i-1}\\ t_{i}^{0}&=1-t_{i}^{+}-t_{i}^{- } \label{tc_i}.\end{aligned}\ ] ] in each time step , a randomly chosen individual evaluates its success in the evolutionary game , given by eqs .
( [ payoff_c ] ) , ( [ payoff_d ] ) , compares it to the level of aspiration , and then changes strategy with probability lower than @xmath32 if its payoff exceeds the aspiration .
otherwise , it switches with probability greater than @xmath32 , except when the aspiration level is exactly met , in which case it switches randomly ( note that this is very unlikely to ever be the case ) . compared to imitation ( pairwise comparison ) dynamics , our self - learning process , which is essentially an ehrenfest - like markov chain , has some different characteristics . without the introduction of mutation or random strategy exploration ,
there exists a stationary distribution for the aspiration - driven dynamics . even in a homogenous population
, there is a positive probability that an individual can switch to another strategy due to the dissatisfaction resulting from payoff - aspiration difference .
this facilitates the escape from the states that are absorbing in the pairwise comparison process and other moran - like evolutionary dynamics .
hence there exists a nontrivial stationary distribution of the markov chain satisfying detailed balance .
specifically , for the case of @xmath26 ( neutral selection ) , the dynamics defined by eqs .
( [ tc_i+1])([tc_i ] ) are characterized by linear rates , while these rates are quadratic for the neutral imitation dynamics and moran process . in the following analysis and discussion
, we are interested in the limit of weak selection , @xmath22 , and its ability to aptly predict the success of cooperation in commonly used evolutionary @xmath33player games .
the limit of weak selection , which has a long standing history in population genetics and molecular evolution @xcite also plays a role in social learning and cultural evolution .
recent experimental results suggest that the intensity with which human subjects adjust their strategies might be low @xcite .
although it has been unclear to what degree and in what way human strategy updating deviates from random @xcite , the weak selection limit is of importance to quantitatively characterize the evolutionary dynamics . in the limiting case of weak selection , we are able to analytically classify strategies with respect to the neutral benchmark , @xmath34 @xcite .
we note that a strategy is favored by selection , if its average equilibrium frequency under weak selection is greater than one half . in order to come to such a quantitative observation
, we need to calculate the stationary distribution over frequencies of strategy @xmath2 . the markov chain given by eqs .
( [ tc_i+1])([tc_i ] ) is a one dimensional birth and death process with reflecting boundaries .
it satisfies the detailed balance condition @xmath35 , where @xmath36 is the stationary distribution over frequencies of @xmath2 in equilibrium @xcite . considering @xmath37
, we find the exact solution by recursion , given by @xmath38 where @xmath39 is the probability of successive transitions from @xmath40 to @xmath4 . the analytical solution eq .
( [ stationary distribution ] ) allows us to find the exact value of the average abundance of strategy @xmath2 , @xmath41 for any strength of selection .
it has been shown that imitation processes are similar to each other under weak selection @xcite . thus in order to compare the essential differences between imitation processes and aspiration process , we consider such selection limit . to better understand the effects of selection intensity , aspiration level , and payoff matrix on the average abundance of strategy @xmath2 ,
we further analyze which strategy is more abundant based on eq .
( [ stationary distribution ] ) . for a fixed population size , under weak selection ,
i.e. @xmath42 , the stationary distribution @xmath43 can be expressed approximately as @xmath44_{\omega=0 } , \label{stationary distribution approximation}\end{aligned}\ ] ] where the neutral stationary distribution is simply given by @xmath45 , and the first order term of this taylor expansion amounts to @xmath46_{\omega=0 } = \frac{c_{n}^{j}}{2^{n+1}}\ , \left\ { \sum\limits_{k=1}^{j}[\,\pi_{a}(k)-\pi_{b}(k-1)\ , ] -\frac{1}{2^{n}}\,\sum\limits_{k=1}^{n}\,c_{n}^{k}\sum\limits_{l=1}^{k}[\,\pi_{a}(l)-\pi_{b}(l-1)\ , ] \right\}. \label{psi'_j(0)}\end{aligned}\ ] ] interestingly , in the limiting case of weak selection , the first order approximation of the stationary distribution of @xmath2 does not depend on the aspiration level . for higher order terms of selection intensity , however
, @xmath43 does depend on the aspiration level . in the following we discuss the condition under which a strategy is favored and
compare the predictions for stationary strategy abundance under self - learning and under imitation dynamics .
thereafter we consider three prominent examples of games with multiple players through analytical , numerical and simulation methods , the results of which are detailed in figs .
[ fig : pgg][fig : sdg ] and appendix [ appendix ii ] .
all three examples are social dilemmas in the sense that the nash equilibrium of the one - shot game is not the social optimum .
first , the widely studied public goods game represents the class of games where there is only one pure nash equilibrium @xcite .
next , the public goods game with a threshold , a simplified version of the collective risk dilemma @xcite , represents the class of coordination games with multiple pure nash equilibria , depending on the threshold .
last , we consider the @xmath0-player volunteer s dilemma , or snowdrift game , which has a mixed nash equilibrium @xcite .
based on the approximation ( [ stationary distribution approximation ] ) , for any symmetric multi - player game with two strategies of normal form ( [ payoff_matrix ] ) , we can now calculate a weak selection condition such that in equilibrium @xmath2 is more abundant than @xmath3 .
since for neutrality , @xmath47 holds and thus @xmath48 , it is sufficient to consider positivity of the sum of @xmath49_{\omega=0}/n$ ] over all @xmath50 . under weak selection ,
strategy @xmath2 is favored by selection , i.e. , @xmath51 , if @xmath52 which holds for any @xmath0-player games with two strategies in a population with more than two individuals . for a detailed derivation of our main analytical result , see appendix [ appendix ] .
note that for a two - player game , @xmath53 , the above condition simplifies to @xmath54 , which is similar to the concept of risk - dominance translated to finite populations @xcite . the left hand side expression of inequality ( [ criterion condition ] )
can also be compared to a similar condition under the class of pairwise comparison processes @xcite , where two randomly selected individuals compare their payoffs and switch with a certain probability based on the observed inequality .
typically , weak selection results for pairwise comparison processes lead to the result that strategy @xmath2 is favored by selection if @xcite @xmath55 which applies both , to evaluate whether fixation of @xmath2 is more likely than fixation of @xmath3 , or whether the average abundance of @xmath2 is greater than one half under weak mutation and weak selection , that can be shown using properties of the embedded markov chain @xcite .
the sums on the left hand sides of ( [ criterion condition ] ) and ( [ pairwisecomp_condition ] ) can thus be compared with each other in order to reveal the nature of our self - learning process driven by a global aspiration level . our main result , eq .
( [ criterion condition ] ) , holds for a variety of self - learning dynamics , not only for the probability function given by eq .
( [ updating_rule ] ) . considering the general self - learning function @xmath56
$ ] with @xmath57 , here @xmath58 is strictly increasing with increasing @xmath23 . denoting @xmath59 , we have @xmath60 .
then , for @xmath42 , @xmath61 , and eq .
( [ psi_j(0 ) ] ) can be rewritten in a more general form @xmath46_{\omega=0 } = \frac{g'(0)}{g(0)}\,\frac{c_{n}^{j}}{2^{2n}}\ , \left\ { 2^{n}\,\sum_{k=1}^{j}[\,\pi_{a}(k)-\pi_{b}(k-1)\ , ] -\sum_{k=1}^{n}\,c_{n}^{k}\sum_{i=1}^{k}[\,\pi_{a}(i)-\pi_{b}(i-1)\ , ] \right\}. \label{general_psi'_j(0)}\end{aligned}\ ] ] since @xmath62 is a positive constant , eq .
( [ criterion condition ] ) is still valid for any such probability function @xmath58 , see appendix [ appendix ] .
public goods games emerge when groups of players engage in the sustenance of common goods .
cooperators @xmath2 pay an individual cost in form of a contribution @xmath63 that is pooled into the common pot .
defectors @xmath3 do not contribute .
the pot is then multiplied by a characteristic multiplication factor @xmath64 and shared equally among all individuals in the group , irrespective of contribution .
if the multiplication factor is smaller than the size of the group @xmath0 , each cooperator recovers only a fraction of the initial investment .
switching to defection would always be beneficial in a pairwise comparison of the two strategies .
the payoff matrix thus reads @xmath65 where @xmath66 is typically assumed .
since @xmath67 is a negative constant for any number of cooperators in the group , we find that @xmath68 is always negative .
cooperation can not be the more abundant strategy in the well - mixed population ( see fig .
[ fig : pgg ] ) .
however , if the self - learning dynamics are driven by a sufficiently high aspiration level , individuals are constantly dissatisfied and switch strategy frequently , even as defectors , such that cooperation can break even if selection is strong enough , namely @xmath69 for all values @xmath21 . on the other hand , if the aspiration level is low , cooperators switch more often than defectors such that the average abundance of @xmath2 assumes a value closer to the evolutionary stable state of full defection , which depends on @xmath21 . in the extreme case of very low @xmath20 and strong selection ,
defectors fully dominate , thus the stationary measure retracts to the all defection state . , population size @xmath70 , and cost of cooperation @xmath71 .
in both panels , the group sizes are @xmath72 ( dark shaded ) , and @xmath73 ( light shaded ) .
panel * a * shows the mean fraction of cooperators as a function of selection intensity for @xmath74 , the inset shows a detail for lower selection intensities .
panel * b * shows the mean fraction of cooperators as a function of selection intensity for @xmath75 .
the inset shows the stationary distribution for @xmath72 , @xmath76 and @xmath77 .
, scaledwidth=95.0% ] here we consider the following public goods game with a threshold in the sense that the good becomes strictly unavailable when the number of cooperators in a group is below a critical threshold , @xmath78 . this threshold becomes a new strategic variable .
here , @xmath63 is an initial endowment given to each player , which is invested in full by cooperators .
whatever the cooperators manage to invest is multiplied by @xmath64 and redistributed among all players in the group irrespective of strategy , if the threshold investment @xmath79 is met .
defectors do not make any investment , and thus have an additional payoff of @xmath63 , as long as the threshold is met .
once the number of cooperators is below @xmath78 , all payoffs are zero , which compares to the highest risk possible ( loss of endowment and investment with certainty ) in what is called the collective - risk dilemma @xcite .
the payoff matrix for the two strategies , cooperation @xmath2 , and defection @xmath3 , reads @xmath80 we can examine when the self - learning process favors cooperation .
we can also seek to make a statement about whether under self - learning dynamics , cooperation performs better than under pairwise comparison process .
for self - learning dynamics , we find @xmath81 while the equivalent statement for pairwise comparison processes based on the same payoff matrix would be @xmath82 .
thus , the criterion of self - learning dynamics can be written as @xmath83 whereas positivity of the imitation processes condition , @xmath84 , simply leads to @xmath85 . comparing the two conditions
, we find @xmath86 since the first factor on the right hand side of eq .
( [ tpgg_03 ] ) is always positive , the factor @xmath87 determines the relationship between self - learning dynamics and pairwise comparison processes : for sufficiently large threshold @xmath78 , expression ( [ tpgg_04 ] ) is positive . in conclusion , the aspiration - level - driven self - learning dynamics can afford to be less strict than the pairwise comparison process .
namely , it requires less reward for cooperators contribution to the common pool ( lower levels of @xmath64 ) in order to promote the cooperative strategy .
the amount of cooperative strategy depends on the threshold : higher thresholds support cooperation , even for lower multiplication factors @xmath64 ( see fig . [
fig : tpgg ] ) . for fixed @xmath64 ,
our self - learning dynamics are more likely to promote cooperation in a threshold public goods game , if the threshold for the number of cooperators needed to support the public goods is large enough , i.e. , not too different from the total size of the group . for small thresholds , and
thus higher temptation to defect in groups with less cooperators , we approach the regular public goods games , and the conclusion may be reversed . under such small @xmath78 cases , imitation - driven ( pairwise comparison ) dynamics
are more likely to lead to cooperation than aspiration dynamics . , population size @xmath70 , group size @xmath72 , and cost of cooperation @xmath71 .
in both panels , the threshold sizes are @xmath88 ( dark shaded ) , and @xmath89 ( light shaded ) .
panel * a * shows the mean fraction of cooperators as a function of selection intensity for @xmath74 .
the inset shows the critical multiplication factor above which cooperation is more abundant as a function of the threshold @xmath78 .
high thresholds lower the critical multiplication factor of the public good such that cooperation can become more abundant than defection .
panel * b * shows the mean fraction of cooperators as a function of selection intensity for @xmath90 .
the inset shows the stationary distribution for @xmath72 , @xmath90 , @xmath91 , and @xmath92 .
, scaledwidth=95.0% ] evolutionary games between two strategies can have mixed evolutionary stable states @xcite .
strategy @xmath2 can invade @xmath3 and @xmath3 can invade @xmath2 ; a stable coexistence of the two strategies typically evolves . in the replicator dynamics of the snowdrift game , cooperators can be invaded by defectors as the temptation to defect is still larger than the reward of mutual cooperation @xcite .
in contrast to the public goods game , cooperation with a group of defectors now yields a payoff greater than exclusive defection .
the act of cooperation provides a benefit to all members of the group , and the cost of cooperation is equally shared among the number of cooperators @xcite .
hence , the payoff matrix reads @xmath93 cooperation can maintain a minimal positive payoff from the cooperative act , then cooperation and defection can coexist .
the snowdrift game is a social dilemma , as selection does not favor the social optimum of exclusive cooperation .
the level of coexistence depends on the amount of cost that a particular cooperator has to contribute in a certain group . evaluating the weak selection condition , ( [ criterion condition ] ) in case of the @xmath0-player snowdrift game leads to the condition @xmath94 in order to observe @xmath95 in aspiration dynamics under weak selection . for imitation processes , on the other hand , we find @xmath96 .
note that , except for @xmath97 , @xmath98 holds for any other @xmath4 .
because of this , the different nature of these two conditions , given by the positive coefficients @xmath99 for any @xmath100 , reveals that self - learning dynamics narrow down the parameter range for which cooperation can be favored by selection . in the snowdrift game ,
self - learning dynamics are less likely to favor cooperation than pairwise comparison processes .
larger group size @xmath0 hinders cooperation : the larger the group , the higher the benefit of cooperation , @xmath101 , has to be in order to support cooperation ( see fig . [
fig : sdg ] ) . , population size @xmath70 , and cost of cooperation @xmath71 . in both panels , the group sizes are @xmath102 ( dark shaded ) , and @xmath72 ( light shaded ) .
panel * a * shows the mean fraction of cooperators as a function of selection intensity for @xmath103 . the inset shows the cooperation condition as a function of group size @xmath0 for benefits @xmath104 . only for high benefit and low group size , cooperation can be more abundant than defection .
panel * b * shows the mean fraction of cooperators as a function of selection intensity for @xmath105 .
the inset shows the stationary distribution for @xmath105 , @xmath91 , and @xmath106 .
, scaledwidth=95.0% ]
previous studies on self - learning mechanism have typically been investigated on graphs via simulations , which often employ stochastic aspiration - driven update rules @xcite .
although results based on the mean field approximations are insightful @xcite , further analytical insights have been lacking so far .
thus it is constructive to introduce and discuss a reference case of stochastic aspiration - driven dynamics of self - learning in well - mixed populations . to this end , here we introduce and discuss such an evolutionary process .
our weak selection analysis is based on a simplified scenario that implements a non - adaptive self - learning process with global aspiration level .
probabilistic evolutionary game dynamics driven by aspiration are inherently innovative and do not have absorbing boundaries even in the absence of mutation or random strategy exploration .
we study the equilibrium strategy distribution in a finite population and make a weak selection approximation for the average strategy abundance for any multi - player game with two strategies , which turns out to be independent of the level of aspiration .
this is different from the aspiration dynamics in infinitely large populations , where the evolutionary outcome crucially depends on the aspiration level @xcite .
thus it highlights the intrinsic differences arising from finite stochastic dynamics of multi - player games between two strategies .
based on this we derive a condition for one strategy to be favored over the other .
this condition then allows a comparison of a strategy s performance to other prominent game dynamics based on pairwise comparison between two strategies .
most of the complex strategic interactions in natural populations , ranging from competition and cooperation in microbial communities to social dilemmas in humans , take place in groups rather than pairs .
thus multi - player games have attracted increasing interest in different areas @xcite .
the most straightforward form of multi - player games makes use of the generalization of the payoff matrix concept @xcite .
such multi - player games are more complex and show intrinsic difference from @xmath107 games .
hence , as examples here we have studied the dynamics of one of the most widely studied multi - player games the linear public goods game @xcite , a simplified version of a threshold public goods game that requires a group of players to coordinate contributions to a public good @xcite , as well as a multi - player version of the snowdrift game @xcite where coexistence is possible .
our analytical finding allows a characterization of the evolutionary success under the stochastic aspiration - driven update rules introduced here , as well as a comparison to the well known results of pairwise comparison processes . while in coordination games , such as the threshold public goods game , the self - learning dynamics support cooperation on a larger set in parameter space ; the opposite is true for coexistence games , where the condition for cooperation to be more abundant becomes more strict
. it will be interesting to derive analytical results that either hold for any intensity of selection , or at least for the limiting case of strong selection @xcite in finite populations .
on the other hand , the update rule presented here does not seem to allow a proper continuous limit in the transition to infinitely large populations @xcite , which might give rise to interesting rescaling requirements of the demographic noise in the continuous approximation @xcite in self - learning dynamics .
our simple model illustrates that aspiration - driven self - learning dynamics in well - mixed populations alone may be sufficient to alter the expected strategy abundance . on previous studies of such processes in structured populations @xcite
, this effect might have been overshadowed by the properties of the network dynamics studied _ in silico_. our analytical results hold for weak selection , which might be a useful framework in the study of human interactions @xcite , where it is still unclear to what role model individuals compare their payoffs and with what strength players update their strategies @xcite .
although weak selection approximations are widely applied in the study of frequency dependent selection @xcite , it is not clear whether the successful spread of behavioral traits operates in this parameter regime .
thus , by numerical evaluation and simulations we show that our weak selection predictions also hold for strong selection .
models such as the one presented here may be used in attempts to predict human strategic dynamics @xcite .
such predictions , likely to be falsified in their simplicity @xcite , are essential to our fundamental understanding of complex economic and social behavior and may guide statistical insights to the effective functioning of the human mind .
this work is supported by the national natural science foundation of china ( nsfc ) under grants no .
61020106005 and no .
61375120 . b.w . gratefully acknowledges generous sponsorship from the max planck society .
greatefully acknowledges support from the deutsche akademie der naturforscher leopoldina , grant no .
lpds 2012 - 12 .
in this appendix , we detail the deducing process of the criterion of @xmath51 for general @xmath0-player game . we consider the first order approximation of stationary distribution , @xmath43 , and get the criterion condition ( shown in sec .
[ sec : res ] ) , as follows : @xmath108_{\omega=0}\,\omega>0 .
\label{a1}\end{aligned}\ ] ] inserting eq .
( [ stationary distribution ] ) , we have @xmath109 denoting @xmath110 , the above equation can be simplified as @xmath111 where @xmath112\ , ( \prod_{i=1}^{j}t_{i}^{- } ) - ( \prod_{i=0}^{j-1}t_{i}^{+})\ , [ \sum_{i=1}^{j}(t_{i}^{-})'\,(\prod_{k=1,k\neq i}^{j } t_{k}^{- } ) ] } { ( \prod_{i=1}^{j}t_{i}^{-})^{2 } } , \label{a4 } \\ \psi_{d } ' & = ( 1+\sum_{k=0}^{n-1}\frac{\prod_{i=0}^{k}t_{i}^{+}}{\prod_{i=1}^{k+1}t_{i}^{- } } ) ' = \sum_{k=0}^{n-1 } \frac { ( \prod_{i=0}^{k}t_{i}^{+})'\ , ( \prod_{i=1}^{k+1}t_{i}^{- } ) -(\prod_{i=0}^{k}t_{i}^{+})\ , ( \prod_{i=1}^{k+1}t_{i}^{- } ) ' } { ( \prod_{i=1}^{k+1}t_{i}^{-})^{2 } } \nonumber \\ & = \sum_{k=0}^{n-1 } \frac { [ \sum_{i=0}^{k}(t_{i}^{+})'\ , ( \prod_{s=0,s\neq i}^{k}t_{s}^{+})]\ , ( \prod_{i=1}^{k+1}t_{i}^{- } ) - ( \prod_{i=0}^{k}t_{i}^{+})\ , [ \sum_{i=1}^{k+1}(t_{i}^{-})'\ , ( \prod_{s=1,s\neq i}^{k+1}t_{s}^{- } ) ] } { ( \prod_{i=1}^{k+1}t_{i}^{-})^{2}}. \label{a5}\end{aligned}\ ] ] we have @xmath113}}\right\ } ' = \frac{n - i}{n}\ , \frac { \left\{e^{-\omega\,[\alpha-\pi_{b}(i)]}\right\}\ , [ \alpha-\pi_{b}(i ) ] } { \{1+e^{-\omega\,[\alpha-\pi_{b}(i)]}\}^{2 } } , \label{a6 } \\ ( t_{i}^{- } ) ' & = \frac{i}{n}\ , \left\{\frac{1}{1+e^{-\omega\,[\alpha-\pi_{a}(i)]}}\right\ } ' = \frac{i}{n}\ , \frac { \left\{e^{-\omega\,[\alpha-\pi_{a}(i)]}\right\}\ , [ \alpha-\pi_{a}(i ) ] } { \left\{1+e^{-\omega\,[\alpha-\pi_{a}(i)]}\right\}^{2}}. \label{a7}\end{aligned}\ ] ] since @xmath42 , @xmath114 , \label{a8 } \\ & \left.(t_{i}^{-})'\right|_{\omega=0}=\frac{i}{4n}\,[\alpha-\pi_{a}(i ) ] , \label{a9 } \\ & \left.(\prod_{i=0}^{j-1}t_{i}^{+})\right|_{\omega=0 } = \prod_{i=0}^{j-1}\frac{n - i}{2n } = \frac{n!}{(n - j)!\,(2n)^{j } } , \label{a10 } \\ & \left.(\prod_{i=1}^{j}t_{i}^{-})\right|_{\omega=0 } = \prod_{i=1}^{j}\frac{i}{2n } = \frac{j!}{(2n)^{j } } , \label{a11}\end{aligned}\ ] ] @xmath115\right|_{\omega=0 } = \frac{n!\,\sum_{i=0}^{j-1}[\alpha-\pi_{b}(i)]}{2\,(n - j)!\,(2n)^{j } } , \label{a12 } \\ & \left.[\sum_{i=1}^{j}(t_{i}^{-})'\,(\prod_{k=1,k\neq i}^{j } t_{k}^{-})]\right|_{\omega=0 } = \frac{j!\,\sum_{i=1}^{j}[\alpha-\pi_{a}(i)]}{2\,(2n)^{j}}. \label{a13}\end{aligned}\ ] ] then , inserting eqs .
( [ a10])([a13 ] ) into eq .
( [ a4 ] ) , @xmath116\right\}\ , \frac{j!}{(2n)^{j } } - \frac{n!}{(n - j)!\,(2n)^{j}}\ , \frac{j!}{2\,(2n)^{j}}\ , \sum_{i=1}^{j}[\alpha-\pi_{a}(i ) ] } { [ \frac{j!}{(2n)^{j}}]^{2 } } \nonumber \\ & = \frac{n!}{2\,j!\,(n - j)!}\ , [ \,-\sum_{i=0}^{j-1}\pi_{b}(i)+\sum_{i=1}^{j}\pi_{a}(i)\ , ] \nonumber\\ & = \frac{c_{n}^{j}}{2}\ , \sum_{i=1}^{j}\left[\,\pi_{a}(i)-\pi_{b}(i-1)\,\right ] .
\label{a14}\end{aligned}\ ] ] similarly , we can get @xmath117\ , \frac{(k+1)!}{(2n)^{k+1 } } } { [ \frac{(k+1)!}{(2n)^{k+1}}]^{2 } } \nonumber \\ & - \frac { \frac{n!}{(n - k-1)!\,(2n)^{k+1}}\ , \frac{(k+1)!}{2\,(2n)^{k+1}}\ , \sum_{i=1}^{k+1}[\alpha-\pi_{a}(i ) ] } { [ \frac{(k+1)!}{(2n)^{k+1}}]^{2 } } \,\ }
\nonumber\\ & = \sum_{k=1}^{n}\frac{c_{n}^{k}}{2}\ , \sum_{i=1}^{k}\left[\,\pi_{a}(i)-\pi_{b}(i-1)\,\right ] .
\label{a15}\end{aligned}\ ] ] and @xmath118 therefore , inserting eqs .
( [ a14])([a17 ] ) into eq .
( [ a3 ] ) , @xmath119_{\omega=0 } & = \frac { c_{n}^{j}\ , \{\sum_{i=1}^{j}[\pi_{a}(i)-\pi_{b}(i-1)]\}\,2^{n } } { 2\,(2^{n})^{2 } } \nonumber \\ & -\frac { c_{n}^{j}\ , \sum_{k=0}^{n-1 } \ { c_{n}^{k+1}\,\sum_{i=1}^{k+1}[\pi_{a}(i)-\pi_{b}(i-1)]\ } } { 2\,(2^{n})^{2}}. \label{a18}\end{aligned}\ ] ] combined with eq .
( [ a1 ] ) , the criterion is rewritten as @xmath120\}\ , 2^{n } } { 2\,(2^{n})^{2 } } \nonumber \\ & -\frac { c_{n}^{j}\ , \sum_{k=0}^{n-1 } \ { c_{n}^{k+1}\ , \sum_{i=1}^{k+1}[\pi_{a}(i)-\pi_{b}(i-1 ) ] \ } } { 2\,(2^{n})^{2 } } \,)>0 , \label{a19}\end{aligned}\ ] ] where @xmath121 and @xmath122 refer to eqs .
( [ payoff_c ] ) , and ( [ payoff_d ] ) .
hence , @xmath123 therefore the criterion equals to @xmath124>0 \nonumber \\ \longleftrightarrow & \frac{\omega}{2\,n\,(2^{n})^{2}}\ , [ \,\sum_{j=1}^{n}j\,c_{n}^{j}\,\sum_{i=1}^{j } \sum_{k=0}^{d-1}\frac{c_{i-1}^{k}\,c_{n - i}^{d-1-k}}{c_{n-1}^{d-1}}\,(a_{k}-b_{k})\,2^{n } \nonumber \\ -&\sum_{j=1}^{n}j\,c_{n}^{j}\,\sum_{m=1}^{n}c_{n}^{m}\,\sum_{i=1}^{m}\sum_{k=0}^{d-1}\frac{c_{i-1}^{k}\,c_{n - i}^{d-1-k}}{c_{n-1}^{d-1}}\,(a_{k}-b_{k})\,]>0 \nonumber \\ \longleftrightarrow & \frac{\omega}{4\,n\,(2^{n})}\ , [ \,\sum_{j=1}^{n}2\,j\,c_{n}^{j}\,\sum_{i=1}^{j } \sum_{k=0}^{d-1}\frac{c_{i-1}^{k}\,c_{n - i}^{d-1-k}}{c_{n-1}^{d-1}}\,(a_{k}-b_{k})\ , ] \nonumber \\ -&\frac{\omega\,n\,2^{n-1}}{2\,n\,(2^{n})^{2}}\ , [ \,\sum_{m=1}^{n}c_{n}^{m}\,\sum_{i=1}^{m}\sum_{k=0}^{d-1}\frac{c_{i-1}^{k}\,c_{n - i}^{d-1-k}}{c_{n-1}^{d-1}}\,(a_{k}-b_{k})\,]>0 \nonumber \\ \longleftrightarrow & \frac{\omega}{4\,n\,(2^{n})}\ , [ \,\sum_{j=1}^{n}(2j - n)\,c_{n}^{j}\ , \sum_{i=1}^{j } \sum_{k=0}^{d-1}\frac{c_{i-1}^{k}\,c_{n - i}^{d-1-k}}{c_{n-1}^{d-1}}\,(a_{k}-b_{k})\,]>0 . \label{a21}\end{aligned}\ ] ] we can prove that the above inequality leads to a general criterion as follows @xmath125>0 . \label{a22}\end{aligned}\ ] ] this is the result we want to show . for this
, we only need to demonstrate @xmath126 = \frac{\omega}{4\,(2^{d})}\ , \left[\,\sum_{k=0}^{d-1 } c_{d-1}^{k}\,(a_{k}-b_{k})\,\right ] .
\label{a23}\end{aligned}\ ] ] this equals to @xmath127 since such equation should hold for any choice of ( @xmath128)s , thus @xmath129 using the identity @xmath130 , we can simplify the equivalent condition as @xmath131 this can be easily proved through mathematical induction .
it is found that for the examples we discussed , namely the linear public goods game , the threshold collective risk dilemma and a multi - player snowdrift game , our result under weak selection can be generalized for a wide range of parameters ( higher values of @xmath21 , small and large populations ) . | studying strategy update rules in the framework of evolutionary game theory , one can differentiate between imitation processes and aspiration - driven dynamics . in the former case
, individuals imitate the strategy of a more successful peer . in the latter case ,
individuals adjust their strategies based on a comparison of their payoffs from the evolutionary game to a value they aspire , called the level of aspiration . unlike imitation processes of pairwise comparison , aspiration - driven updates do not require additional information about the strategic environment and can thus be interpreted as being more spontaneous .
recent work has mainly focused on understanding how aspiration dynamics alter the evolutionary outcome in structured populations .
however , the baseline case for understanding strategy selection is the well - mixed population case , which is still lacking sufficient understanding .
we explore how aspiration - driven strategy - update dynamics under imperfect rationality influence the average abundance of a strategy in multi - player evolutionary games with two strategies .
we analytically derive a condition under which a strategy is more abundant than the other in the weak selection limiting case .
this approach has a long standing history in evolutionary game and is mostly applied for its mathematical approachability .
hence , we also explore strong selection numerically , which shows that our weak selection condition is a robust predictor of the average abundance of a strategy .
the condition turns out to differ from that of a wide class of imitation dynamics , as long as the game is not dyadic .
therefore a strategy favored under imitation dynamics can be disfavored under aspiration dynamics .
this does not require any population structure thus highlights the intrinsic difference between imitation and aspiration dynamics . |
trapped cold ions provide a leading platform to implement quantum information processing .
separating ion chains is in the toolkit of basic operations required .
( merging chains is the corresponding reverse operation so we shall only refer to separation hereafter . )
it has been used to implement two - qubit quantum gates @xcite ; also to purify entangled states @xcite , or teleport material qubits @xcite . moreover ,
an architecture for processing information scalable to many ions could be developed based on shuttling , separating and merging ion crystals in multisegmented traps @xcite .
ion - chain separation is known to be a difficult operation @xcite .
experiments have progressed towards lower final excitations and shorter times but much room for improvement still remains @xcite .
problems identified include anomalous heating , so devising short - time protocols via shortcuts to adiabaticity ( sta ) techniques was proposed as a way - out worth exploring @xcite .
sta methods intend to speed up different adiabatic operations @xcite without inducing final excitations .
an example of an elementary ( fast quasi - adiabatic ) sta approach @xcite was already applied for fast chain splitting in @xcite . here , we design , using a more general and efficient sta approach based on dynamical normal modes @xcite , protocols to effectively separate two equal ions , initially in a common electrostatic linear harmonic trap , into a final configuration where each ion is in a different well .
the motion is assumed to be effectively one dimensional due to tight radial confinement .
the external potential for an ion at @xmath0 is approximated as @xmath1 which is experimentally realizable with state - of - art segmented paul traps @xcite . using dynamical normal modes ( nm ) @xcite a hamiltonian
will be set which is separable in a harmonic approximation around potential minima . by means of lewis - riesenfeld invariants
@xcite we shall design first the approximate dynamics of an unexcited splitting , taking into account anharmonicities in a perturbative manner , and from that inversely find the corresponding protocol , i.e. the @xmath2 and @xmath3 functions .
( color online ) scheme of the separation process . at @xmath4 ( left ) , both ions are trapped within the same external harmonic potential . at final time
@xmath5 ( right ) , the negative harmonic term , and a quartic term build a double well external potential .
the aim of the process is to set each of the ions in a different well without any residual excitation.,width=302 ] the hamiltonian of the system of two ions of mass @xmath6 and charge @xmath7 is , in the laboratory frame , @xmath8 where @xmath9 are the momentum operators for both ions , @xmath10 their position operators , and @xmath11 , @xmath12 being the vacuum permittivity .
we use on purpose a c - number notation since we shall also consider classical simulations .
the context will make clear if @xmath13numbers or @xmath14numbers are required .
we suppose that , due to the strong coulomb repulsion , @xmath15 . by minimizing the potential part of the hamiltonian @xmath16
, we find for the equilibrium distance between the two ions , @xmath17 , the quintic equation @xcite @xmath18 which will be quite useful for inverse engineering the ion - chain splitting , even without an explicit solution for @xmath17 . at @xmath4 a single external well
is assumed , @xmath19 and @xmath20 , whereas in the final double - well configuration @xmath21 . at some intermediate time
@xmath22 the potential becomes purely quartic ( @xmath23 ) .
our aim is to design the functions @xmath2 and @xmath3 so that each of the ions ends up in a different external well as shown in fig .
[ scheme ] , in times as short as possible , and without any final excitation .
to define dynamical nm coordinates , we calculate first at equilibrium ( the point @xmath24 in configuration space where the potential is a minimum , @xmath25 ) the matrix @xmath26 .
the equilibrium positions are @xmath27 , @xmath28 , and the matrix takes the form @xmath29 the eigenvalues are @xmath30 which define the nm frequencies as @xmath31 corresponding to center - of - mass ( @xmath32 ) and relative ( stretch ) motions ( + ) .
these relations , with eq .
( [ beta ] ) written as @xmath33 allow us to write @xmath2 and @xmath17 as functions of the nm frequencies , @xmath34{\frac{4c_c}{m\left(\omega_+^2-\omega_-^2\right)}}.\end{aligned}\ ] ] substituting these expressions into eq .
( [ beta2 ] ) , @xmath3 may also be written in terms of nm frequencies .
the normalized eigenvectors are @xmath35 which we denote as @xmath36 . the ( mass weighted ) dynamical nm coordinates are defined in terms of the laboratory coordinates as @xmath37 the unitary transformation of coordinates is @xmath38 where @xmath39\delta [ q_2-q_2({\sf q}_+,{\sf q}_-)]$ ] .
the hamiltonian in the dynamical equation for @xmath40 , where @xmath41 is the lab - frame time - dependent wave function evolving with @xmath42 , is given by @xmath43 plus qubic and higher order terms in the potential that we neglect by now ( they will be considered in sec .
iv below ) . similarly to @xcite , we apply a further unitary transformation @xmath44 , to write down an effective hamiltonian for @xmath45 with the form of two independent harmonic oscillators in nm space , @xmath46 , @xmath47 these oscillators have dynamical invariants of the form @xcite
@xmath48 ^ 2 \nonumber\\ & + & \frac{1}{2}\omega_{0\pm}^2\left(\frac{{\sf q}_\pm - x_\pm}{\rho_\pm}\right)^2 , \end{aligned}\ ] ] where the auxiliary functions @xmath49 and @xmath50 satisfy @xmath51 with @xmath52 , and , due to symmetry , @xmath53 .
the physical meaning of the auxiliary functions may be grasped from the solutions of the time - dependent schrdinger equations ( for each nm hamiltonian @xmath54 in eq .
( [ hamnm ] ) ) proportional to the invariant eigenvectors @xcite .
they form a complete basis for the space spanned by each hamiltonian @xmath54 and take the form @xmath55 } \frac{\phi_n(\sigma_\pm)}{\rho_\pm^{1/2}},\ ] ] where @xmath56 , and @xmath57 are the eigenfunctions of the static harmonic oscillator at time @xmath4 .
thus @xmath49 are scaling factors proportional to the state `` width '' in nm coordinates , whereas the @xmath58 are the dynamical - mode centers in the space of nm coordinates . within the harmonic approximation
there are dynamical states of the factorized form @xmath59 for the ion chain dynamics , where the nm wave functions @xmath60 evolve independently with @xmath54 .
they may be written as combinations of the form @xmath61 , with constant amplitudes @xmath62 .
the average energies of the @xmath63-th basis states for the two nm are @xmath64 , @xmath65
once the hamiltonian and lewis - riesenfeld invariants are defined , we proceed to apply the invariant based inverse engineering technique and design shortcuts to adiabaticity . the results for the simple harmonic oscillator in @xcite serve as a reference but have to be extended here since the two modes are not really independent from the perspective of the inverse problem .
this is because a unique protocol , i.e. , a single set of @xmath2 and @xmath3 functions has to be designed .
we first set the initial and target values for the control parameters @xmath2 and @xmath3 . at time @xmath4 , the external trap -for a single ion- is purely harmonic , with ( angular ) frequency @xmath66 . from eq .
( [ lambdaeq ] ) , we find that @xmath67 and @xmath68 .
the equilibrium distance is @xmath69{\frac{2c_c}{m\omega_0 ^ 2}}$ ] . for the final time
, we set a tenfold expansion of the equilibrium distance , @xmath70 , and @xmath71 .
this also implies @xmath72 , _
i.e. _ , the final frequencies of both nm are essentially equal , the coulomb interaction is negligible , and the ions can be considered to oscillate in independent traps .
the inverse problem is somewhat similar to the expansion of a trap with two equal ions in ref .
@xcite , but complicated by the richer structure of the external potential .
the hamiltonian ( [ hamnm ] ) and the invariant ( [ invariant ] ) must commute at both boundary times @xmath73=0 $ ] , @xmath74 to drive initial levels into final levels via a one - to - one mapping .
this is achieved by applying appropriate boundary conditions ( bc ) to the auxiliary functions @xmath49 , @xmath58 and their derivatives , @xmath75 where @xmath76 .
let us recall that @xmath53 for all times so this parameter does not have to be considered further .
inserting the bc for @xmath77 and @xmath78 in eq .
( [ auxiliarynew ] ) we find that @xmath79 .
additionally , @xmath80 is to be imposed so that @xmath81 . according to eq .
( [ distance ] ) , @xmath82 by imposing @xmath83 . with @xmath82
the hamiltonians and wave functions coincide at the boundaries , @xmath84 , @xmath85 , which simplifies the calculation of the excitation energy . from eq .
( [ auxiliaryerm ] ) , the nm frequencies may be written as @xmath86 thus the bc @xmath83 are satisfied by imposing on the auxiliary functions the additional bc @xmath87 we may now design ansatzes for the auxiliary functions @xmath49 that satisfy the ten bc in eqs .
( [ rho1],[rho2],[rho3 ] ) , plus the bc for @xmath77 and @xmath88 in eq .
( [ xbc ] ) ( since @xmath79 , @xmath89 is then automatically satisfied , see eq .
( [ auxiliarynew ] ) ) . finally , from the nm frequencies given by eq .
( [ omega ] ) we can inverse engineer the control parameters @xmath2 and @xmath3 from eqs .
( [ alpha],[distance ] ) and ( [ beta2 ] ) .
a simple choice for @xmath90 is a polynomial ansatz of 9th order @xmath91 , where @xmath92 . substituting this form in the ten bc in eqs .
( [ rho1],[rho2],[rho3 ] ) , we finally get @xmath93 for @xmath94 we will use an 11-th order polynomial @xmath95 to satisfy as well @xmath96 . the parameters @xmath97 are fixed so that the 10 bc for @xmath94 are fulfilled ( see the appendix ) , whereas @xmath98 , @xmath99 are left free , and will be numerically determined by a shooting program @xcite ( ` fminsearch ' in matlab which uses the nelder - mead simplex method for optimization ) , so that the remaining bc for @xmath77 and @xmath88 are also satisfied .
specifically , for each pair @xmath100 , @xmath101 and @xmath17 are determined from eqs .
( [ auxiliaryerm ] ) and ( [ distance ] ) , to solve eq .
( [ auxiliarynew ] ) for @xmath102 with initial conditions @xmath103 .
the free constants are changed until @xmath104 and @xmath105 are satisfied . numerically a convenient way to find the solution is to minimize the energy @xmath106 in eq .
( [ energy ] ) .
[ parameters ] ( a ) and ( b ) depict the control parameters @xmath2 and @xmath3 found with this method , using eqs .
( [ alpha ] ) and ( [ beta2 ] ) , for some value of @xmath5 and @xmath66 , see the caption , while fig .
[ parameters ] ( c ) represents the equilibrium distance between ions as a function of time ( [ distance ] ) , and fig .
[ parameters ] ( d ) the nm frequencies . in fig .
[ excitation ] ( a ) the excitation energy is shown , versus final time , for optimized parameters given in fig . [ excitation ]
the initial state is the ground state of the two ions .
it is calculated with an `` imaginary time evolution '' .
this is a variation of any numerical time evolution method ( here we used a variation of the split - operator method ) , where instead of the real time one defines imaginary time in the evolution operator . by letting an ansatz wave function
evolve for the static initial potential , it will eventually converge to the ground state of the system .
the excitation energy is @xmath107 , where @xmath108 is the final energy , calculated in the lab frame , and @xmath109 is the final ground - state energy .
the wave function evolution is calculated using the `` split - operator method '' with the full hamiltonian ( [ labhamiltonian ] ) .
if the harmonic approximation were exact , there would not be any excitation with this sta method , @xmath110 , see eq .
( [ energy ] ) .
the actual result is perturbed by the anharmonicities and nm couplings . the final ground state is also calculated with an `` imaginary time evolution '' .
the corresponding final ground state energy is essentially two times a harmonic oscillator ground state energy plus the ( negligible ) coulomb repulsion at distance @xmath111 . for the final times of all the examples ,
as it was noted in previous works @xcite , classical simulations ( solving hamilton s equations from the equilibrium configuration instead of schrdinger s equation ) give indistinguishable results in the scale of figure [ excitation ] ( a ) . the excitation energy in figure [ excitation](a ) ( solid line ) increases at short times since the harmonic nm approximation fails @xcite .
however , it goes down rapidly below one excitation quantum at times which are still rather small compared to experimental values used so far @xcite . in the following section we shall apply a perturbative technique to minimize the excitation further .
( color online ) a ) final excitation energy vs. final time using the inverse engineering design of sec .
iii ( solid blue ) , and the design that takes into account anharmonicities in sec .
iv ( dashed red ) .
b ) values of the free parameters @xmath98 ( solid blue ) and @xmath99 ( dashed red ) that minimize the excitation energy for the 11th order polynomial ( [ ansatz2 ] ) .
c ) parameters @xmath112 ( solid blue ) , @xmath113 ( dashed red ) and @xmath114 ( dash - dotted green ) that minimize the excitation energy for the 12th order polynomial ( [ ansatz3 ] ) .
two @xmath115be@xmath116 ions where split , with @xmath117 mhz.,title="fig:",width=207 ] ( color online ) a ) final excitation energy vs. final time using the inverse engineering design of sec .
iii ( solid blue ) , and the design that takes into account anharmonicities in sec .
iv ( dashed red ) .
b ) values of the free parameters @xmath98 ( solid blue ) and @xmath99 ( dashed red ) that minimize the excitation energy for the 11th order polynomial ( [ ansatz2 ] ) .
c ) parameters @xmath112 ( solid blue ) , @xmath113 ( dashed red ) and @xmath114 ( dash - dotted green ) that minimize the excitation energy for the 12th order polynomial ( [ ansatz3 ] ) .
two @xmath115be@xmath116 ions where split , with @xmath117 mhz.,title="fig:",width=207 ] ( color online ) a ) final excitation energy vs. final time using the inverse engineering design of sec .
iii ( solid blue ) , and the design that takes into account anharmonicities in sec .
iv ( dashed red ) .
b ) values of the free parameters @xmath98 ( solid blue ) and @xmath99 ( dashed red ) that minimize the excitation energy for the 11th order polynomial ( [ ansatz2 ] ) .
c ) parameters @xmath112 ( solid blue ) , @xmath113 ( dashed red ) and @xmath114 ( dash - dotted green ) that minimize the excitation energy for the 12th order polynomial ( [ ansatz3 ] ) .
two @xmath115be@xmath116 ions where split , with @xmath117 mhz.,title="fig:",width=207 ]
an improvement of the protocol is to consider the perturbation of the higher order terms neglected in the hamiltonian ( [ hamnm ] ) .
these `` anharmonicities '' @xcite are cubic and higher order terms in the taylor expansion of the coulomb term @xmath118 , @xmath119^j .
\end{aligned}\ ] ] in nm coordinates the terms take the simple form @xmath120 which may be regarded as a perturbation to be added to @xmath121 in eq .
( [ hamnm ] ) .
( the perturbation does not couple the center - of - mass and relative subspaces . ) to first order , the excess energy due to these perturbative terms at final time is given by @xmath122 where the @xmath123 are the unperturbed states in eq .
( [ expandingmodes ] ) .
inverse engineering the splitting process may now be carried out by considering a 12th order polynomial for @xmath94 ( see ( [ ansatz3 ] ) ) , with three free parameters so as to fix the bc for @xmath50 and also minimize the excitation energy . in practice
we use matlab s ` fminsearch ' function for the shooting to minimize @xmath124 as no significant improvement occurs by including higher order terms . fig .
[ excitation ] ( a ) compares the performance of such a protocol with the simpler one with the 11th - order polynomial ( [ ansatz2 ] ) .
[ excitation ] ( c ) gives the values of optimized parameters at different final times .
a large quartic potential is desirable to control the excitations produced at the point where the harmonic term changes its sign @xcite . at this point , the harmonic potential switches from confining to repulsive , which reduces the control of the system and potentially increases diabaticities and heating . in the inverse approach
proposed here there is no special design of the protocol at this point , but the method naturally seeks high quartic confinements there . in fig .
[ parameters ] ( b ) @xmath125 reaches its maximum value right at the time where @xmath126 changes sign ( see fig . [ parameters ] ( a ) ) .
however , the maximum value that @xmath125 can reach will typically be limited in a paul trap @xcite . in table
[ table ] we summarize the different maximal values of @xmath125 and critical times ( final times at which excitations below 0.1 quanta are reached ) for different values of @xmath66 using the 11-th order polynomial ( [ ansatz2 ] ) for @xmath94 .
.maximum values of @xmath125 , and critical times ( final times at which excitations below 0.1 quanta are reached ) for different values of @xmath66 .
the calculations were performed with the 11-th order polynomial for @xmath94 . [ cols="^,^,^",options="header " , ] the maximum @xmath125 decreases with @xmath5 , such that the shortest possible @xmath5 at a given maximum tolerable excitation energy is limited by the achievable @xmath125 . the trap used in ref .
@xcite yields a maximum @xmath125 of about @xmath127 n / m@xmath128 , at @xmath129 v steering range . in a recent experiment reported in @xcite , where although the purpose was not ion separation a double well potential was produced ,
the value used was @xmath130 n / m@xmath131 .
the numbers reported in the table are thus within reach , as the @xmath125 coefficients scale with the inverse 4th power of the overall trap dimension , and technological improvements on arbitrary waveform generators may allow for operation at an increased voltage range .
another potential limitation the method could encounter in the laboratory is due to biases ( a linear slope ) in the trapping potential , @xmath132 , with @xmath133 constant and unknown @xcite .
[ bias ] represents the excitation energy versus the energy difference between the two final minima of the external potential , @xmath134 ( also vs. @xmath133 ) . to calculate the results , @xmath2 and @xmath3
are designed as if @xmath135 , but the dynamics is carried out with a non - zero @xmath133 , in particular the initial state is the actual ground state , including the perturbation . note that @xmath134 should be more than a thousand vibrational quanta to excite the final energy by one quantum . in ref .
@xcite an energy increase of ten phonons at about 150 zn and 80 @xmath136s separation time was reported , so the sta ramps definitely improve the robustness against bias .
( color online ) excitation energy vs. different tilt values of the external potential in terms of the energy difference between both wells ( upper axis ) and values of the @xmath133 parameter ( lower axis ) , when using the 11-th order polynomial in the evolution .
same parameters as in fig .
[ parameters].,width=302 ] ( color online ) excitation energy vs. final time comparing the 11-th order polynomial ( solid blue ) and a non optimized trajectory experimentally used in @xcite ( dashed red ) in the evolution .
same parameters as in fig .
[ parameters].,width=302 ] further experimental limitations may be due to random fluctuations in the potential parameters , or higher order terms in the external potential .
we leave these important issues for a separate study but note that the structure of the sta techniques used here is well adapted to deal with noise or perturbations @xcite .
finally , we compare in fig .
[ comparison ] the performance of the protocols based on the polynomials ( [ ansatz1 ] ) and ( [ ansatz2 ] ) with a simple non - optimized protocol based on those experimentally used in @xcite .
there , the equilibrium distance @xmath137 is first designed as @xmath138s^2\sin(s\pi/2)$ ] , where @xmath139 . from the family of possible potential ramps consistent with this function , we chose a polynomial that drives @xmath126 from @xmath140 to @xmath141 ( as in fig . [ parameters ] ) and
whose first derivatives are 0 at both boundary times .
@xmath125 is given by eq .
( [ beta ] ) . for the times
analysed in fig .
[ comparison ] , the method based on eqs .
( [ ansatz1 ] ) and ( [ ansatz2 ] ) clearly outperforms the non - optimized ramp . to get excitations below the single motional excitation quantum with the non - optimized protocol ,
final times as long as @xmath142s would be needed , which is in line with current experiments .
we conclude that the method presented here , could bring a clear improvement with respect to the best results experimentally reported so far @xcite .
the parameters required are realistic in current trapped ions laboratories .
the simulations show that , under ideal conditions , the separation of two ions could be performed in a few oscillation periods , at times similar to those required for other operations as transport @xcite or expansions @xcite , also studied with sta .
this work has been possible thanks to the close collaboration with three ion - trap groups at boulder , mainz and zurich .
we acknowledge in particular discussions with didi leibfried , ludwig de clercq , joseba alonso , and jonathan home that provided essential elements to develop the approach .
it was supported by the basque country government ( grant no . it472 - 10 ) , ministerio de economa y competitividad ( grant no . fis2012 - 36673-c03 - 01 ) , the program ufi 11/55 of upv / ehu .
s. m .- g .
acknowledge fellowships by upv / ehu .
this research was funded by the office of the director of national intelligence ( odni ) , intelligence advanced research projects activity ( iarpa ) , through the army research office grant w911nf-10 - 1 - 0284 .
all statements of fact , opinion or conclusions contained herein are those of the authors and should not be construed as representing the official views or policies of iarpa , the odni , or the us government .
the ansatz for @xmath94 that satisfies the bc @xmath143 , @xmath144 , @xmath145 with two free parameters takes the form @xmath146 to minimize the perturbation energy in eq .
( [ perturbationenergy ] ) , three free parameters are introduced , @xmath147 99 hanneke d , home j p , jost j d , amini j m , leibfried d and wineland d j 2010 realisation of a programmable quantum information processor _ nat .
* 6 * 13 reichle r , leibfried d , knill e , britton j , blakestad r b , jost j d , langer c , ozeri r , seidelin s and wineland d j , 2006 experimental purification of two - atom entanglement _ nature ( london ) _ * 443 * 838 jost j d , home j p , amini j m , hanneke d , ozeri r , langer c , bollinger j j , leibfried d and wineland d j , 2009 entangled mechanical oscillators _ nature ( london ) _ * 459 * 683 barrett m d , chiaverini j , schaetz t , britton j , itano w m , jost j d , knill e , langer c , ozeri r and wineland d j 2004 deterministic quantum teleportation of atomic qubits _ nature ( london ) _ * 429 * 737 kielpinski d , monroe c and wineland d j 2002 architecture for a large - scale ion - trap quantum computer _ nature ( london ) _ * 417 * 709 j. p. home and a. m. steane . 2006 electrode configurations for fast separation of trapped ions _ quantum inf . comput . _
* 6 * 289 - 325 bowler r , gaebler j p , lin y , t. r. tan , hanneke d , jost j d , home j p , leibfried d and wineland d j 2012 coherent diabatic transport and separation in a multi - zone trap array _
lett . _ * 109 * 080502 ruster t , warschburger c , kaufmann h , schmiegelow c t , walther a , hettrich m , pfister a , kaushal v , schmidt - kaler f and poschinger u g 2014 experimental realization of fast ion separation in segmented paul traps _ phys . rev .
a _ * 90 * 033410 rowe m a , ben - kish a , demarco b , leibfried d , meyer v , beall j , britton j , hughes j , itano w m , jelenkovic b , langer c , rosenband t and wineland d j 2002 transport of quantum states and separation of ions in a dual rf ion trap _ quantum inf .
* 2 * , 257 ( 2002 ) .
kaufmann h , ruster t , schmiegelow c t , schmidt - kaler f and poschinger u g 2014 dynamics and control of fast ion crystal splitting in segmented paul traps _
new j. phys . _
* 16 * 073012 chen x , ruschhaupt a , schmidt s , del campo a , gu ' ery - odelin d and muga j g 2010 fast optimal frictionless atom cooling in harmonic traps : shortcut to adiabaticity _ phys .
lett . _ * 104 * 063002 torrontegui e , ibez s , martnez - garaot s , modugno m , del campo a , gu ' ery - odelin d , ruschhaupt a , chen xi and muga j g 2013 shortcuts to adiabaticity _ adv . at . mol . opt .
* 62 * 117 palmero m , bowler r , gaebler j p , leibfried d and muga j g 2014 fast transport of mixed - species ion chains within a paul trap _ phys .
a _ * 90 * 053408 palmero m , martnez - garaot s , alonso j , home j p , and muga j g fast expansions and compressions of trapped - ion chains arxiv:1502.00998 nizamani a h and hensinger w k 2012 optimum electrode configurations for fast ion separation in microfabricated surface ion traps _ appl . phys .
b _ * 106 * 327 lewis h r and riesenfeld w b , 1969 an exact quantum theory of the time - dependent harmonic oscillator and of a charged particle in a time - dependent electromagnetic field _ j. math
. phys . _ * 10 * 1458 press w h , teukolsky s a , vetterling w t , flannery b p 2007 `` section 18.1 .
the shooting method '' .
numerical recipes : the art of scientific computing ( 3rd ed . ) .
new york : cambridge university press .
isbn 978 - 0 - 521 - 88068 - 8 .
palmero m , torrontegui e , gu ' ery - odelin d and muga j g 2013 fast transport of two ions in an anharmonic trap _ phys .
a _ * 88 * 053423 morigi g and walther h 2001 two - species coulomb chains for quantum information _ eur .
j. d _ * 13 * 261 torrontegui e , chen xi , modugno m , ruschhaupt a , gu ' ery - odelin d and muga j g 2012 fast transitionless expansion of cold atoms in optical gaussian - beam traps _ phys .
a _ * 85 * 033605 torrontegui e , ibez s , chen x , ruschhaupt a , gu ' ery - odelin d and muga j g 2011 fast atomic transport without vibrational heating _ phys .
rev . a _ * 83 * 013415 wilson a c , colombe y , brown k r , knill e , leibfried d and wineland d j 2014tunable spin - spin interactions and entanglement of ions in separate potential wells _ nature _ * 512 *
, 57 - 60 ruschhaupt a , chen x , alonso d and muga j g 2012 optimally robust shortcuts to population inversion in two - level quantum systems _ new j. phys . _ * 14 * 093040 lu x j , chen x , ruschhaupt a , alonso d , gu ' erin s and muga j g 2013 fast and robust population transfer in two - level quantum systems with dephasing noise and/or systematic frequency errors _ phys .
rev . a _ * 88 * 033406 lu x j , muga j g , chen x , poschinger u g , schmidt - kaler f and ruschhaupt a 2014 fast shuttling of a trapped ion in the presence of noise _ phys . rev
. a _ * 89 * 063414 | we design fast protocols to separate or recombine two ions in a segmented paul trap . by inverse engineering the time evolution of the trapping potential composed of a harmonic and a quartic term
, it is possible to perform these processes in a few microseconds without final excitation .
these times are much shorter than the ones reported so far experimentally .
the design is based on dynamical invariants and dynamical normal modes .
anharmonicities beyond the harmonic approximation at potential minima are taken into account perturbatively .
the stability versus an unknown potential bias is also studied . |
since the discovery of asymptotic freedom in non - abelian gauge field theories , like quantum chromo dynamics ( qcd ) , many perturbative calculations have been performed to hadron - hadron and semi - leptonic processes .
the most reactions are computed up to next - to - leading order ( nlo ) only , except for some semi - leptonic processes .
the latter is due to the simplicity of the born approximation where the basic process is given by the interaction of the intermediate vector boson with a quark .
higher order corrections to the other reactions are still missing because of the very complicated feynman and phase space integrals which arise in the calculations . here
we will discuss some new methods which maybe will enable us to compute the qcd and also qed corrections beyond the nlo level .
in this section we concentrate on the computation of the coefficient functions and anomalous dimensions which show up in deep inelastic lepton - hadron scattering .
the relevant quantity in this process is the structure function defined by @xmath0 \mid p \rangle \quad q^2=-q^2 < 0 \quad x=\frac{q^2}{2\nu}\,.\end{aligned}\ ] ] in the commutator above @xmath1 represents the electro - weak current where we have suppressed lorentz indices for convenience . in the bjorken limit
i.e. @xmath2 and @xmath3 the light cone region dominates the integrand so that one can make an operator product expansion ( ope ) @xmath4 { \raisebox{-2 mm}{$\,\stackrel{=}{{\scriptstyle z^2 \sim 0 } } \ , $ } } \sum_{n } c^n(z^2
\mu^2 ) o^n ( \mu^2,0)\,.\end{aligned}\ ] ] this expansion also holds for the time ordered product corresponding to forward compton scattering @xmath5 \mid p \rangle\ , , \nonumber\\[2ex ] { \rm im}\ , t(x , q^2 ) & = & \pi \ , f(x , q^2 ) \,.\end{aligned}\ ] ]
let us assume that we can write an unsubtracted dispersion relation in the variable @xmath6 @xmath7 so that after substitution of the variables @xmath8 , @xmath9 we can derive the following relation @xmath10 & = & \sum_{n=0}^{\infty } \left ( \frac{2p\cdot q}{q^2 } \right ) ^n \int_0 ^ 1 dx ' \ , x'^{n-1 } f(x',q^2 ) \nonumber\\[2ex ] & = & \sum_{n=0}^{\infty } \left ( \frac{2p\cdot q}{q^2 } \right ) ^n a^{(n)}(\mu^2 ) { \cal c}^{(n ) } \left ( \frac{q^2 } { \mu^2 } \right ) \,,\end{aligned}\ ] ] where the operator matrix element and the coefficient function are given by @xmath11 and @xmath12 respectively .
both quantities satisfy a renormalization group equation @xmath13 a^{(n)}(\mu^2)=0\ , , \nonumber\\[2ex ] & & \left [ \mu \frac{\partial}{\partial \mu } + \beta ( g ) \frac{\partial } { \partial g } - \gamma^{(n)}(g ) \right ] { \cal c}^{(n)}\left(\frac{q^2}{\mu^2 } \right ) = 0\,.\end{aligned}\ ] ] in the equations above the quantities @xmath14 and @xmath15 represent the beta - function and the anomalous dimension respectively .
the latter determine the @xmath16-evolution of the structure function @xmath17 which can be measured in experiment and provides us with one of the tests of perturbative qcd .
if one computes the coefficient functions in the conventional way one encounters phase space integrals and loop integrals ( see e.g. @xcite ) .
however if one wants to compute the coefficient function beyond order @xmath18 it is more convenient to try another method which is explained in @xcite .
taking @xmath19-theory as an example one can expand the propagator in @xmath20 ( [ eqn2.3 ] ) given by @xmath21 as follows @xmath22 & = & -ig^2 \sum_{n=0}^{\infty } \int \frac{d^nk}{(2\pi)^n } \frac{(2k \cdot p)^n } { ( k^2)^{3+n}(k+q)^2 } \,,\end{aligned}\ ] ] ( 260,100)(0,0 ) ( 0,0)(20,20)25 ( 20,20)(20,80)3 ( 0,100)(20,80 ) ( 20,80)(60,80 ) ( 60,80)(100,80)3 ( 100,80)(100,20)3 ( 100,20)(120,0 ) ( 60,80)(60,20 ) ( 60,20)(20,20)3 ( 100,20)(60,20 ) ( 100,80)(120,100)25 ( 15,100)[t]@xmath23 ( 105,100)[t]@xmath24 ( 45,95)[t]@xmath25 ( 85,95)[t]@xmath26 ( 10,55)[t]@xmath27 ( 110,55)[t]@xmath28 ( 45,15)[t]@xmath29 ( 85,15)[t]@xmath30 ( 15,10)[t]@xmath24 ( 105,10)[t]@xmath23 ( 60,50)[t]@xmath31 ( 130,50)[t]@xmath32 ( 140,50)(165,50)25 ( 235,50)(260,50)25 ( 200,50)(35,180,90)3 ( 200,50)(35,90,0)3 ( 200,50)(35,270,180)3 ( 200,50)(35,360,270)3 ( 200,85)(200,15)3 ( 168,85)[t]@xmath27 ( 240,85)[t]@xmath33 ( 165,25)[t]@xmath29 ( 150,45)[t]@xmath24 ( 250,45)[t]@xmath24 ( 235,25)[t]@xmath28 ( 205,65)[t]@xmath34 where we have used n - dimensional regularization to regularize the collinear ( c ) and ultraviolet ( uv ) singularities . the asymptotic expansion of the propagator converts the compton amplitude into a self energy type of integral . the latter is much easier to compute than expressions for box- or triangle graphs .
another feature is that this asymptotic expansion transforms the c - divergence in @xmath35 at @xmath36 into an uv singularity appearing in the expression for the self energy integral .
therefore the operator matrix element in eq .
( [ eqn2.5 ] ) , containing the collinear divergence denoted by @xmath37 , will be replaced by the corresponding operator renormalization constant @xmath38 @xmath39 & & \rightarrow \quad \sum_{n=0}^{\infty } \left ( \frac{2p\cdot q}{q^2 } \right ) ^n z_{o^n } \left ( \frac{1}{\varepsilon_{uv}},\mu^2 \right ) { \cal c}^{(n ) } \left ( \frac{q^2}{\mu^2}\right ) \,,\end{aligned}\ ] ] where @xmath38 depends on the uv singularity denoted by @xmath40 .
this transformation leaves the coefficient function unaltered .
a straightforward calculation yields @xmath41 one can expand the gamma - functions above around @xmath42 and obtain @xmath43 \big [ \frac{2}{\varepsilon } + \gamma_e - \ln 4\pi \nonumber\\[2ex ] & & + \ln \frac { q^2}{\mu^2 } - 1 - \sum_{k=1}^n \frac{1}{k } \big ] \,.\end{aligned}\ ] ] in order @xmath44 eq .
( [ eqn2.10 ] ) can be expressed as @xmath45\,,\end{aligned}\ ] ] so that one can read @xmath46 and @xmath47 . if the method is extended up to order @xmath18 one encounters graphs where at least three propagators , carrying the momentum @xmath23 ( see the solid line in fig .
( [ fig1 ] ) ) , have to be expanded ( 240,160)(0,0 ) ( 20,120)(30,120 ) ( 70,120)(80,120 ) ( 50,140)(50,100 ) ( 50,120)(20,0,360 ) ( 100,120)(110,120 ) ( 150,120)(160,120 ) ( 150,140)(21,180,270 ) ( 130,120)(20,0,360 ) ( 180,120)(190,120 ) ( 250,120)(260,120 ) ( 205,120)(15,0,360 ) ( 235,120)(15,0,360 ) ( 60,40)(70,40 ) ( 110,40)(120,40 ) ( 90,40)(20,0,360 ) ( 110,19)(21,90,180 ) ( 140,40)(150,40 ) ( 210,40)(220,40 ) ( 165,40)(15,0,360 ) ( 195,40)(15,0,360 ) ( 0,125)[t]@xmath48 ( 130,105)[t]@xmath49 ( 207,110)[t]@xmath49 ( 90,65)[t]@xmath49 ( 165,60)[t]@xmath49 this procedure leads to triple sums which are very difficult to unravel @xmath50 to avoid this one has to remove one or more propagators , carrying the momentum @xmath23 , before doing the expansion . first one can adopt the method 1 .
integration by parts + this trick has been successfully applied to compute self energy graphs @xcite .
an example is @xmath51 the integral above appears in the partial integration w.r.t .
the momentum @xmath27 of the expression below ( see fig .
( [ fig2 ] ) ) @xmath52 hence @xmath53 in eq .
( [ eqn2.15 ] ) can be reduced to a set of integrals each containing one propagator less than in the original expression .
recursion relations ( difference equations ) + another method is developed in @xcite .
using the method of `` difference equations '' one is able to knock out one propagator ( see the example in fig .
( [ fig3 ] ) ) .
one can show that all graphs can be reduced to two building blocks only ( see fig .
( [ fig4 ] ) ) .
the latter contain one propagator carrying the momentum @xmath23 so that one only obtains a single sum which is easy to perform . at this moment
it is not clear whether the method can be extended beyond order @xmath18 ( two - loop ) .
( 285,70)(0,0 ) ( 0,65)(5,55 ) ( 5,55)(35,55 ) ( 35,55)(65,55)3 ( 65,55)(70,65)23 ( 5,55)(5,15)3 ( 0,5)(5,15)23 ( 5,15)(35,15)3 ( 65,55)(65,15)3 ( 35,55)(35,15 ) ( 35,15)(65,15 ) ( 65,15)(70,5 ) ( 90,40)[t]@xmath54 ( 0,60)[t]@xmath23 ( 0,20)[t]@xmath24 ( 73,60)[t]@xmath24 ( 73,18)[t]@xmath23 ( 105,65)(110,55 ) ( 110,55)(140,55 ) ( 140,55)(170,55)3 ( 170,55)(175,65)23 ( 110,55)(110,15)3 ( 105,5)(110,15)23 ( 110,15)(140,15)3 ( 170,55)(170,15)3 ( 140,55)(140,15 ) ( 140,15)(170,15 ) ( 170,15)(175,5 ) ( 194,40)[t]@xmath55 ( 210,65)(215,55 ) ( 215,55)(245,55 ) ( 245,55)(275,55)3 ( 275,55)(285,65)23 ( 215,55)(215,15)3 ( 210,5)(215,15)23 ( 215,15)(245,15)3 ( 245,55)(245,15 ) ( 245,15)(275,55)3 ( 240,70)(0,0 ) ( 0,65)(5,55 ) ( 5,55)(35,55 ) ( 35,55)(65,55)3 ( 65,55)(75,65)23 ( 5,55)(5,15)3 ( 0,5)(5,15)23 ( 5,15)(35,15)3 ( 35,55)(35,15)3 ( 35,15)(65,55)3 ( 150,35)(160,35)23 ( 230,35)(240,35)23 ( 160,35)(195,70)3 ( 160,35)(195,0)3 ( 195,70)(230,35)3 ( 195,0)(230,35)3 ( 195,0)(195,35)3 ( 195,70)(195,35 ) ( 195,35)(215,35 ) ( 0,58)[t]@xmath23 ( 0,22)[t]@xmath24 ( 75,60)[t]@xmath24 ( 205,45)[t]@xmath23 ( 155,47)[t]@xmath24 ( 235,47)[t]@xmath24 in higher order it is only possible to obtain the integral over the structure function in eq .
( [ eqn2.5 ] ) for finite moments @xmath56 ( see @xcite ) .
this is achieved by @xmath57 in our example eq .
( [ eqn2.9 ] ) this procedure leads to the result @xmath58 which is again of the self energy type .
a method which turns out to be very useful to compute three - point and four - point functions even up to two - loop level is given by the mellin - barnes technique @xcite .
a simple mellin - barnes transformation takes the form @xmath59 with @xmath60 and the poles in @xmath61 are located on the real axis . to illustrate this technique we take as an example the three - point function where the propagators are raised to an arbitrary power @xmath62^{\nu_1 } [ ( q_2+k)^2-m_2 ^ 2]^{\nu_2}[(q_3+k)^2-m_3 ^ 2]^{\nu_3 } } \nonumber\\[2ex ] & = & i^{1-n } \pi^{n/2 } \frac
{ \gamma(\sum_i \nu_i - n/2)}{\prod_i \gamma(\nu_i ) } \int \prod_i d \alpha_i \ , \alpha_i^{\nu_i-1 } \,\delta(1- \sum_j \alpha_j ) \nonumber\\[2ex ] & & \times \frac{1}{[-\sum_j \alpha_j m_j^2 + \alpha_2\alpha_3 p_1 ^ 2 + \alpha_1\alpha_3 p_2 ^ 2+\alpha_1\alpha_2 p_3 ^ 2]^{\sum_k \nu_k - n/2}}\,,\end{aligned}\ ] ] where @xmath63 , @xmath64 and @xmath65 .
next we apply the double mellin barnes integral transformation given by @xmath66 & & \left ( \frac{x}{z } \right ) ^u\left ( \frac{y}{z } \right ) ^v\,,\end{aligned}\ ] ] so that the integral in eq .
( [ eqn3.2 ] ) can be written as @xmath67 & & \int_{-i\infty}^{i\infty } \frac{du}{2\pi i } \int_{-i\infty}^{i\infty }
\frac{dv}{2\pi i } \gamma(\sum_i \nu_i - n/2 + u + v ) \gamma(-u)\gamma(-v ) \left ( \frac{\alpha_3~p_1 ^ 2}{\alpha_1~p_3 ^ 2 } \right ) ^u \left ( \frac{\alpha_3~p_2 ^ 2}{\alpha_2~p_3 ^ 2 } \right ) ^v \,,\end{aligned}\ ] ] where we have put for simplicity @xmath68 . using @xmath69
the three - point function reads @xmath70 & & \pi^{n/2 } i^{1-n } \int_{-i\infty}^{i\infty } \frac{du}{2\pi i } \int_{-i\infty}^{i\infty } \frac{dv}{2\pi i } \frac{\gamma(n/2-\nu_1-\nu_3)\gamma(n/2-\nu_2-\nu_3)\gamma(\nu_3+u+v ) } { \prod_i \gamma(\nu_i)\gamma(n- \sum_j \nu_j ) } \nonumber\\[2ex ] & & \gamma(\sum_k \nu_k + u+v - n/2)\gamma(-u)\gamma(-v ) \left ( \frac{p_1 ^ 2}{p_3 } \right ) ^u \left ( \frac{p_2 ^ 2}{p_3 } \right ) ^u \left ( p_3 \right ) ^{n/2-\sum_j \nu_j}\,.\end{aligned}\ ] ] a special case is @xmath71 ( uniqueness condition ) .
after application of the residue theorem to expression ( [ eqn3.6 ] ) one obtains @xmath72 ( 180,70)(0,0 ) ( 10,35)(0,35 ) ( 10,35)(60,60 ) ( 10,35)(60,10 ) ( 60,60)(80,70 ) ( 60,10)(80,0 ) ( 35,48)(35,22 ) ( 60,60)(60,10 ) ( 5,32)[t]@xmath73 ( 70,60)[t]@xmath74 ( 70,17)[t]@xmath75 ( 110,35)(100,35 ) ( 110,35)(160,60 ) ( 160,60)(180,70 ) ( 110,35)(160,10 ) ( 160,10)(180,0 ) ( 135,48)(160,10 ) ( 160,60)(135,22 ) ( 105,32)[t]@xmath73 ( 170,60)[t]@xmath74 ( 170,17)[t]@xmath75 ( 200,70)(0,0 ) ( 15,60)(0,60 ) ( 75,60)(90,60 ) ( 15,10)(0,10 ) ( 75,10)(90,10 ) ( 15,60)(75,60 ) ( 15,10)(75,10 ) ( 15,60)(15,10 ) ( 75,60)(75,10 ) ( 45,60)(45,10 ) ( 5,21)[t]@xmath74 ( 5,57)[t]@xmath75 ( 85,21)[t]@xmath73 ( 85,57)[t]@xmath76 ( 125,60)(110,60 ) ( 185,60)(200,60 ) ( 125,10)(110,10 ) ( 185,10)(200,10 ) ( 125,60)(185,60 ) ( 125,10)(185,10 ) ( 125,60)(155,10 ) ( 155,60)(125,10 ) ( 185,60)(185,10 ) ( 115,21)[t]@xmath74 ( 115,57)[t]@xmath75 ( 195,21)[t]@xmath73 ( 195,57)[t]@xmath76 the mellin barnes techniques has been successfully applied to compute two - loop graphs .
examples are the three - point functions in fig .
[ fig5 ] and the four point functions in fig .
[ fig6 ] which are computed in @xcite for @xmath77 .
recent progress has been made for the two - loop box graphs in fig .
[ fig6 ] which has been computed for all external momenta on - shell ( @xmath78 ) @xcite , @xcite .
this is very useful for applications to the second order corrections in bhabha scattering and in di - jet production .
this method has been recently developed in @xcite @xcite to express one - loop integrals containing different masses into special functions . to illustrate the technique we take the two - point function as an example .
the latter is given by the integral @xmath79 using the schwinger representation for the propagator @xmath80 expression ( [ eqn4.1 ] ) becomes equal to @xmath81 the expression above can be evaluated in two different ways .
the first way proceeds by expanding the exponents in a power series .
this yields @xmath82 & & \prod_{i=1}^2 \frac{(-1)^{\nu_i}}{\gamma(\nu_i ) } \int_0^{\infty } d x_i \ , x_i^{\nu_i-1 }
\sum_{n_1=0}^{\infty}\sum_{n_2=0}^{\infty } i_2^n(-n_1,-n_2,q^2,m_1 ^ 2,m_2 ^ 2 ) \prod_{j=1}^2 \frac{x_j^{n_j}}{n_j!}\,.\end{aligned}\ ] ] using the identity @xmath83 we obtain @xmath84 there is a second way to compute eq .
( [ eqn4.3 ] ) .
first one shifts the momentum @xmath85 which becomes @xmath86 .
subsequently one uses the following identities @xmath87 notice that since @xmath88 one obtains @xmath89 .
this the reason for the name `` negative dimension approach '' .
the expression in eq .
( [ eqn4.3 ] ) can now be written as @xmath90 the integrand can be written as @xmath91 & & = \sum_{l=0}^{\infty } \frac{1}{l ! } ( x_1+x_2)^{-l - n/2}(x_1x_2q^2)^l \sum_{m=0}^{\infty } \frac{(-x_1m_1 ^ 2-x_2m_2 ^ 2)^m}{m!}\,.\end{aligned}\ ] ] using the identities @xmath92 one can write @xmath93 & & \sum_{m_1=0}^{\infty } \sum_{m_2=0}^{\infty } \frac{(-x_1m_1 ^ 2)^{m_1}}{m_1 ! } \frac{(-x_2m_2 ^ 2)^{m_2}}{m_2 ! }
\quad \mbox{with } \quad p_1+p_2=-l - n/2\,.\end{aligned}\ ] ] equating eqs .
( [ eqn4.6 ] ) and ( [ eqn4.11 ] ) one obtains the result @xmath94 & & ( q^2)^{l+m_1+m_2 } \times \left ( \frac{-m_1 ^ 2}{q^2 } \right ) ^{m_1 } \left ( \frac{-m_2 ^ 2}{q^2 } \right ) ^{m_2}\,,\end{aligned}\ ] ] with @xmath95 choosing two independent summation indices out of five i.e. @xmath96 one can express @xmath97 into the following special functions @xcite . given by the hyper - geometric functions @xmath98 , @xmath99 , the appell functions @xmath100
( @xmath101 ) , the horn function @xmath102 and the kamp de friet functions @xmath103 ( @xmath104 ) . in the case of the two - point function in eq .
( [ eqn4.1 ] ) there are eight possibilties to express the integral into the special functions above depending on the convergence of the sums in a specific kinematic region .
if we choose @xmath105 one has to take @xmath106 .
using the following identity @xmath107 the expression in eq .
[ eqn4.12 ] can be written as @xmath108 & & \times \frac{\gamma(1-\nu_1 ) \gamma(1-\nu_2)}{\gamma(1+m_1)\gamma(1+m_2 ) } ( q^2)^{n/2-\nu_1-\nu_2 } \times \left ( \frac{-m_1 ^ 2}{q^2 } \right ) ^{m_1 } \left ( \frac{-m_2 ^ 2}{q^2 } \right ) ^{m_2 } \nonumber\\[2ex ] & & = ( -1)^{n/2 } ( q^2)^{n/2-\nu_1-\nu_2 } \frac{\gamma(n/2-\nu_1)\gamma(n/2-\nu_2 ) \gamma(\nu_1+\nu_2-n/2)}{\gamma(\nu_1)\gamma(\nu_2)\gamma(n-\nu_1-\nu_2 ) } \nonumber\\[2ex ] & & f_4(1+\nu_1+\nu_2-n,\nu_1+\nu_2-n/2,1+\nu_1-n/2,1+\nu_2-n/2,\frac{m_1 ^ 2}{q^2 } , \frac{m_2 ^ 2}{q^2})\,.\end{aligned}\ ] ] here the fourth appell function is given by @xmath109 & & \times \frac{x^m}{\gamma(m+1 ) } \frac{y^n}{\gamma(n+1)}\,.\end{aligned}\ ] ] we conclude that using the method above one can find different representations for an one - loop integral in different kinematic regions depending on the radius of convergence of the ratios between the scales @xmath110 .
it is unclear whether this method , which seems to us very complicate , can be used to compute two - loop integrals .
van neerven , nucl . phys .
* b268 * ( 1986 ) 453 ; + e.b .
zijlstra and w.l .
van neerven , nucl .
b383 * ( 1992 ) 525 .
s. moch and j.a.m .
vermaseren , hep - ph/9912355 , hep - ph/9909269 .
k.g . chetyrkin and f.v .
tkachov , nucl . phys .
* b192 * ( 1981 ) 159 .
larin , t. van ritbergen and j.a.m .
vermaseren , nucl . phys .
* b427 * ( 1994 ) 41 ; + s.a .
larin , p. nogueira , t. van ritbergen and j.a.m .
vermaseren , nucl .
* b492 * 338 . n.i .
ussyukina and a.i .
davydychev , phys . lett . *
b298 * ( 1993 ) 363 , ibid . *
b332 * ( 1994 ) 159 , * b348 * ( 1993 ) 503 .
smirnov , phys.lett .
* b460 * ( 1999 ) 397 .
tausk , phys.lett . *
b469 * ( 1999 ) 225 . l.g.halliday and r.m .
ricotta , phys.lett . *
b193 * ( 1987 ) 241 ; + a.t .
suzuki and a.g.m .
schmidt , eur .
j. * c5 * ( 1998 ) 175 , phys . rev .
* d58 * ( 1998 ) 047701 . c. anastasiou , e.w.n . glover and c. oleari , hep - ph/9907494 . | we present some techniques which have been developed recently or in the recent past to compute feynman graphs beyond one - loop order . these techniques are useful to compute the three - loop splitting functions in qcd and to obtain the complete second order qed corrections to bhabha scattering .
inlo - pub 05/00 + march 2000 + * higher order corrections in perturbative * * quantum chromo dynamics * + |
the concept of _ s - property _ for logarithmic potentials appeared for the first time in the works of h. stahl @xcite , in the study of the limiting distribution of poles of pad approximants , see also @xcite for a recent survey and extension of these results . inspired by stahl s works ,
gonchar and rakhmanov extended the s - property to situations with external field @xcite .
they then characterized the limiting distribution of zeros of certain _ non - hermitian _ orthogonal polynomials , subject to the existence of a certain curve with the s - property in an external field - the _ s - curve _ , see section [ section_s_property ] below for a precise definition .
very recently , rakhmanov returned to the question of existence of s - curves from a general perspective @xcite .
he considered an associated _ max - min equilibrium problem _ in logarithmic potential theory , and in an `` ideal '' situation he proved that this max - min problem has a solution and the solution has the s - property .
this general approach by rakhmanov is very similar in spirit to his previous work with kamvissis @xcite , where a max - min problem for green energy in a particular external field was considered . in @xcite ,
rakhmanov also pointed out some examples where his proposed `` ideal '' situation does not apply , although similar considerations should also lead to the existence of a curve with the s - property .
the present work is aimed at studying this problem in a very particular , but important , model , namely when the external field is given by the real part of a polynomial . following the approach of rakhmanov , for a polynomial @xmath1 we consider a max - min problem of the form @xmath4,\ ] ] where the maximum is taken over a suitable class of contours @xmath5 ( see section [ section_admissible_contours ] below for precise definitions ) , and the infimum is taken over the set of borel probability measures
@xmath6 on @xmath5 .
the aim of this work is to prove that the max - min problem has a solution , and this solution leads to an s - curve @xmath5 . the max - min approach is not the only possible approach to the existence of the s - curve . in some particular cases
, it is possible to construct the s - curve explicitly from specific properties of the problem at hand , see e.g. @xcite for recent contributions . in general ,
the explicit determination of the s - curve - or more specifically the determination of the support of its equilibrium measure - is a hard problem , and some numerical studies have also been carried out , see @xcite . motivated by questions in random matrix theory
, bertola @xcite also studied the existence of the s - curve for polynomial external field in the same framework as the present paper . by using deformation techniques
, bertola was able to show the existence of the s - curve .
however , his approach just works when the underlying equilibrium measure vanishes as a square root on the endpoints of its support , although it is very likely that his approach could also be adapted to remove this restriction .
it is worth emphasizing that in our work there is no restriction on the underlying equilibrium measure - so we get the same result as bertola but valid in the general situation .
an extension of the concept of s - curves to vector equilibrium problems appears in the context of hermite - pad approximation , see e.g. @xcite .
vector equilibrium problems also appear in certain problems in random matrix theory , for example for random matrix models with external source @xcite and coupled random matrices @xcite , see @xcite for an overview .
the asymptotic analysis has been mostly restricted to cases with enough symmetry so that the relevant contours are contained in the real or imaginary axis . a suitable existence theory for curves with s - property appropriate to vector equilibrium problems with external fields , and possibly also upper constraints ,
would be a major step towards the analysis of situations with less symmetry .
see @xcite for a first example in this direction .
to state our main result , we first need to introduce a few notations and notions . throughout this paper , @xmath1 always denotes a complex polynomial of degree @xmath7 and @xmath8 also , we denote @xmath9 given a finite borel measure @xmath6 on @xmath10 , its _ logarithmic potential _ at @xmath11 is defined by @xmath12 and its _ logarithmic energy _ by @xmath13 whenever these integrals make sense .
for a closed set @xmath14 and @xmath15 the real part of a polynomial as before , we define @xmath16 as the set of borel probability measures @xmath6 supported on @xmath17 , satisfying the growth condition @xmath18 and by @xmath19 we denote the subset of @xmath20 consisting of all measures @xmath6 for which @xmath21 is finite and @xmath22 .
the condition assures the quantities @xmath23 are well defined elements of @xmath24 $ ] and @xmath25 is finite for a.e .
@xmath26 with respect to planar lebesgue measure . for @xmath27 , the quantity @xmath28 is called the _ weighted logarithmic energy of @xmath6 in the external field @xmath15 _ , or just weighted energy of @xmath6 .
the following minimization problem is classical @xcite : find a measure @xmath29 for which the infimum @xmath30 is attained , i.e. , for which @xmath31 .
the measure @xmath32 is called the _ equilibrium measure _ of the set @xmath17 in the external field @xmath15 . for the case without external field ,
that is , when @xmath33 , the measure @xmath34 is also called the _ robin measure _ of @xmath17 , the value @xmath35 is the _ robin constant _ of @xmath17 , the constant @xmath36 is called _ ( logarithmic ) capacity _ of @xmath17 and denoted by @xmath37 . for @xmath15 the real part of a polynomial as before ,
if @xmath5 is a contour in @xmath10 satisfying the growth condition @xmath38 the equilibrium measure @xmath39 uniquely exists @xcite and can be characterized through the _ euler - lagrange variational conditions_. it is the only measure in @xmath40 for which there exists a constant @xmath41 satisfying @xmath42 [ definition_s_property ] a contour @xmath5 satisfying the growth condition is said to possess the _ s - property in the external field _
@xmath15 if its equilibrium measure @xmath43 in the external field satisfies the following .
there is a finite set @xmath44 such that @xmath45 is locally an analytic arc at each of its points , and its potential @xmath46 satisfies @xmath47 where @xmath48 are the unit normal vectors to @xmath49 at @xmath50 .
if a contour has the s - property in the external field @xmath15 , we also call it an _ s - curve _ in the external field @xmath15 .
if the external field @xmath15 is understood ( as it is in this paper ) , we briefly call it an s - curve .
as it is clear already from the definition [ definition_s_property ] , the s - property is intrinsic to the equilibrium measure in the external field of the contour .
if we modify the contour away from the support , keeping the equilibrium measure in the external field the same for the modified contour , the s - property is preserved .
although in the present work we are just interested in the s - property stated in terms of logarithmic potentials , its analogue for potentials coming from other kernels , such as the green potential , have also their own interest @xcite .
there exist @xmath51 sectors @xmath52 in the complex plane along which @xmath53 as @xmath54 in @xmath55 .
these sectors are determined by the leading coefficient of @xmath1 . if @xmath56 , @xmath57 , then @xmath58 for @xmath59 .
we say that a set @xmath14 _ stretches out to infinity _ in the sector @xmath55 if there exist @xmath60 and @xmath61 such that for every @xmath62 there is @xmath63 with @xmath64 thus in particular we have @xmath65 if @xmath66 within @xmath67 .
a partition @xmath68 of @xmath69 is called _ noncrossing _ if it satisfies the following property : @xmath70 here @xmath71 denotes the equivalence relation that is associated with the partition @xmath68 .
the term _ noncrossing _ becomes more evident in a graphical representation of a partition , see figure [ examples_partitions_crossing_noncrossing ] .
.3 ( 0,0 )
( 3,-2 )
( 3,0 ) ; ( 3,-2 )
( 4,-2 ) ( 4,0 ) ; ( 1,0 )
( 1,-1 ) ( 5,-1 )
( 5,0 ) ; ( 2,0 )
( 2,-1 ) ; at ( 0,0 ) 1 ; at ( 1,0 ) 2 ; at ( 2,0 ) 3 ; at ( 3,0 ) 4 ; at ( 4,0 ) 5 ; at ( 5,0 ) 6 ; .3 ( 0,0 )
( 5,0 ) ; ( 4,0 )
( 4,-2 ) ; ( 1,0 )
( 2,0 ) ; ( 3,0 )
( 3,-1 ) ; at ( 0,0 ) 1 ; at ( 1,0 ) 2 ; at ( 2,0 ) 3 ; at ( 3,0 ) 4 ; at ( 4,0 ) 5 ; at ( 5,0 ) 6 ; to any noncrossing partition we associate a collection @xmath72 of admissible contours .
[ definition_admissible_sets ] let @xmath68 be a noncrossing partition of @xmath69 and @xmath73 the subset of @xmath68 obtained by removing from @xmath68 all singleton sets .
we assume @xmath74 .
we associate with @xmath68 the collection @xmath75 of admissible contours @xmath5 defined as follows .
* each @xmath76 is a finite union of @xmath77 jordan arcs . *
each @xmath76 has at most @xmath78 connected components , all of them unbounded and stretching out to infinity in at least two of the sectors @xmath52 .
* for each @xmath79 there is a connected component @xmath80 of @xmath5 that stretches out to infinity in each sector @xmath55 with @xmath81 . * if @xmath82 then there exists @xmath83 for which @xmath84 for simplicity , we will mostly write @xmath72 instead of @xmath75 . .3 possible_configurations_1.pdf ( 80,35)@xmath85 ( 60,80)@xmath86 ( 5,70)@xmath87 ( 5,10)@xmath88 ( 55,5)@xmath89 .3 and representative contours in @xmath75 for a polynomial @xmath1 of degree @xmath90 and positive leading coefficient .
the sectors @xmath91 are indicated on the top left figure . in gray is represented the region where @xmath15 is very negative.,title="fig : " ] .3 and representative contours in @xmath75 for a polynomial @xmath1 of degree @xmath90 and positive leading coefficient .
the sectors @xmath91 are indicated on the top left figure . in gray
is represented the region where @xmath15 is very negative.,title="fig : " ] + .3 and representative contours in @xmath75 for a polynomial @xmath1 of degree @xmath90 and positive leading coefficient .
the sectors @xmath91 are indicated on the top left figure . in gray
is represented the region where @xmath15 is very negative.,title="fig : " ] .3 and representative contours in @xmath75 for a polynomial @xmath1 of degree @xmath90 and positive leading coefficient .
the sectors @xmath91 are indicated on the top left figure . in gray
is represented the region where @xmath15 is very negative.,title="fig : " ] .3 and representative contours in @xmath75 for a polynomial @xmath1 of degree @xmath90 and positive leading coefficient .
the sectors @xmath91 are indicated on the top left figure . in gray
is represented the region where @xmath15 is very negative.,title="fig : " ] + .3 and representative contours in @xmath75 for a polynomial @xmath1 of degree @xmath90 and positive leading coefficient .
the sectors @xmath91 are indicated on the top left figure . in gray
is represented the region where @xmath15 is very negative.,title="fig : " ] .3 and representative contours in @xmath75 for a polynomial @xmath1 of degree @xmath90 and positive leading coefficient .
the sectors @xmath91 are indicated on the top left figure . in gray
is represented the region where @xmath15 is very negative.,title="fig : " ] figure [ table_configurations ] shows all noncrossing partitions ( up to rotation ) for a polynomial of degree @xmath92 and representative contours @xmath5 from each of the collections @xmath75 .
the restriction to noncrossing partitions in definition [ definition_admissible_sets ] is not essential .
we could as well have defined the admissible contours for arbitrary partitions , but then the class of admissible contours obtained for a given crossing partition would coincide with the class of admissible contours for some noncrossing partition . for the polynomial @xmath1 and a fixed noncrossing partition @xmath68 as above , the _ max - min energy problem _ for the pair @xmath93 asks for finding a contour @xmath94 for which @xmath95 , see ,
attains its maximum on @xmath72 .
that is , @xmath96 in what follows we use @xmath97 to denote the cauchy transform of a measure @xmath6 . the main result of this paper
is the following .
[ main_result ] let @xmath1 be a polynomial of degree @xmath7 and let @xmath68 be a noncrossing partition with class @xmath75 of admissible contours as in definition [ definition_admissible_sets ] .
then there exists a solution @xmath98 to the max - min energy problem for @xmath93 .
the contour @xmath99 has the s - property in the external field @xmath100 .
the equilibrium measure @xmath101 of @xmath99 in the external field @xmath15 is supported on a finite union of analytic arcs that are critical trajectories of the quadratic differential @xmath102 , where @xmath103 is a polynomial of degree @xmath104 .
furthermore @xmath105 is absolutely continuous with respect to arclength and @xmath106 see section [ section_quadratic_differentials ] for a brief discussion on quadratic differentials and their critical trajectories .
in we use a complex line element @xmath107 on @xmath99 , which induces an orientation on @xmath99 .
then @xmath108 denotes the limiting value of @xmath109 as @xmath110 from the left - hand side .
the contour @xmath99 that is the maximizer for the max - min energy property in theorem [ main_result ] is not unique .
we can modify the contour @xmath99 outside the support of @xmath105 slightly , preserving the equilibrium measure in the external field . in this way we obtain a new contour that also solves the max - min problem for @xmath111 .
although the contour is not unique , the equilibrium measure in the external field @xmath105 and its support turn out to be uniquely determined by @xmath1 and @xmath68 .
we do not know of a direct way to prove this , but there is an indirect way using orthogonal polynomials .
assume for simplicity that @xmath112 for some @xmath113 .
this means that any contour @xmath76 connects two fixed distinguished sectors at infinity where @xmath114 .
the ( non hermitian ) bilinear form defined on polynomials @xmath115 and @xmath116 , @xmath117 is then well defined and independent of the choice of contours @xmath118 .
a polynomial @xmath119 of degree at most @xmath120 is _ orthogonal _ with respect to @xmath121 if @xmath122 for any polynomial @xmath116 of degree @xmath123 . associated to the polynomial @xmath119 is its normalized zero counting measure @xmath124 @xmath125 in studying @xmath126 , gonchar and rakhmanov proved [ gonchar_rakhmanov_theorem_zeros ] if there exists @xmath76 with the s - property in the external field @xmath100 and @xmath127 is connected , then @xmath128 in other words , if a contour @xmath76 has the s - property , then its equilibrium measure in the external field is the weak limit of @xmath126 . by theorem [ main_result ]
any contour @xmath129 with the max - min property has the s - property .
also @xmath130 is connected , since otherwise we could remove self - intersecting subarcs of @xmath5 and increase its energy . thus ,
if @xmath99 , @xmath131 are two solutions to the max - min problem @xmath93 , then by theorem [ gonchar_rakhmanov_theorem_zeros ] both equilibrium measures @xmath132 and @xmath133 are the weak limit of the sequence @xmath126 . by uniqueness of limits , it follows that @xmath134 .
the above argument applies if @xmath135 for some @xmath113 . for other choices of @xmath73
, we need to modify the bilinear form @xmath121 to get a right family of orthogonal polynomials @xmath136 . for more details
, we refer to @xcite .
a polynomial @xmath137 of degree @xmath104 , @xmath7 , defines on the riemann sphere @xmath138 the _ quadratic differential _
a basic reference for quadratic differentials is the book by strebel @xcite .
the _ critical points _ of @xmath102 are the zeros of @xmath137 , whose orders as critical points are the same as their respective orders as zeros of @xmath137 , and the point @xmath139 , which is a pole of order @xmath140 .
the other points in @xmath138 are regular points of @xmath102 .
an arc @xmath141 is called a _ horizontal arc _ of @xmath102 if , along @xmath141 @xmath142 at a regular point , this condition can be locally stated as @xmath143 a maximal horizontal arc is called a _ horizontal trajectory _ , or shortly _ trajectory _ , of @xmath102 , and a trajectory that ends at a zero of @xmath137 is called a _
critical trajectory_. also of interest are the _ vertical trajectories _ , which are defined instead by the requirement @xmath144 which locally at a regular point is equivalent to @xmath145 the local structure of trajectories is well understood . through any regular
point @xmath146 passes exactly one horizontal and one vertical trajectory , which are analytic arcs that intersect at @xmath146 only , where they are orthogonal .
from a zero of @xmath102 of order @xmath115 emanate exactly @xmath147 trajectories , at equal consecutive angles @xmath148 . at the pole @xmath139 of order @xmath140 ,
the local behavior of trajectories is as follows ( * ? ? ?
* theorem 7.4 ) .
there are @xmath149 directions at @xmath150 forming equal angles , such that any trajectory ending at @xmath150 does so in one of these directions .
there is a neighborhood @xmath151 of @xmath150 such that any trajectory entering @xmath151 ends at @xmath150 in at least one direction , and trajectories entirely contained in @xmath151 end at @xmath150 in two consecutive directions , see figure [ local_trajectories ] , which shows the trajectories at infinity for a pole of order @xmath152 after inversion @xmath153 .
figure_3.pdf ( 220,80)@xmath154 as mentioned in the introduction , our work is very much influenced by the recent work of rakhmanov @xcite , which we now explain .
the first result obtained by rakhmanov in @xcite , which is relevant to us is the following .
[ rakhmanov_theorem_1 ] suppose @xmath17 is closed in @xmath155 .
if @xmath156 then the equilibrium measure @xmath157 in external field @xmath15 exists and is unique .
in contrast with previous results , theorem [ rakhmanov_theorem_1 ] does not contain any assumption on the growth of @xmath15 at @xmath150 .
actually the theorem in @xcite is more general as it also deals with situations where the external field is continuous except at a finite number of points , and no growth restriction near these singularities .
rakhmanov considers the energy functional @xmath158 defined on a metric space of compacts , which we explain next .
the riemann sphere @xmath155 is mapped to the sphere @xmath159 centered at @xmath160 with radius @xmath161 , @xmath162 via the inverse stereographic map @xmath163 given by @xmath164 and @xmath165 the _ hyperbolic metric _ on @xmath155 is then given by @xmath166 where the norm @xmath167 on the right - hand side above is the euclidean norm in @xmath168 .
the following relation holds @xmath169 this metric induces a metric on compact subsets of @xmath155 . if @xmath170 are compacts , the _ hausdorff metric _ between them is given by the formula @xmath171 where the distance between points and sets above is the usual one induced by @xmath172 , @xmath173 alternatively , denoting @xmath174 it can be equivalently defined by @xmath175 the hausdorff metric is thus defined on closed sets of @xmath138 .
naturally it induces a metric on closed sets of @xmath10 by adding the point at infinity to unbounded closed sets . in what follows ,
when we talk about the hausdorff metric on closed subsets of @xmath10 , we mean the metric obtained with this natural identification to closed subsets of @xmath138 .
we equip @xmath72 with the hausdorff metric induced by the hyperbolic distance .
so whenever we refer to a topological notion on @xmath72 such as convergence or continuity , we will always mean with respect to the hausdorff metric just introduced .
suppose @xmath176 is a class of closed subsets of @xmath10 for which there exists an absolute constant @xmath177 such that every compact set in @xmath176 has at most @xmath177 connected components .
equip @xmath176 with the hausdorff metric just introduced . for such classes @xmath176 , rakhmanov proves the following .
[ theorem_upper_semicontinuity ] the energy functional @xmath178 $ ] is upper semicontinuous .
if the class @xmath176 is closed , then it follows from there exists a set @xmath179 maximizing @xmath180 .
however , it is easy to see that the class of admissible contours @xmath181 introduced in definition [ definition_admissible_sets ] is not closed in the hausdorff metric . an easy way to see
this is through space - filling curves .
one can approximate sets of large planar measure with smooth contours , and the limiting set is not a smooth contour anymore .
a natural idea would then be to work with the closure @xmath182 of the class @xmath72 in the hausdorff metric .
there is however an issue with that , since sets in @xmath182 may stretch out to infinity in unlikely directions , and we might then recover the s - curve for the admissible class @xmath183 corresponding to another choice of partition @xmath184 .
our technique to overcome this last issue is to introduce a suitable subclass @xmath185 , see , consisting of contours that do not enter a `` forbidden region '' @xmath186 .
then we take the closure @xmath187 , see also . for the closed class @xmath188 , theorem [ theorem_upper_semicontinuity ]
assures us that there exists a set @xmath189 maximizing the energy functional @xmath95 on @xmath188 .
moreover , with @xmath190 chosen large enough , we can guarantee that the set @xmath189 does not come to the boundary of the forbidden region @xmath186 , which implies that small deformations of @xmath189 in any direction also belong to @xmath188 .
any @xmath191 function @xmath192 generates a one - parameter deformation @xmath193 of this maximizer set by @xmath194 now , since @xmath193 also belongs to @xmath176 for @xmath195 sufficiently close to @xmath196 , we have that ( the limit actually exists ) @xmath197 because @xmath189 is a maximizer set .
rakhmanov @xcite then proves that this limit implies that the equilibrium measure @xmath198 in the external field @xmath15 is critical , in the sense introduced by martnez - finkelshtein and rakhmanov @xcite , see also section [ critical_measures ] below . as it was already observed by martnez - finkelshtein and rakhmanov @xcite , the cauchy transform of a critical measure is an algebraic function .
in the present context , this means that the cauchy transform @xmath199 of the equilibrium measure in external field satisfies an algebraic equation of the form @xmath200 where @xmath137 is a polynomial of degree @xmath201 , see proposition [ proposition_algebraic_equation_critical_measure ] below .
the main technical difference between @xcite and the present work is that in @xcite this is obtained under the assumption that the critical measure is compactly supported , while here we obtain the algebraic equation first , and as one of its consequences we find that @xmath105 must be compactly supported . as in @xcite ,
the algebraic equation also implies that the measure @xmath105 is supported on a finite union of analytic arcs and its density can be recovered from the polynomial @xmath137 , see propositions [ proposition_density_critical_measure ] , [ proposition_support_critical_measure ] .
also , the s - property ultimately follows from the algebraic equation , see corollary [ corollary_s_property ] . as a final step ,
once some properties of the equilibrium measure in external field @xmath105 of the maximizer set @xmath189 are known , we drop the parts of @xmath189 that do not belong to @xmath202 , and we give in section [ proof_main_result ] a construction of a curve @xmath203 whose equilibrium measure in the external field @xmath15 is the desired measure @xmath105 .
the rest of the paper is organized as follows . in section [ critical_measures ]
we discuss critical measures .
the results therein presented are similar to the ones obtained for the first time by martnez - finkelshtein and rakhmanov @xcite .
however , as mentioned before and further explained later , the setup in @xcite is not the same as ours , so we preferred to present detailed proofs in section [ critical_measures ] . in section [ critical_measures_critical_sets ]
we explain the link between critical measures and critical sets , following the lines of @xcite .
section [ proof_main_theorem ] is devoted to the proof of theorem [ main_result ] .
as before , @xmath1 is a polynomial of degree @xmath204 and the external field @xmath15 is given by @xmath205 with proper modifications , all the results and proofs in this section are also valid for external fields given by the real part of an analytic function , possibly multivalued , with rational ( and so single - valued ) derivative , but for our purposes it is enough to consider the polynomial case .
the results we are going to show here were proven before by martnez - finkelshtein and rakhmanov @xcite under the assumption that the measures involved are a priori compactly supported and the external field has a rational derivative with simple poles .
the techniques we are using are also similar , but sometimes simplified and adapted to our needs .
the main difference here is that we do not impose a priori that a critical measure must have compact support .
ultimately , we do find that critical measures must be compactly supported ( for external fields as in ) , and then the results here are the same as the ones obtained in @xcite .
consider the set of test functions @xmath191 , consisting of @xmath206 complex - valued functions in @xmath10 with compact support .
for a function @xmath207 and @xmath208 denote @xmath209 any function @xmath210 generates local variations of sets @xmath211 and also of measures @xmath212 , where @xmath213 is the pushforward measure induced by @xmath214 .
if @xmath6 has finite weighted logarithmic energy then the same is true for @xmath213 . for a measure @xmath215 and @xmath207
, we define the _ derivative _ of the weighted energy functional at @xmath6 in the direction of @xmath216 by @xmath217 whenever the limit exists .
a first result is that @xmath218 indeed always exists .
[ proposition_formula_derivative ] given @xmath219 and @xmath207 , the derivative @xmath218 exists and is given by @xmath220 the proof here is the same as in ( * ? ? ?
* proof of lemma 3.7 ) . by the change of variables formula
@xmath221 which implies @xmath222 since @xmath207 , we have for @xmath223 , @xmath224 \\ & = t \operatorname{re}\left(\frac{h(x)-h(y)}{x - y}\right)+{\mathcal{o}}(t^2 ) , \label{equacao_auxiliar_9}\end{aligned}\ ] ] with the implicit constant in @xmath225 uniform in @xmath226 , and analogously @xmath227 where again the @xmath228 term is uniform in @xmath229 .
this implies @xmath230 and the result follows by plugging this last equation and into . [ definition_critical_measures ]
we say a measure @xmath231 is @xmath15_-critical _ if @xmath232 for every @xmath233 .
the definition of @xmath15-critical measures introduced in @xcite allows the external field to have singularities .
in that situation , the test functions @xmath216 appearing in definition [ definition_critical_measures ] should also vanish on a set @xmath234 of zero capacity containing the singular points of @xmath15 , leading to the notion of @xmath235-critical measures .
in @xcite it is also imposed that critical measures must have compact support . in the case
considered here the only singularity of the external field is at @xmath139 , and for convenience we are initially considering functions that are not just vanishing at infinity but also in a neighborhood of it , that is , compactly supported functions .
considering @xmath216 and @xmath236 separately , we easily obtain the following from proposition [ proposition_formula_derivative ] .
[ corollary_integral_equation_critical_measure ] a measure @xmath237 is @xmath15-critical if , and only if , @xmath238 for every @xmath207 .
equation appears frequently in the context of random matrices , although usually in different forms and names as for example _ loop equations _ , _ schwinger - dyson equations _ or _ ward identities _ , see @xcite .
we need to extend to larger classes of test functions .
[ lemma_test_function_at_infinity ] let @xmath6 be a @xmath15-critical measure .
then equation remains valid for @xmath239 satisfying @xmath240 introduce the function @xmath241 by @xmath242 then @xmath243 , @xmath244 and @xmath245 for @xmath239 satisfying , define @xmath246 then @xmath247 belongs to @xmath191 , so that equation is valid for it . from assumption and the fact that @xmath248 , we see @xmath249 and from the definition of @xmath247 and the dominated convergence theorem , it is easily seen @xmath250 the measure @xmath6 has finite logarithmic energy , so it has no mass points .
in particular , the diagonal @xmath251 has zero @xmath252 measure , implying @xmath253 using , we get @xmath254 from we find that both terms on the right - hand side are bounded , and therefore @xmath252-integrable .
the dominated convergence theorem can be applied , yielding @xmath255 and the equation for @xmath216 follows from this last limit and , keeping in mind that is valid for each @xmath247 .
[ lemma_test_function_singularity ] let @xmath6 be a @xmath15-critical measure .
if @xmath256 is such that @xmath257 then is valid for every function of the form @xmath258 the same idea used to prove lemma [ lemma_test_function_at_infinity ] can also be used here , namely approximating @xmath216 by a sequence of functions for which is valid .
but now the approximating sequence needs to be modified . by translating @xmath259
we can assume @xmath260 , and then we rewrite @xmath261 define @xmath262 where @xmath263 is a sequence of positive numbers converging to zero .
clearly @xmath264 , and so equation is valid for it .
note that assures us that @xmath216 is @xmath6-integrable .
proceeding then similarly as in the proof of lemma [ lemma_test_function_at_infinity ] , @xmath265 and also @xmath266 moreover , @xmath267 where in the last inequality we used @xmath268 for @xmath269 .
now , using the trivial decompositions @xmath270 in , we get @xmath271 since @xmath272 , the quotient @xmath273 is bounded , as well as @xmath274 , say both by @xmath190 .
this last inequality then implies @xmath275 from assumption the right - hand side above is @xmath252-integrable . from the dominated convergence theorem and we conclude @xmath276 since is valid for the @xmath247 s , this last limit together with implies for @xmath216 . a combination of lemmas [ lemma_test_function_at_infinity ] and [ lemma_test_function_singularity ] leads to the following . [ corollary_test_function ]
let @xmath6 be a @xmath15-critical measure .
suppose @xmath277 are distinct points such that @xmath278 then equation is valid for the function @xmath279 let @xmath280 .
for each @xmath281 , we can take a @xmath191 function @xmath282 $ ] supported in @xmath283 with @xmath284 in @xmath285 .
then @xmath286 satisfies the condition of lemma [ lemma_test_function_singularity ] and @xmath287 satisfies the condition of lemma [ lemma_test_function_at_infinity ] .
thus the equality holds for these functions , and then by linearity it also holds for @xmath216 .
given a finite positive borel measure @xmath6 , consider the function @xmath288 $ ] @xmath289 it is the convolution of a finite borel measure with the function @xmath290 , which is in @xmath291 , where @xmath292 is the planar lebesgue measure .
thus tonelli s theorem tells us it also belongs to @xmath291 , being then finite @xmath292-a.e .
this implies that the cauchy transform @xmath293 of @xmath6 , @xmath294 is well - defined and finite @xmath292-a.e . , and @xmath295 as observed by martnez - finkelshtein and rakhmanov ( * ? ? ?
* lemma 5.1 ) , an essential feature of critical measures is the fact that their cauchy transform satisfies an algebraic equation of degree @xmath296 .
[ proposition_algebraic_equation_critical_measure ] let @xmath6 be a @xmath15-critical measure .
then there exists a polynomial @xmath137 of degree @xmath104 such that @xmath297 where @xmath292 is the lebesgue measure on @xmath10 . as we will see later ,
one of the consequences of proposition [ proposition_algebraic_equation_critical_measure ] is that @xmath15-critical measures are always compactly supported , see proposition [ proposition_support_critical_measure ] below .
however , since we can not assume this _ a priori _ , our proof , at the technical level , is different from the one given in ( * ? ? ?
* proof of lemma 5.1 ) , although the key ideas are similar . since @xmath298 is finite @xmath292-a.e .
, we can fix @xmath299 distinct points @xmath300 for which @xmath301 take @xmath302 . since we want to have @xmath292-a.e .
and @xmath298 is finite @xmath292-a.e .
, it is enough to prove the identity under the assumption @xmath303 define @xmath304 then @xmath216 satisfies the conditions of corollary [ corollary_test_function ] ( with @xmath305 ) and so is valid for this @xmath216 .
we can write @xmath306 where @xmath307 is the rational function @xmath308 whose only possible poles are the points @xmath300 , all of them simple , and @xmath309 on the other hand , we can write @xmath310 the first term in the right - hand side is bounded away from @xmath300 , and since we are assuming this term is @xmath311-integrable .
the second term is also @xmath312-integrable because of assumptions and . from this
we conclude @xmath313 where @xmath314 is also a rational function with simple poles at @xmath300 and no other poles .
moreover , @xmath315 as is valid for this @xmath216 , equations and give us @xmath316 which is equivalent to if we set @xmath317 note that the poles of @xmath318 cancel out the possible poles of @xmath319 , and due also to the behavior of @xmath307 and @xmath320 at infinity given in , we see that @xmath137 is indeed a polynomial of degree @xmath104 . from the polynomial @xmath137 in
, we can recover @xmath6 .
[ proposition_density_critical_measure ] suppose @xmath6 is a measure on @xmath321 for which there exist polynomials @xmath322 and @xmath137 such that @xmath323 then @xmath6 is supported on a union of analytic arcs , which are maximal trajectories of the quadratic differential @xmath324 .
moreover , in the interior of any arc of @xmath49 , the measure @xmath6 is absolutely continuous with respect to the arclength measure , with density given by @xmath325 where @xmath107 is the complex line element , chosen according to a fixed orientation of the arcs of @xmath49 .
observe that by proposition [ proposition_algebraic_equation_critical_measure ] , the condition is valid for @xmath15-critical measures with the choice @xmath326 . since proposition [ proposition_density_critical_measure ] is essentially about local properties of the measure @xmath6 , it is not important if the measure @xmath6 has unbounded support or not .
so the proof presented here is essentially the same as the one given in ( * ? ? ?
* lemma 5.2 ) for measures with bounded support , which in turn is modelled after a result in @xcite .
we decided to give a full detailed proof here for the benefit of the reader , and also to correct a sign misprint in @xcite : in the first centered formula following formula ( 5.23 ) in @xcite the right - hand side should have a @xmath327-sign .
let @xmath328 be a simply connected domain not containing zeros of @xmath137 and intersecting @xmath49 .
fix a point @xmath329 and select a branch of @xmath330 in @xmath328 , and define @xmath331 reducing @xmath328 if necessary , we can assume @xmath332 is a conformal mapping between @xmath328 and @xmath333 , @xmath334 intervals both containing @xmath260 . for this same branch of @xmath330 , define for @xmath292-a.e .
@xmath335 a function @xmath336 by the formula @xmath337 due to @xmath336 assumes values in @xmath338 . a simple application of the cauchy - green formula ( * ? ?
* pg . 491 ) gives us @xmath339 in the sense of distributions , and we conclude ( also in the sense of distributions ) @xmath340 now , @xmath332 is a conformal map from @xmath328 onto @xmath341 , with inverse denoted by @xmath342 , which induces a map between distributions in the @xmath50-variable and the @xmath332-variable , say @xmath343 , where @xmath344 is the distribution acting on test functions in @xmath345 via the formula @xmath346 where @xmath347 .
now we will calculate the ( distributional ) derivative of @xmath348 .
if @xmath349 is the pushforward measure of @xmath6 induced by @xmath332 in @xmath341 then and the chain rule applied to @xmath347 imply @xmath350 since @xmath351 and @xmath352 .
thus by the change of variables formula for the lebesgue measure @xmath353 where we used that @xmath354 .
the result is that @xmath355 which can be rewritten as @xmath356 where @xmath357 with @xmath358 .
as @xmath6 is a real measure , we conclude @xmath359 , which means @xmath360 , for a real function @xmath274 defined on the interval @xmath361 . if @xmath362 is the pushforward measure of @xmath349 induced by @xmath363 , using we can conclude in a similar fashion that @xmath364 where @xmath365 is a positive constant , and so @xmath366 for some real constant @xmath367 . but
@xmath274 only assumes values in @xmath368 , and since @xmath369 is non - decreasing , we conclude @xmath274 is non - increasing and also that there exists @xmath370 in @xmath361 for which @xmath371 , where @xmath177 is a positive constant , and then @xmath372 in terms of @xmath349 and @xmath373 , this means that there exists a vertical segment @xmath374 in @xmath341 with @xmath373 equal to @xmath375 at the left side of @xmath376 and @xmath377 at the right side of @xmath376 , and @xmath378 is contained in @xmath376 . but for such @xmath373 a direct computation shows @xmath379 where @xmath380 is the lebesgue measure on @xmath376 , so by @xmath381 where @xmath382 is the line element in @xmath376 , oriented such that @xmath383 on the positive side of @xmath376 . since @xmath332 is a conformal map we can pullback the measures @xmath384 and @xmath382 , obtaining @xmath385 where @xmath107 is the complex line element on the trajectory @xmath386 chosen with orientation such that @xmath387 on the positive side ( i.e. , on the left - hand side ) of @xmath141 . finally , the trajectory arcs of @xmath49 must be maximal , because the arguments above show that @xmath49 is an analytic arc in a neighborhood of any point of its support that is not a zero of @xmath137 . for the next proposition ,
it is important that @xmath100 with @xmath1 polynomial .
[ proposition_support_critical_measure ] for any @xmath15-critical measure @xmath6 , the support @xmath49 is compact and can be written as @xmath388 where each @xmath389 is an analytic arc and the union is taken over a finite set of indices .
each @xmath389 is a critical trajectory of the quadratic differential @xmath390 .
propositions [ proposition_algebraic_equation_critical_measure ] and [ proposition_density_critical_measure ] tell us that the measure @xmath6 is supported on maximal trajectories of the quadratic differential @xmath391 which has a pole of order @xmath140 at @xmath139 .
the general theory of quadratic differentials assures us that there exists a neighborhood @xmath151 of @xmath139 such that every trajectory intersecting @xmath151 ends up at @xmath139 in at least one direction ( * ? ? ?
* theorem 7.4 ) .
in particular , if @xmath392 for some maximal trajectory @xmath141 of @xmath49 , then @xmath141 would be unbounded , and due to the measure @xmath6 would not be finite . since @xmath137 is a polynomial ,
the only bounded trajectories are critical trajectories that connect two different zeros of @xmath137 . since there are only finitely many zeros of @xmath137 , the union in is certainly over a finite number of arcs .
it is worth noting that the expression for @xmath137 in terms of the function @xmath318 defined in , depends on the chosen points @xmath300 at which the cauchy transform of @xmath6 is convergent .
since we now know by proposition [ proposition_support_critical_measure ] that @xmath49 is compact , we can take the constant function @xmath393 in , and the rest of the proof of then works fine for this @xmath318 . repeating all the computation , we also end up with a nicer representation for @xmath137 , @xmath394 which is well - known for equilibrium problems in polynomial external fields on the real line , see e.g. @xcite .
this representation also allows us to derive some extra properties for @xmath137 .
for example , from one easily sees that @xmath395 is a polynomial of degree @xmath396 , whose leading coefficient coincides with minus the leading coefficient of @xmath397 . also , if @xmath1 is quadratic , say @xmath398 then reduces to @xmath399 since @xmath6 is a probability measure .
then the zeros of @xmath137 are @xmath400 for some choice of the branch of the square root , and the support of the @xmath15-critical measure is the segment from @xmath401 to @xmath402 . for @xmath1 cubic ,
the polynomial @xmath395 is linear , so there is one free parameter to be determined by the requirement of extra conditions , see for instance @xcite .
see also @xcite for a similar situation . by proposition
[ proposition_density_critical_measure ] , the support of a @xmath15-critical measure @xmath6 consists of analytic arcs and their endpoints .
the analytic arcs are trajectories of the quadratic differential @xmath390 .
each regular point @xmath403 has a neighborhood , say @xmath328 , for which @xmath404 is an analytic arc .
the only non - regular points are the zeros of @xmath137 .
the next result is taken from @xcite .
[ proposition_s_property_critical_measures ] let @xmath6 be a @xmath15-critical measure , with @xmath405 as in proposition [ proposition_support_critical_measure ] .
a. then there exist constants @xmath406 such that @xmath407 b. if @xmath408 is regular , then @xmath409 where @xmath48 are the normal vectors to @xmath49 at @xmath50 pointing in opposite directions .
see lemma 5.4 of @xcite . an immediate consequence of part ii ) of proposition [ proposition_s_property_critical_measures ] is .
[ corollary_s_property ] if @xmath1 is a polynomial , @xmath100 and @xmath410 is a contour satisfying the growth condition and whose equilibrium measure @xmath39 is a @xmath15-critical measure , then @xmath5 has the s - property in the external field @xmath15 . as a final remark , we notice that the cornerstone of the present section is the proposition [ proposition_algebraic_equation_critical_measure ] , which gives us that the cauchy transform of a @xmath15-critical measure is an algebraic function .
one could also analyse from another perspective , namely given @xmath397 , find a polynomial @xmath137 and a probability measure @xmath6 such that the algebraic equation is satisfied .
this is the spirit of some works already present in the literature , e.g. @xcite .
it is expected that the pairs @xmath411 satisfying depend on continuous parameters , but as it follows from the present work , only for a finite number of pairs @xmath411 the measure @xmath6 is an equilibrium measure ( of some admissible contour ) in the external field @xmath100 .
more precisely , just for a finite number of pairs @xmath411 obtained in this way , all the constants @xmath412 appearing in proposition [ proposition_s_property_critical_measures ] i ) coincide .
in such a case there is a choice of partition @xmath68 and a contour @xmath75 for which @xmath6 is the equilibrium measure of @xmath5 in the external field @xmath15 .
we continue to use the notation introduced in the beginning of section [ critical_measures ] .
so for @xmath210 we have @xmath214 as in , and we denote by @xmath213 the pushforward of a measure @xmath6 induced by @xmath214 . for a subset @xmath14 , we denote @xmath413 .
in section [ critical_measures ] we studied variations of the energy functional @xmath180 induced by @xmath216 , viewed as acting on measures , see .
but we might as well study variations of @xmath414 when viewed as acting on sets , that is , we might consider the limit @xmath415 just as the vanishing of @xmath416 defines critical measures , the vanishing of @xmath417 defines critical sets .
[ definition_critical_set ] we say that a closed set @xmath17 is a @xmath15-critical set ( or simply a critical set ) if the limit in exists for every @xmath418 , and @xmath419 for every @xmath418 .
the remarkable fact is that the limit coincides with the derivative of @xmath180 at the equilibrium measure in external field of @xmath17 .
[ proposition_derivative_energy_compact ] let @xmath14 be a closed set with @xmath420 .
then the limit in exists and @xmath421 the existence of the equilibrium measure in the external field @xmath157 is guaranteed by theorem [ rakhmanov_theorem_1 ] .
proposition [ proposition_derivative_energy_compact ] is due to rakhmanov , who gives a proof in ( * ? ? ?
* section 9.10 ) and refers to an unpublished manuscript where the argument is first given .
we found that one of the steps in the proof in @xcite , namely the equivalent of below , might require some additional explanation , and therefore we decided to present here a detailed proof . to prove proposition [ proposition_derivative_energy_compact ]
, we will need an auxiliary lemma .
we use @xmath422 to denote the set of signed measures defined on @xmath10 , whose positive and negative parts belong to @xmath423 ( see the beginning of section [ potential_theory_notions ] for the definition of @xmath423 ) . recall that a sequence of finite signed measures @xmath424 converges vaguely to a finite signed measure @xmath425 if @xmath426 for every continuous functions @xmath427 with compact support .
[ lemma_vague_convergence ] the functional @xmath428 is well - defined on @xmath422 and strictly positive for @xmath429 .
if @xmath424 is a sequence of signed measures in @xmath422 with @xmath430 then the sequence @xmath424 converges vaguely to the null measure .
the lemma is well known if we consider signed measures supported in the unit disc , see for example ( * ? ? ?
80 , theorem 1.16 , pg .
88 , lemma 1.3 ) .
the main issue here is that we are dealing with signed measures with possibly unbounded support .
the proof we are going to give is actually obtained by collecting known results together and fitting them to our needs here . given two measures @xmath431 with finite energies @xmath21 ,
@xmath432 , the integral @xmath433 is well defined and finite ( * ? ? ?
* theorem 2.2 ) , that is , @xmath434 .
in particular , if @xmath435 , then the value @xmath436 is also well defined and finite . due to a result of mattner ( * ? ? ? * theorem 2.2 ) it is positive , i.e. , @xmath437 and @xmath438 if and only if @xmath425 is the zero measure
. for @xmath439 compactly supported and absolutely continuous with respect to the lebesgue measure @xmath292 , say @xmath440 , with @xmath441 and real - valued satisfying @xmath442 , cegrell et al .
* lemma 2.4 ) obtained the representation @xmath443 for @xmath444 , where @xmath445 is the fourier transform of @xmath427 ( induced by @xmath446 , so that @xmath447 is just the inner product in @xmath446 ) . for @xmath448 and a given sequence of approximation to the identity @xmath449 (
@xmath450 is smooth and independent of @xmath425 ) , satisfying @xmath451 cegrell et al .
also constructed a sequence of absolutely continuous signed measures @xmath452 with @xmath453 where @xmath454 and @xmath362 is the normalized lebesgue measure on the unit disc and @xmath455 a sequence of discs @xmath456 whose respective sequence of radii @xmath457 converges to @xmath458 when @xmath459 .
so now consider a sequence @xmath460 in @xmath422 with @xmath461 .
for each @xmath462 consider the measures @xmath463 defined as in for @xmath464 .
for each @xmath462 , we can take @xmath465 large enough , such that for @xmath466 , @xmath467 , for all @xmath462 .
the explicit form of @xmath468 and this last condition on the indices imply in particular @xmath469 the measures @xmath468 are absolutely continuous w.r.t .
@xmath292 with a smooth and compactly supported density , say @xmath470 . equations and and the convergence @xmath461
then imply @xmath471 if @xmath472 , then @xmath473 and also @xmath474 , and the plancherel theorem and cauchy - schwarz inequality imply for @xmath475 , @xmath476 so that @xmath477 finally , since @xmath478 is dense in @xmath479 w.r.t .
the uniform norm and @xmath480 is uniformly bounded , the inequality @xmath481 for a sequence @xmath482 of functions in @xmath478 converging to @xmath483 in the uniform norm shows the limit is also true for @xmath483 , that is , @xmath484 vaguely . in virtue of this
is equivalent to the vague convergence @xmath485 .
for @xmath195 small enough and @xmath214 as in , the inverse map @xmath486 is a well defined @xmath191 function which satisfies @xmath487 with the @xmath488 terms uniform in @xmath489
. for simplicity , denote by @xmath490 and @xmath491 the equilibrium measures in the external field @xmath15 of @xmath17 and @xmath492 , respectively . by rakhmanov
s theorem [ rakhmanov_theorem_1 ] and the assumption @xmath233 these measures certainly exist . clearly , in view of the definitions and and the proposition [ proposition_formula_derivative ] , it is enough to prove that @xmath493 which comes down to proving @xmath494 for a given @xmath195 , denote by @xmath495 the pushforward measure of @xmath491 induced by the inverse mapping @xmath496 .
bearing in mind equations , and mimicking the proof of proposition [ proposition_formula_derivative ] we obtain @xmath497 and also from proposition [ proposition_formula_derivative ] , @xmath498 this in turn implies @xmath499 so we are done if we can show @xmath500 to obtain , assume for a moment @xmath501 that is , @xmath502 vaguely , to be proved later . since @xmath233 the function @xmath503 is continuous with compact support , so that by @xmath504 the function @xmath505 belongs also to @xmath191 , so again by @xmath506 using proposition [ proposition_formula_derivative ] and , , the limit follows and the proof is done , provided that we have .
the proof of will be done in two steps , namely @xmath507 and @xmath508 vaguely .
the first of these limits follows directly from the change of variables formula , @xmath509 because @xmath427 has compact support , and so it is bounded , and @xmath510 as @xmath511 , and the dominated convergence theorem can be applied .
the difference @xmath512 is bounded , thanks to proposition [ proposition_formula_derivative ] , so implies @xmath513 .
now , following notations of the previous lemma , @xmath514 as @xmath223
. the inequality in is obtained by noting that @xmath515 , so its weighted energy is larger than @xmath516 .
lemma [ lemma_vague_convergence ] then implies @xmath517 vaguely .
an immediate consequence of proposition [ proposition_derivative_energy_compact ] is the following .
[ cor_critical_set ] the equilibrium measure @xmath157 in the external field @xmath15 of a @xmath15-critical set @xmath17 is a @xmath15-critical measure .
in this section we are going to prove theorem [ main_result ] .
recall we are assuming that @xmath1 is a polynomial of degree @xmath51 , @xmath100 and @xmath518 is the class of admissible contours associated to a fixed noncrossing partition @xmath68 , as given in definition [ definition_admissible_sets ] .
this setup defines the max - min problem @xmath519 , see . in the first step we are going to restrict the class of contours @xmath72 .
the contours in @xmath72 stretch out to infinity in certain directions in which @xmath521 , as specified by the partition @xmath68 .
however , it is not forbidden that some parts of @xmath522 are in regions of the plane where @xmath15 is very negative
. then also @xmath523 will be very negative , and so such @xmath5 will be far from optimal for the max - min problem .
for @xmath524 large enough , the level set @xmath525 consists of @xmath51 disjoint analytic arcs , each of them stretching out to infinity in its both ends .
this is so because @xmath15 is the real part of a polynomial of degree @xmath51 .
then , if @xmath524 is large enough , @xmath526 is an open non - empty set and its boundary consists of a union of @xmath51 pairwise disjoint analytic arcs .
the distance in is the usual euclidean distance @xmath527 the set @xmath186 is indicated by the gray regions in figures [ table_configurations ] and [ level_curves ] .
we also assume @xmath190 satisfies @xmath528 then for such @xmath190 we consider the subclass @xmath529 and its closure @xmath530 with respect to hausdorff metric on closed subsets of @xmath10 with the hyperbolic distance .
the sets in @xmath188 are closed subsets of @xmath10 , but they are not necessarily finite unions of contours , since this property is not preserved under taking closure in the hausdorff metric . however , because of definition [ definition_admissible_sets ] ii ) , each @xmath531 has at most @xmath78 components that are all unbounded in @xmath10 , since this property is preserved by taking closure in the hausdorff metric .
also @xmath532 we start with the proof of the first inequality in . to prove this we take @xmath534 .
since all connected components of @xmath5 are unbounded , see item ii ) of definition [ definition_admissible_sets ] , and @xmath5 intersects the set @xmath186 from , we have that @xmath5 contains a connected subset @xmath99 satisfying @xmath535 then @xmath536 , which means that @xmath537 , see for example @xcite .
let @xmath538 be the equilibrium measure on @xmath99 ( without external field ) .
then @xmath539 , and @xmath540 because @xmath99 is contained in the region where @xmath541 .
then by @xmath542 it remains to prove that @xmath545 is finite . to this end
, we construct a closed curve @xmath141 , contained in @xmath546 and intersecting all the connected components of @xmath547 , see figure [ level_curves ] for a pictorial configuration of @xmath141 .
recall that by item ii ) of definition [ definition_admissible_sets ] the connected components of sets in @xmath72 are unbounded and stretch out to infinity in at least two of the sectors @xmath52 .
assuming in addition that @xmath548 , that is , @xmath549 , it is easy to see that @xmath550 for any connected component @xmath80 of @xmath5 .
this property is preserved by taking closures with respect to hausdorff metric , that is , it is also valid @xmath551 for any connected component @xmath552 of a set @xmath17 belonging to @xmath188 .
fix @xmath553 for which @xmath554 . a similar reasoning as the one that led to shows that @xmath555 for any connected component @xmath552 of @xmath556 . from
and we see that there exists a connected compact @xmath557 with @xmath558 this means that @xmath559 , because @xmath554 .
then @xmath560 , which implies @xmath561 , and then @xmath562 since @xmath137 is independent of @xmath531 , the inequality @xmath563 follows .
the class @xmath188 is closed in the hausdorff metric .
the upper semicontinuity of @xmath180 given by theorem [ theorem_upper_semicontinuity ] and proposition [ proposicao_fundamental ] give the desired result .
let @xmath418 be a test function .
since @xmath216 is compactly supported , it follows that there is a @xmath567 such that @xmath572 we also take @xmath573 small enough so that @xmath574 is a @xmath206-diffeomorphism on @xmath321 for @xmath575 .
let @xmath569 , and let @xmath576 be a sequence in @xmath520 that converges to @xmath189 in hausdorff metric .
it is then an easy fact that the convergence @xmath577 in the hausdorff
metric also holds true .
take @xmath578 such that @xmath579 .
since @xmath214 is a diffeomorphism that leaves @xmath580 invariant , we then have @xmath581 for all @xmath120 . since @xmath582 , it does not intersect @xmath186 , and so by @xmath583 the convergence , together with the fact that @xmath193 does not intersect the closed set @xmath584 , see , implies that for large enough @xmath120 , @xmath585 then by
and we have @xmath586 for @xmath120 large enough , and so @xmath587 by the definition of the class @xmath520 .
because of , we then conclude that @xmath588 . from lemma [ lemma_small_variations ]
it follows that the maximizer @xmath189 is a @xmath15-critical set .
thus by corollary [ cor_critical_set ] its equilibrium measure in the external field @xmath15 is a @xmath15-critical measure and the results of section [ critical_measures ] apply to this measure .
thus we obtain the following proposition that summarizes the work we did so far .
[ characterization_critical_measures_polynomial ] the equilibrium measure @xmath589 of @xmath189 is @xmath15-critical . there exists a polynomial @xmath137 such that @xmath590 moreover , @xmath202 is compact and consists of a finite union of analytic arcs , which are maximal trajectories of the quadratic differential @xmath102 , with each trajectory connecting two distinct zeros of @xmath137 .
the measure @xmath105 is absolutely continuous with respect to the arclength measure on those arcs and @xmath591 where @xmath107 is the complex line element . at any point @xmath592 which is not a zero of @xmath137 , its logarithmic potential satisfies @xmath593 where @xmath48 are the unit normal vectors to @xmath202 at @xmath50 pointing in opposite directions . having in hands the equilibrium measure in the external field @xmath589 of the maximizer set @xmath189 given by corollary [ corollary_maximizing_set ] ,
our final task is to construct a contour @xmath94 for which @xmath594 .
note that we know that the support of @xmath105 consists of a finite union of analytic arcs , but we do not know that the full set @xmath189 is a contour .
there is actually no reason why this should be the case , and so we will modify @xmath189 outside of the support of its equilibrium measure in the external field , while preserving the equilibrium measure in the external field .
we also have to show that the contour @xmath99 that we obtain this way belongs to the class @xmath72 . [ lemma_description_level_sets ] for any @xmath597
, there exists @xmath598 such that for every @xmath599 the following holds .
the set @xmath600 has exactly @xmath51 connected components , each of them contained in precisely one of the sectors @xmath601 and containing in its interior the half - rays @xmath602 the boundary of @xmath604 consists of level curves of @xmath605 , and these are exactly the trajectories of the quadratic differential @xmath102 given in .
the quadratic differential has a pole of order @xmath140 at @xmath139 .
lemma [ lemma_description_level_sets ] then follows from the local behavior of trajectories of a quadratic differential near a pole of order @xmath606 , see ( * ? ? ?
* theorem 7.4 ) .
fix a number @xmath607 and also @xmath608 for which @xmath609 for @xmath610 , let @xmath611 be the connected component of @xmath612 for which @xmath613 then @xmath611 is pathwise connected , and we can connect any point in @xmath611 to infinity through a curve entirely contained in @xmath611 and stretching out to infinity in the sector @xmath55 ( see for the definition of @xmath55 ) . from and lemma [ lemma_description_level_sets ]
we see that @xmath624 .
given @xmath618 , we have by lemma [ lemma_refinement_partition ] that the points @xmath625 with @xmath81 , belong to the same connected component of @xmath626
. then there exists a connected set @xmath627 which is a finite union of bounded @xmath77 jordan arcs satisfying @xmath628 for @xmath629 given by , define @xmath630 each @xmath80 is then a finite union of @xmath77 arcs .
@xmath80 is connected and @xmath80 stretches out to infinity in each sector @xmath55 with @xmath81 .
the union @xmath631 then satisfies all the requirements in definition [ definition_admissible_sets ] and it follows that @xmath632 hence @xmath633 on the other hand , since @xmath636 , the variational conditions , are valid for @xmath637 , which means that @xmath105 is the equilibrium measure of @xmath637 in the external field @xmath15 .
thus @xmath638 since @xmath105 is the equilibrium measure in the external field of @xmath189 as well .
then we recall that @xmath189 is the maximizer of @xmath414 over the class @xmath188 which contains @xmath520 . then by proposition [ proposicao_fundamental ] @xmath639 combining the inequalities in , we finally get @xmath640 and so equality holds throughout in .
let @xmath641 be the equilibrium measure in the external field @xmath15 of @xmath99 .
then because of the equalities in , we have @xmath642 thus @xmath643 is a measure supported on @xmath644 whose energy in external field @xmath15 is equal to the minimal energy for measures on @xmath645 . by uniqueness of equilibrium measure
it follows that @xmath646 and @xmath647 is the equilibrium measure in external field @xmath15 of @xmath99 .
then @xmath94 has the s - property in external field @xmath15 , see in proposition [ characterization_critical_measures_polynomial ] .
also the remaining statements of theorem [ main_result ] follow from proposition [ characterization_critical_measures_polynomial ] .
this completes the proof of theorem [ main_result ] , assuming the validity of lemma [ lemma_refinement_partition ] .
let @xmath655 . from item
iii ) of definition [ definition_admissible_sets ] it follows that @xmath656 has a connected component that stretches out to infinity in the sectors @xmath55 and @xmath657 .
then by dropping parts of that component that go to infinity in other sectors , and parts that make loops , we can find a subset @xmath658 that is a simple piecewise @xmath77 contour that goes from infinity in the sector @xmath55 to infinity in the sector @xmath657 . here
simple means that it has no points of self - intersection .
we consider @xmath659 as an oriented contour , and we talk about points on the contour that lie before or after other points . from the definition of @xmath520 , see , we know that @xmath660 the region @xmath186 consists of @xmath51 connected components , all of them stretching out to infinity , as shown with the gray shaded regions in figure [ level_curves ] for the case @xmath661 . figure_5.pdf ( 250,120)@xmath662 ( 175,250)@xmath663 ( 03,230)@xmath664 ( 03,25)@xmath665 ( 175,05)@xmath666 ( 230,107)@xmath667 ( 180,220)@xmath668 ( 50,205)@xmath669 ( 60,45)@xmath670 ( 137,23)@xmath671 ( 240,162)@xmath672 ( 45,175)@xmath673 ( 43,88)@xmath674 ( 61,61)@xmath675 ( 172,32)@xmath676 ( 175,55)@xmath677 ( 208,160)@xmath678 let @xmath137 be large enough such that the circle @xmath679 intersects each of the components of @xmath186 along a circular arc
. then there are also @xmath51 circular subarcs of @xmath680 outside of @xmath186 , that we call @xmath681 where @xmath682 is in the direction of the sector @xmath683 , for @xmath684 . by taking @xmath137 large enough
, we can also make sure that @xmath685 for each @xmath686 , where @xmath687 is as in .
we now follow the contour @xmath659 starting at infinity in the sector @xmath55 .
it will meet the arc @xmath389 .
we let @xmath673 be the _
last point _ on @xmath659 that is on @xmath389 .
then @xmath659 enters into @xmath688 and never returns to @xmath389 .
since it ends up at infinity again , the contour has to leave @xmath688 again , and it will do so along one of the arcs @xmath682 , with @xmath689 .
we let @xmath674 be the _ first point _
after @xmath673 which is on one of the @xmath682 again , say @xmath690 .
if @xmath691 , then we stop .
if @xmath692 , then we continue and we let @xmath675 be the _ last point _ on @xmath659 that is on @xmath693 .
after @xmath675 the contour is going into @xmath688 again , since the contour has to end in the sector @xmath657 and @xmath692 .
the contour will meet one of the @xmath682 again , and we let @xmath676 be the _ first point _
after @xmath675 which is on one of the @xmath682 , say @xmath694
. then @xmath695 and @xmath696 , since @xmath673 is the last point on @xmath389 and @xmath675 is the last point on @xmath693 , and @xmath676 comes after these two points . if @xmath697 then we stop , and otherwise we continue , see figure [ level_curves_7 ] for an illustration .
continuing this process , we find a sequence of distinct indices @xmath698 for some @xmath699 , where @xmath700 and points @xmath701 on the contour @xmath659 with @xmath702 furthermore , if we use @xmath703 to denote the part of @xmath659 lying strictly between two points @xmath332 and @xmath704 of the contour , then @xmath705 the numbers @xmath706 depend on the contour @xmath659 and so they depend on @xmath120 .
but these numbers are always positive integers smaller than or equal to @xmath51 , so there are just finite possible choices for them . by passing to a subsequence of the sequence @xmath576 , we may assume they are independent of @xmath120 . the set @xmath707 = \widetilde{\gamma}_n(\xi_{i , n } , \eta_{i , n } ) \cup \{\xi_{i , n } , \eta_{i , n}\}\ ] ] is a connected closed subset of @xmath708 lying in the compact set @xmath709 . by compactness of the hausdorff metric
, there is a subsequence , say @xmath710 , such that the limit @xmath711.\ ] ] exists in the hausdorff metric for every @xmath712 .
the limit @xmath713 is connected , because this property is preserved by taking limits in the hausdorff metric if all the sets involved are contained in a large , but fixed , compact set of @xmath10 , in this case @xmath714 , see . also @xmath715 as follows from and the fact @xmath716 \subset \gamma_n$ ] . by compactness of @xmath682
we may also assume ( by taking a further subsequence , if necessary ) that the sequences @xmath717 and @xmath718 for @xmath719 converge along the same subsequence @xmath720 , say @xmath721 g. lvarez , l. martnez alonso , and e. medina , _ determination of s - curves with applications to the theory of non - hermitian orthogonal polynomials _
, j. stat .
theory exp . , * 2013 * ( 2013 ) , no .
06 , p06006 .
a. i. aptekarev , a. b. j. kuijlaars , and w. van assche , _ asymptotics of hermite - pad rational approximants for two analytic functions with separated pairs of branch points ( case of genus 0 ) _ , int .
imrp 2008 , art .
i d rpm007 , 128 pp .
a. i. aptekarev , v. g. lysov , and d. n. tulyakov , random matrices with an external source and the asymptotics of multiple orthogonal polynomials , mat .
202 ( 2011 ) , no .
english translation in sb .
( 2011 ) , 155206 .
m. bertola and a. tovbis , _ asymptotics of orthogonal polynomials with complex varying quartic weight : global structure , critical point behaviour and the first painlev equation _ , preprint arxiv:1108.0321 .
n. m. ercolani and k. d. t - r mclaughlin , _ a quick derivation of the loop equations for random matrices _ , probability , geometry and integrable systems , math .
55 , cambridge univ . press , cambridge , 2008 , pp .
185198 .
a. a. gonchar and e. a. rakhmanov , _ equilibrium distributions and the rate of rational approximation of analytic functions _ ,
( n.s . ) * 134(176 ) * ( 1987 ) , no . 3 , 306352 .
english translation in math .
ussr - sb .
* 62 * ( 1989 ) , no . 2 , 305348 . a. b. j. kuijlaars , _ multiple orthogonal polynomials in random matrix theory _ , in : proceedings of the international congress of mathematicians .
volume iii , ( r. bhatia ed . ) , hindustan book agency , new delhi , india , 2010 , pp . 14171432 .
a. martnez - finkelshtein , e. a. rakhmanov , and s. p. suetin , _ variation of equilibrium energy and the s - property of a stationary compact set _ , mat .
* 202 * ( 2011 ) , 113136 .
english transl . in sb
* 202 * ( 2011 ) , 18311852 .
e. a. rakhmanov and s. p. suetin , _ the distribution of the zeros of the hermite - pad polynomials for a pair of functions forming a nikishin system _ , mat .
* 204 * ( 2013 ) , no . 9 , 115160 .
english transl . in sb
* 204 * ( 2013 ) , 13471390 . , _ orthogonal polynomials with respect to complex - valued measures _ , orthogonal polynomials and their applications ( erice , 1990 ) , imacs ann .
, vol . 9 , baltzer , basel , 1991 , pp . | curves in the complex plane that satisfy the s - property were first introduced by stahl and they were further studied by gonchar and rakhmanov in the 1980s .
rakhmanov recently showed the existence of curves with the s - property in a harmonic external field by means of a max - min variational problem in logarithmic potential theory .
this is done in a fairly general setting , which however does not include the important special case of an external field @xmath0 where @xmath1 is a polynomial of degree @xmath2 . in this paper
we give a detailed proof of the existence of a curve with the s - property in the external field @xmath0 within the collection of all curves that connect two or more pre - assigned directions at infinity in which @xmath3 .
our method of proof is very much based on the works of rakhmanov on the max - min variational problem and of martnez - finkelshtein and rakhmanov on critical measures . |
according to the dirac s arguments , the large dimensionless numbers provided by atomic physics and astronomy of our universe are connected with each other @xcite .
these numbers include : ( i ) the ratio of the electric to the gravitational force between an electron and a proton @xmath8 ; ( ii ) the age of the universe expressed in terms of the atomic unit @xmath9 , where @xmath10 is the hubble constant ; ( iii ) the mass of the part of the universe which is receding from with a velocity @xmath11 , expressed in the units of the proton mass , say , @xmath12 .
dirac introduced a large number hypothesis @xmath13 based on this hypothesis , dirac suggested a model of cosmology with a varying gravitational constant @xmath14 and an increase in the amount of matter in the universe . based on the large number hypothesis , a number of cosmological models with a varying gravitational constant have been proposed @xcite-@xcite .
however most of them met many difficulties . noting that gravitational constant @xmath14 can not vary in general relativity ( gr ) , the first difficulty is that one must explain the contradiction between einstein gr and dirac cosmology .
though much effort , for example , milne two time scale hypothesis @xcite , weyl s geometry @xcite etc .
, has been devoted to reconcile the requirements of these two theories , it is still an open question to establish a theory which can unify the dirac cosmology and einstein gr .
the second difficulty comes from experiments .
almost all experiments at the scales of solar system and galaxies have not found the variation of @xmath14 @xcite-@xcite .
a possible variation of @xmath14 has been investigated with no success through geophysical and astronomical observations . from experimental results
one tends to believe that @xmath14 is a constant for local system with large scale .
the third difficulty belongs to the conservation of energy and momentum .
usually we use a perfect - fluid energy - momentum tensor to describe the matter of universe and it is conserved in the cosmic evolution .
the addition creation , the new matter created uniformly in the whole space , and the multiplication creation , the new matter created in regions where old matters exist , must come from other mechanism as suggested by dirac .
many years ago a possible unified theory of dirac cosmology and gr to overcome the above difficulties was suggested in @xcite .
the basic idea is as follows .
though one would expect a constant value of @xmath14 at the local system such as solar system , binary system and galaxies , it must be stressed that the cosmological observations still can not put strong limits on the time variation of @xmath14 in the cosmological scale , especially at the late time of the universe evolution @xcite@xcite . though einstein gr has been proved to be correct by many experiments such as the excess perihelion precession of mercury , gravitational redshift etc . at the scales of local system
, it probably needs to be modified in the cosmological scale , especially if we want to use it to explain the recent observational result of the acceleration .
using the idea of dirac cosmology with the variation of @xmath14 one could provide a possible way to modify the einstein gr . as is well known , the lagrangian density of einstein gr with a cosmological term is @xmath15 where @xmath16 is a dimensionless parameter .
the cosmological constant @xmath17 can be generally explained as the background fluctuation of the cosmological vacuum , and @xmath18 is the 4-dimensional scalar curvature . in ordinary astrophysical problems of local systems ,
the magnitude of @xmath18 is about @xmath19 .
since the vacuum energy density is much smaller compared to the density @xmath20 in local system , @xmath21 is a very small quantity and can be neglected .
but in cosmological problems , because @xmath20 is small and has the same order as that of the vacuum energy density , @xmath21 can not be neglected . in the limit of vacuum or the matter domination area , @xmath22 , @xmath23
, @xmath21 can attain the magnitude @xmath24 .
this means that @xmath21 plays an important role at cosmological scale .
however , there is only a first order term of @xmath21 in the equations of einstein cosmology . instead of the factor @xmath25
, we argue that in a perfect cosmological theory , the lagrangian density of the gravitational field could contain higher - order terms of @xmath21 .
we take the lagrangian density as @xmath26 where @xmath27 the einstein action becomes @xmath28 starting from eq.([e5 ] ) , we can establish a dirac cosmology with varying @xmath14 and matter creation .
obviously , this theory can unify gr and dirac cosmology because in a local system with large scale , @xmath29 and the theory reduces to einstein gr . only at cosmological scale , the terms @xmath21 , @xmath30 ... become important and our theory reduces to dirac cosmology .
recently , many authors have considered the terms @xmath31 and/or @xmath32 on gravity @xcite-@xcite . but none of them has connected with dirac cosmology and the large number hypothesis . in this paper , after introducing the higher order terms of @xmath21 , we will establish a unified theory of gr and dirac cosmology and explain the acceleration of the contemporary universe , variation of @xmath14 , cosmological supernova type ia data , and the creation of matter .
we find that the addition creation of matter looks like the dark energy and the multiplication creation of matter like the dark matter .
the organization of this paper is as follows : we will present our model in sec.ii . in sec.iii , by using the present cosmological parameters as our initial conditions , we will calculate numerically the physical quantities of our universe and study the addition creation and multiplication creation . we will prove that our results are in good agreement with present knowledge in cosmology .
finally , a summary and conclusion will be given in sec.iv .
up to the second order of magnitude @xmath33 , the action of our model is @xmath34 where @xmath0 is a expansion parameter and we will adjust it from supernova data later .
the field equation for the metric is then @xmath35 where @xmath36 is the energy - momentum tensor of matter and @xmath37 is the covariant derivative induced by the metric .
when @xmath38 , the constant curvature vacuum solutions require @xmath39 we find the constant - curvature vacuum solutions are not minkovski space , but rather de sitter space ( @xmath40 ) or anti - de sitter space ( @xmath41 ) .
eq.([e7 ] ) can be rewritten as @xmath42 where @xmath43 is the einstein tensor , and @xmath44 @xmath45 the physical meaning of @xmath1 and @xmath46 is obvious .
we see from eq.([e9 ] ) that @xmath1 plays the role of gravitational constant and @xmath46 the creation of energy - momentum in the universe .
they depend on the second order term @xmath47 of the expansion @xmath48 .
if @xmath49 , @xmath50 , @xmath46 reduces to the ordinary vacuum energy cosmological term . since @xmath1 ,
@xmath46 depend on @xmath18 and then are functions of time @xmath51 , this is just the character predicted by dirac large number hypothesis .
futhermore , using eq.([e9 ] ) and the bianchi identity , we get @xmath52 @xmath46 is no longer conserved . to understand the physical meaning of eq.([e12 ] ) , we rewrite it as following : @xmath53 @xmath54 where @xmath55 .
it is clear that @xmath56 is created around the normal matter , while the creation of @xmath57 spreads over the whole universe . in the words of dirac cosmology
, @xmath56 is the multiplication creation having the basic character of dark matter and @xmath58 is the addition creation having the basic character of dark energy .
both of them depend on the variation of @xmath1 . in sec.iii ,
the suggestions that addition and multiplication creations being dark energy and dark matter respectively will be elaborated numerically . employing flat robertson - walker metric @xmath59 and the perfect - fluid energy momentum tensor @xmath60 where @xmath61 is the 4-velocity in fluid - rest frame , @xmath62 , @xmath63 are the energy density and pressure respectively . taking @xmath64 as the equation of state , @xmath65 for matter , @xmath66 for radiation , we find that the field equations become @xmath67 @xmath68 where @xmath69 , @xmath70 , @xmath71 @xmath72 are the energy density and pressure of the created matter , respectively .
comparing to the kinetic term @xmath73 , @xmath74 which corresponds to a potential . here
we only consider the case of positive curvature scalar @xmath18 .
it is easy to show that @xmath18 is always positive in the evolution of the universe with matter and radiation . the kinetic term and the potential determine the evolution of @xmath1 explicitly .
@xmath75 will reach its maximum at @xmath76 $ ] as shown in fig.([f1 ] ) .
we also define @xmath77 , @xmath78 , @xmath79 and @xmath80 @xmath72 as the energy density and pressure of both of the multiplication and addition creation matter .
the components of @xmath81 of eq.([e12.1 ] ) and eq.([e12.2 ] ) can be written as @xmath82 @xmath83 the other components of eq.([e12.1 ] ) and eq.([e12.2])@xmath84 give the identities about pressures @xmath85 and @xmath86 and no further information . since the multiplication creation @xmath87 looks like dark matter , we set @xmath88 eq.([e15])-eq.([e12.5 ] ) and the equation of scalar curvature @xmath89 = 6(\dot h + 2h^2 ) \ ] ] given by the flat robertson - walker metric form a complete set . in principle
, we can solve the set of equations to find the behavior of the cosmological evolution .
we now study the behavior of our universe containing matter and radiation . to find the cosmological solution , we need the initial conditions of eq.([e15])-eq.([e12.5 ] ) in the early universe which
, unfortunately , is little known now .
hence , we have done our numerical computation by using the present cosmological parameters as our initial conditions , which include the hubble parameters constant @xmath90 , the present deceleration @xmath91 , the cosmological constant @xmath92 ( as the scale ) , the density of baryon @xmath93 , the ratio of the baryon - to - matter ( including baryon and multiplication creation ) @xmath94 and the ratio of baryon - to - photon @xmath95@xcite .
we will start from the present epoch and trace back the history of our universe .
fig.[f4 ] shows the variation of @xmath1 in the evolution . the present time is set as @xmath96 .
the three curves correspond to @xmath97 and @xmath98 respectively .
we see that the gravitational constant @xmath1 increases at first when the universe expands and then decreases at the late time .
the ratio of the density of radiation to baryon floats up quickly and the scale factor @xmath99 as @xmath51 approaches the age of the universe , which are shown in fig.[f5 ] and fig.[f6 ] respectively .
therefore , the radiation dominates the early universe .
and the age of the universe given by @xmath100 is @xmath101 , @xmath102 and @xmath103 for @xmath97 and @xmath98 respectively . since there is no mechanism of inflation in our model , it is easy to understand that our model can consider the universe back to the epoch of radiation domination .
fig.[f7 ] shows the evolution of the deceleration factor @xmath104 .
we see that the universe decelerates in the early era and gradually stops decelerating and starts accelerating .
this is in a good agreement with what we understand the universe nowadays . through the deceleration factor @xmath104
, we notice that the scalar curvature @xmath18 is always positive . obviously , @xmath0 is the parameter in our model which needs to be adjusted . to determine this parameter ,
let us compare our model with the supernova data .
noting that the uncertainty of the present value of the deceleration factor @xmath105 is still large , we will give a better estimation of the parameter @xmath0 and @xmath105 to fit the supernova data .
fig.[f8 ] presents the supernova data as a phase portrait of the universe @xcite@xcite , where the universe was decelerating at high redshift and started accelerating when it was about two - third of the present size .
our model is favored by this set of data : @xmath106 and @xmath107 . comparing fig.[f8 ] and
the curves given by fig.2 of ref@xcite , we find that the fit of our curve is better than that of the standard model .
fig.[f9 ] shows the evolution of the pressure of the density of the addition creation matter with @xmath108 .
the pressure is highly positive in the beginning of the universe , falls quickly blow zero and has a negative value in the recent and present era .
this negative pressure looks like the character of the dark energy .
if we take @xmath109 as the equation of state of addition creation , we show the parameter of the pressure - to - density of the addition creation @xmath110 versus time @xmath51 in fig.[f10 ] and find @xmath111 which is much smaller than @xmath112 and is consistent with the requirement of dark energy to explain the accelerating universe . to compare our results with other models , we have also calculated the ratio of the density of addition creation and find @xmath113 and that of the density of matter ( multiplication creation and normal matter ) in our model reads @xmath114 these results have the same magnitude as that dark energy and dark matter respectively .
hence , we suggest that the dark energy comes from the addition creation and the dark matter from the multiplication creation .
we have suggested a model to unify the einstein gr and dirac cosmology . in local system , our theory reduces to gr , but in the cosmological scale , our theory refers to the dirac cosmology .
the variation of the gravitational constant comes from the scalar curvature . in the local system ,
the variation of the gravitational constant is negligible .
this result is in good agreement with present experiments .
but in the cosmological scale , the change of @xmath1 is remarkable .
the acceleration of the present universe is a proof of the decrease of the gravitational constant @xmath1 , because the decrease of @xmath1 corresponds to an effective repulsion . after introducing the term @xmath115 in our theory and
fit the adjust parameter @xmath0 by supernova data , we have calculated the physical quantities of @xmath116 , @xmath2 , @xmath3 , @xmath4 by using the data of present epoch as the initial conditions .
we have found that the results are self - consistent and in good agreement with present knowledge of cosmology . according to dirac large number hypothesis ,
matter will be created in the universe .
we have calculated the matter comes from addition creation and that from multiplication .
an interesting picture in our theory is that the addition creation , which spreads over the universe , looks like the dark energy and the multiplication creation , which clusters around the normal matter , like the dark matter .
we have found that the pressure of addition creation has a big positive value initially and fall down quickly to become a negative value .
it has a negative pressure region corresponding @xmath117 .
the equation of state parameter @xmath118 at present .
this value is in agreement with present dark energy models .
we have also calculated the ratio of the density of the addition creation and found @xmath6 .
the same parameter but for multiplication creation , radiation and normal matter has also been computed , which reads @xmath119 . both of them have the same magnitude of the observational value of the dark energy and matter .
this result suggests that the dark energy and dark matter are just the addition creation and the multiplication creation in dirac cosmology .
finally , we would like to emphasize that this model can not be extended to the big bang epoch .
we have not add the terms with @xmath120 ( or @xmath121 ) in the lagrangian density . obviously , these terms are very important to the very early universe , especially in the inflationary epoch
. our model can only be used in the time evolution regions starting from the radiation dominated epoch to the present time .
this work was supported in part by nnsf of china , by the national basic research program 2003cb716300 and the foundation of education of ministry of china .
b. wang s work was also supported in part by shanghai education commission .
su and s.w .
zhang , _ proc . of the 3rd grossmann meeting on gen .
p1381 , sci . press and north - holland pub company ( 1983 ) + r.k . su and s.w .
zhang , _ acta . math .
* 3 * ( 1983 ) 321 , _
kexue tongbao ( sci .
bullium of china ) _ * 27 * ( 1982 ) 944 g. brans and r.h .
dicke , _ phys rev . _
* 124 * ( 1961 ) 925 f. hoyle and j.v .
narlikav , _ proc .
_ * a227 * ( 1964 ) 1 h.w .
peng , gr - qc/0401105 , gr - qc/0405002 v. canuto , et .
al . , _ phys .
* 39 * ( 1977 ) 429 t. damour and k. nordtvedt , jr .
* 70 * ( 1993 ) 2217 , _ phys
_ * d48 * ( 1993 ) 3436 t. damour , f. piazza and g. veneziano , _ phys .
* 89 * ( 2002 ) 081601 , _ phys .
_ * d66 * ( 2002 ) 046007 k. nordtvedt , jr .
, gr - qc/0301024 s. carneiro and j.a.s .
lima , gr - qc/0405141 , _ gen .
* 26 * ( 1994 ) 909 j.p .
ozan , _ rev .
* 75 * ( 2003 ) 403 j.g .
williams , s.g .
turyshev and d.h .
boggs , _ phys .
_ * 93 * ( 2004 ) 261101 g. magnano and l.m .
sokolowski , _ phys .
* d50 * ( 1994 ) 5039 a. dobado and a.l .
maroto , _ phys .
_ * d52 * ( 1995 ) 1895 s. capozziello , v.f .
cardone , s. cardoni and a. troisi , _ phys .
lett . _ * a326 * ( 2004 ) 292 s.m .
carroll , v. duvvuri , m. trodden and m.s .
turner , _ phys .
* d70 * ( 2004 ) 043528 s. nojiri and s.d .
odintsov , _ phys .
rev . _ * d68 * ( 2003 ) 123512 a.a .
starobinsky , _ phys .
lett . _ * b91 * ( 1980 ) 99 t. padmanabhan , t.r .
choudhury , _ mon . not .
* 344 * ( 2003 ) 828 t.r .
choudhury and t. padmanabhan , _ astron .
astrophys . _
* 429 * ( 2005 ) 807 c.l .
bennett , et al . ,
_ apjs _ * 148 * ( 2003 ) 1 | a model is suggested to unify the einstein gr and dirac cosmology .
there is one adjusted parameter @xmath0 in our model . after adjusting the parameter @xmath0 in the model by using the supernova data , we have calculated the gravitational constant @xmath1 and the physical quantities of @xmath2 , @xmath3 and @xmath4 by using the present day quantities as the initial conditions and found that the equation of state parameter @xmath5 equals to -0.83 , the ratio of the density of the addition creation @xmath6 and the ratio of the density of the matter including multiplication creation , radiation and normal matter @xmath7 at present .
the results are self - consistent and in good agreement with present knowledge in cosmology .
these results suggest that the addition creation and multiplication creation in dirac cosmology play the role of the dark energy and dark matter . |
the iron k - shell lines near the galactic center ( gcdx ) was first detected with the ginga satellite @xcite .
the line energy is about 6.7 kev and the emission extends about 12@xmath3 over the galactic center ( gc ) region .
the asca satellite observed more details of the gc region , and confirmed the nature of the gcdx @xcite .
moreover , the asca ccd resolved the iron k - shell line detected with ginga into three distinct lines at 6.4 , 6.7 and 6.9 kev @xcite .
the 6.7 and 6.9 kev lines are likely due to he - like ( fe ) and hydrogenic ( fe ) ions of iron .
there are three plausible scenarios that have been proposed for the origin of the iron k - shell lines .
one is collisional excitation ( ce ) by low energy electrons ( lee ) , similar to the galactic ridge emission .
this process , however , contributes mainly to the 6.4 kev line emission , and not to the 6.7 and 6.9 kev lines ( valinia et al .
2000 ) . with respect to the 6.7 kev and 6.9 kev lines ,
ce in a thin , high temperature plasma , and charge exchange ( cx ) recombination are both considered . for the ce case ,
the best - fit thin thermal plasma temperature of the gcdx spectrum was @xmath410 kev .
if the gcdx is really due to a thin hot and uniformly distributed plasma , the total thermal energy is estimated to be 10@xmath5 erg , or equivalent to 10@xmath610@xmath7 supernovae explosions .
the temperature of @xmath410 kev is higher than that bounded by the galactic gravity , hence the plasma should escape from the gc region .
the time scale estimated by the plasma size ( 1@xmath3 or @xmath4 150 pc at 8.5 kpc ) divided by the sound velocity of the @xmath410 kev plasma is @xmath410@xmath8 years .
therefore the energy of @xmath410@xmath5 erg should be supplied in the past @xmath410@xmath8 years , e.g. 1 sn every 1001000 year is required in the gc region .
this huge energy budget in the gc region suggests that cx recombination may be a more reasonable explanation for the gcdx . in this scenario , during a collision between neutral hydrogen and fe or fe , the bound hydrogen electron is captured into an excited state of the iron , and then the excited state decays producing an x - ray .
the x - ray spectral signature is distinct from the signature produced by ce in a collisional ionization equilibrium ( cie ) plasma , such as the high - temperature plasma mentioned above . in the cx case ,
the forbidden line at @xmath9 ev , is stronger than the resonance line at @xmath10 ev @xcite , whereas the resonance line is stronger in the case of a plasma in cie .
thus , even if the resonance and forbidden lines are not completely resolved , the energy centroid of the k@xmath0 line of fe produced by cx will be lower than produced by ce .
we also note that an energy shift towards lower energy as the cx would be observed in the case of x - ray emission following recombination with low energy electrons , as in a photo - ionized plasma @xcite .
the 6.4 kev line is likely to be k@xmath0 from the neutral and/or a blend of low charge states of iron ( here , fe ) .
in fact , the 6.4 kev line is localized near molecular clouds such as the radio complex sgr b2 and the radio arc regions .
sgr b2 , the most massive giant molecular cloud with ultra compact h ( uch ) and many maser sources , has been extensively studied with asca and chandra ( @xcite ; @xcite , 2001 ) . as a result of those studies
, the authors proposed that the 6.4 kev line arises in an x - ray reflection nebula ( xrn ) irradiated by sgr a@xmath11 during an exceptionally bright period that occurred about 300 years ago , i.e. , a time span equivalent to the light traveling time between sgr b2 and sgr a@xmath11 .
this putative high activity of sgr a@xmath11 could provide the energy supply to the gcdx .
accordingly , a high quality observation of the iron and nickel k - shell lines can provide key information for understanding the origin of the gcdx . using the x - ray imaging spectrometer ( xis ) ( koyama et al .
2006a ) , we have measured x - ray emission for the spectral band containing these lines .
the xis is the x - ray ccd camera system on board suzaku ( @xcite ) .
the excellent energy resolution and well - calibrated performance for diffuse sources , coupled with the low background orbit of suzaku , make the xis well - suited to provide crucial information about the gcdx .
here we present the results of the first measurement of the gcdx with the xis focusing on the k - shell emission from iron and nickel .
[ cols= " < , < , < , < , < , < , < " , ] + @xmath12 from wargelin et al . ( 2005 ) .
+ the relatively high resolution of the xis also affords us the ability to extract crucial information from the line feature of fe ly@xmath0 provides further evidence against the cx process for x - ray production . as discussed in section 4.3 ,
the intrinsic line broadening of ly@xmath0 is nearly zero , and it follows that if cx is the source of the x - ray emission , then recombination would occur when the bare iron velocity is nearly zero , i.e. , the collision energy between the iron ion and neutral hydrogen is small . in such a low - collision energy case ,
electrons are non - statistically captured into levels with large principal quantum numbers ( @xmath13 ) @xcite .
we , however , see no enhancement at the fe rydberg series limit at 9.2 kev .
our measurements gives an upper limit of @xmath14 photons s@xmath15 @xmath16 for the high-@xmath17 rydberg series line emission ( we assumed a gaussian line with width of 30 ev ) , only about 6% of the intensity of ly@xmath0 .
this upper limit corresponds to a lower limit of collision energy of 100 ev amu@xmath15 ( @xcite ; @xcite ) between bare fe nuclei and neutral h. for that collision energy , the velocity of bare iron nuclei must be larger than @xmath4150 km s@xmath15
. therefore the cx process , if present in the gcdx , gives a lower limit on the velocity of @xmath4150 km s@xmath15 .
@xcite suggested that the cx process may have the maximum cross section at a higher iron velocity of @xmath18 km s@xmath15 , hence the subsequent decay lines have widths of about 100 ev .
no such large broadening is observed in fe ly@xmath0 .
together with the discussion of section 4.4 , we therefore propose that the 6.7 kev and 6.9 kev lines in the gcdx are likely due to a cie plasma and would not arise from a cx process .
the k - shell lines of fe k@xmath0 and fe ly@xmath0 were discovered with the data from asca @xcite , chandra @xcite and xmm - newton @xcite . with suzaku
, we have , for the first time , detected ni k@xmath0 , and fe k@xmath1 and fe ly@xmath1 .
we have also tentatively identified fe k@xmath2 and fe ly@xmath2 ( see table 2 ) . as discussed in sections 4.4 and 4.5 ,
the emission lines from these highly ionized atoms are likely due to collisional excitation in a thin thermal plasma .
the best - fit flux ratio of fe - ly@xmath0/fe - k@xmath0 is [email protected] , which provides an ionization temperature of [email protected] kev .
the best - fit flux ratio of fe - k@xmath1/fe - k@xmath0 is 0.092@xmath20 , which gives the electron temperature of 6.6@xmath21 kev .
although the error is still large , this electron temperature is similar to the ionization temperature .
the line flux ratio of ni - k@xmath0/fe - k@xmath0 is 0.078 . with the further assumption that the relative abundances of iron and nickel are proportional to the solar value
, this ratio is converted to the plasma temperature of @xmath45.4 kev .
we therefore conclude that the gcdx plasma has a temperature of @xmath457 kev in collisional ionization equilibrium . the observed flux ratios of fe - ly@xmath1/fe - ly@xmath0 , fe - k@xmath2/fe - k@xmath0 fr and fe - ly@xmath2/fe - ly@xmath0 from are consistent with this plasma condition .
although we show that the gcdx is attributable to a high temperature plasma of @xmath4 57 kev in collisional ionization equilibrium , whether the origin is diffuse or the integrated emission of a large number of unresolved point sources , is not yet clear .
we therefore investigated the spatial variation of the gcdx in temperature and flux .
we made a series of x - ray spectra in @xmath22 arc - min rectangles along the galactic plane of @xmath23 ( figure [ fig : boxreg ] ) , then extracted the fluxes of the 6.7 and 6.9 kev lines , applying the same model as is given in table [ tabl : lines ] .
the plots of the flux ratio between the 6.7 and 6.9 kev lines , and the fluxes of the 6.7 kev line are given in figure [ fig : ratio ] and figure [ fig : ps_diff ] , respectively , in which the flux and flux ratio of the bright snr sgr a east are excluded .
we see that the 6.7 and 6.9 kev line ratio is constant at @xmath24 in the negative longitude region from @xmath25 to @xmath26 , but decreases to @xmath27 at the positive longitude near at @xmath28
. the ratios of 0.38 and 0.30 suggest ionization temperatures of 6.8 kev and 6.2 kev , respectively .
the smooth and monotonic variation of the ionization temperature along the galactic longitude argues that the gcdx is not due to the superposition of the outputs of a large number of unresolved point sources .
we simulated the point - source flux in the 4.78.0 kev band observed with xis , putting the positions and fluxes ( 2.08.0 kev band ) of the chandra cataloged sources @xcite in the same rectangles of figure 4 . to preserve completeness limit , we selected the sources with flux larger than 5@xmath29 photons @xmath16 s@xmath15 in the 2.08.0 kev band .
after convolving the response of the xrt + xis , we extracted the integrated point - source fluxes from each rectangle .
the results are plotted in the same figure of the 6.7 kev line ( figure [ fig : ps_diff ] ) .
the integrated point - source flux is symmetrically distributed between positive and negative galactic longitude with a sharp peak at sgr a@xmath11 ( @xmath30 ) , which is significantly different from the 6.7 kev line flux distribution .
the 6.7 kev line flux shows an asymmetric distribution ( larger flux at the positive galactic longitude than the negative side ) and a gradual increase towards sgr a@xmath11 .
these facts suggest that the major fraction of the 6.7 kev line flux is not due to the integrated emission of unresolved point sources , but is truly diffuse .
( 70mm,50mm)figure4.eps ( 70mm,50mm)figure5.eps ( 70mm,50mm)figure6.eps from the line diagnostics , the gcdx is most likely produced by a thin hot plasma with a temperature of @xmath457 kev in cie , while the bremsstrahlung temperature determined with the continuum shape in the 5.511.5 kev band is @xmath415 kev(see table 2 ) , significantly higher than @xmath457 kev .
this suggests that the continuum flux may contain an additional hard component .
we therefore fit the spectrum with a model of a cie plasma and power - law component with 3 gaussians for the k - shell emissions from fe and ni . in this model ,
the k - edge absorption with @xmath31 is also implemented .
although its contribution is minor , we added the model spectrum of the cosmic x - ray background ( cxb ) .
as discussed in section 3.2 , there are systematic errors in the line width ( @xmath430 ev ) and center energy ( @xmath32 ev ) .
we therefore introduced an artificial random motion and small red - shift to cancel these systematic errors .
the fit is reasonably good and the best - fit results and parameters are given in figure [ fig : apecfit ] and table [ tabl : vapec ] .
we note that the forbidden line of ni and some of the satellite lines are missing in the apec model , hence the ni abundance may be overestimated .
( 70mm,50mm)figure7.eps lcccc +
parameter & unit & & & value + @xmath33 & @xmath16 & & & 6@xmath34 ( fixed ) + @xmath31 & @xmath16 & & & 8.5@xmath35 + @xmath36 & kev & & & 6.5@xmath37 + @xmath38 & solar & & & 1.2@xmath39 + @xmath40 & solar & & & 2.1@xmath41 + velocity & km s@xmath15 & & & 1192 + redshift & & & & 6.0@xmath42 + + @xmath43 & & & & 1.4@xmath44 + flux at 8 kev & photons s@xmath15 @xmath45kev@xmath15 & & & 4.5@xmath46 + + center energy & & width & intensity + ( ev ) & line & energy ( ev ) & ( ev ) & ( photons s@xmath15 @xmath16 ) + 6409@xmath47 & fe@xmath48 k@xmath0 & & 36@xmath49 & 4.41@xmath50 + 7069.1@xmath11 & fe@xmath48 k@xmath1 & & 39@xmath11&7.8@xmath51 + 7492@xmath52 & ni@xmath48 k@xmath0 & & 0(@xmath5330 ) & 2.3@xmath54 + + @xmath48 or low ionization state + @xmath11 fixed to 1.103 @xmath55 ( fe k@xmath0 ) .
+ one notable finding is the presence of power - law hard tail . the power - law index @xmath43 is sensitive to the slope of the high energy continuum . because the effective area of the xrt - xis combination drops rapidly at high energy , our determination of the power - law index @xmath43 has a relatively large statistical error of @xmath561.4@xmath44 .
the low numbers of counts make it challenging to determine the error contribution from systematic effects . to provide a more reliable estimate of the systematic error , we checked it using the spectrum of the crab nebula .
we examined the variation of the best - fit @xmath43 by fitting the crab spectrum in the 410 , 510 , 610 and 710 kev bands , and found no systematic variations with the increasing energy band .
the mean variations of @xmath43 are within 5% for all the xiss , which is far smaller than the statistical error of @xmath440% .
since the flux level of the nxbg becomes comparable to the hard tail flux at 810 kev , it may be argued that the error due to the nxbg - subtraction should not be ignored .
the possible flux error of the nxbg is estimated to be @xmath573% ( koyama et al . 2006a ) . including this error in the spectrum , the best - fit @xmath43 are 1.4 ( adding 3% of the nxbg ) and 1.2 ( 3% reduction of the nxbg ) . further systematic analysis of the nxbg found a hint that the flux of the night earth data is smaller than that from a blank sky .
the details of the flux difference is still in study , but is at a level of @xmath58 10% at the 811 kev band .
if we use a nxbg of 1.1@xmath59 of the night earth , then the best - fit @xmath43 becomes 1.5 .
thus any possible systematic error of nxbg does not change the conclusion that a flat hard tail is present in the gcdx .
in addition , the flux of the cxb is more than one order of magnitude smaller than the hard tail in the hard x - ray band , so that the fluctuation and/or error of the cxb can be ignored .
a possible contributor of the power - law hard tail is non - thermal filament structures found with chandra ( @xcite ; @xcite ; @xcite ) .
in fact , the power - law flux in the positive galactic longitude ( gc_sr1 ) is significantly larger than that of the negative galactic longitude ( gc_sr2 ) , which is a trend similar to the population of the non - thermal filaments ( @xcite ; @xcite ; @xcite ) .
the origin of the non - thermal emission is a key for understanding the presence of high energy particles , which are suggested by the detection of very high energy gamma - rays ( @xcite ; @xcite ; @xcite ) .
further analysis of the xis data in combination with those of the hard x - ray detector(hxd ) ( @xcite ) could yield an important contribution to the study of this hard component .
since the difference between the theoretical ( 6966 ev ) and observed ( 6969@xmath60 ev ) values of fe ly@xmath0 is only 3 ev , we can assume that the detector gain and uniformity are well - calibrated to study features at this level of accuracy .
2 . the abundance of fe in the interstellar gas near the gc appears to be 3.5 times solar .
the 6.4 kev ( k@xmath0 of neutral iron ) line from the gc region could be contaminated by those of low ionization states , but still characteristic of gas at temperature much lower than those in which the helium - like lines arise .
the observed line centroid energy of fe k@xmath0 is 6680@xmath191 ev , which is consistent with that of the collisional ionization plasma .
the gcdx spectrum is consistent with that of a cie plasma with mean temperature 57 kev .
the plasma temperature of gxdx is constant at 6.8 kev in the galactic longitude region from @xmath61 to @xmath62 , while exhibiting a significant decrease to 6.2 kev near @xmath63 .
the flux distribution of the 6.7 kev line along the galactic plane is different from that of the integrated point sources .
we found a clear hard tail in the gcdx .
the authors express sincere thanks to all the suzaku team members , especially a. bamba , h. uchiyama , h. yamaguchi , and h. mori for their comments and supports
. y.h . and
h.n . are supported by jsps research fellowship for young scientists .
this work is supported by the grants - in - aid for the 21st century coe `` center for diversity and universality in physics '' and also by grants - in - aid ( no 18204015 ) of the ministry of education , culture , sports , science and technology ( mext ) .
work by uc llnl was performed under the auspices of the u.s .
department of energy by university of california , lawrence livermore national laboratory under contract w-7405-eng-48 .
since the gcdx emits strong lines covering over the entire field view of the xis , we can calibrate the gain and its variation in the full xis imaging area ( ia ) .
for this purpose , we divide each xis area into 4 sections along the acty - axis and designate as no.1 to 4 in the order of the distance from the signal read - out nodes ( no.4 corresponds the largest acty , while no.1 is the smallest ) . the definition of acty is given in koyama et al .
( 2006a ) .
figures [ fig:4xis - cti0sheka ] , [ fig:4xis - cti0feneuka ] and [ fig:4xis - cti0feheka ] are the plots of the line center energy of s k@xmath0 , fe k@xmath0 and fe k@xmath0 , respectively . from figures [ fig:4xis - cti0sheka ] , [ fig:4xis - cti0feneuka ] and [ fig:4xis - cti0feheka ] , we see a clear trend that the line center energy is lower at longer distance from the read - out nodes . the effect is clearly instrumental ( rather than astronomical ) , since the trend is observed in all four xis detectors , with differing read - out orientations on the sky ( koyama et al .
therefore , we can safely conclude that the systematic trend that the center energy decreases as increasing acty is due to the charge transfer inefficiency ( cti ) . assuming that cti is a linear function of the distance from the read - out nodes
, we correct the cti so that the center energies of the three k@xmath0 lines become constant along the acty axis .
the cti - corrected plots are given in figures [ fig:4xis - aftctisheka ] , [ fig:4xis - aftctifeneuka ] and [ fig:4xis - aftctifeheka ] .
the systematic decreases of the center energy along the acty - axis disappear , demonstrating that the cti correction is properly made .
we should note that the cti at 6.4 and 6.7 kev is consistent with that at 5.9 kev from the calibration sources reported by the xis team ( koyama et al .
note also that we have corrected the cti for each xis independently , but the cti is assumed to be identical in the 4-segments .
also absolute energy gain tuning is not yet made at this stage .
we find significant gain differences from segment to segment in each xis and from xis to xis in the segment - averaged energy .
the absolute gains for the segments including the calibration sources are fine - tuned to the energy of the calibration lines ( k@xmath0 and k@xmath1 lines of mn ) the gains of the other segments are adjusted to match the center energies of the 3 k - shell lines ( s , fe and fe ) to those of the fine - tuned energies of the calibration segments .
the results are given in figures [ fig:4xis - gainsheka ] , [ 4xis - gainfeneuka ] and [ fig:4xis - gainfeheka ] .
note that the vertical axis scales are finer than those of figures [ fig:4xis - cti0sheka][fig:4xis - actxfeheka ] .
we see the energy variations from xis to xis and/or from segment to segment become smaller than those of the non - correction in both the s and fe k@xmath0-lines . | we have observed the diffuse x - ray emission from the galactic center ( gc ) using the x - ray imaging spectrometer ( xis ) on suzaku .
the high - energy resolution and the low - background orbit provide excellent spectra of the gc diffuse x - rays ( gcdx ) .
the xis found many emission lines in the gcdx near the energy of k - shell transitions of iron and nickel .
the most pronounced features are fe k@xmath0 at 6.4 kev and k - shell absorption edge at 7.1 kev , which are from neutral and/or low ionization states of iron , and the k - shell lines at 6.7 kev and 6.9 kev from he - like ( fe k@xmath0 ) and hydrogenic ( fe ly@xmath0 ) ions of iron . in addition
, k@xmath0 lines from neutral or low ionization nickel ( ni k@xmath0 ) and he - like nickel ( ni k@xmath0 ) , and fe k@xmath1 , fe k@xmath1 , fe ly@xmath1 , fe k@xmath2 and fe ly@xmath2 are detected for the first time .
the line center energies and widths of fe k@xmath0 and fe ly@xmath0 favor a collisional excitation ( ce ) plasma for the origin of the gcdx .
the electron temperature determined from the line flux ratio of fe - k@xmath0/ fe - k@xmath1 is similar to the ionization temperature determined from that of fe - k@xmath0/fe - ly@xmath0 .
thus it would appear that the gcdx plasma is close to ionization equilibrium .
the 6.7 kev flux and temperature distribution to the galactic longitude is smooth and monotonic , in contrast to the integrated point source flux distribution .
these facts support the hypothesis that the gcdx is truly diffuse emission rather than the integration of the outputs of a large number of unresolved point sources .
in addition , our results demonstrate that the chemical composition of fe in the interstellar gas near the gc is constrained to be about 3.5 times solar . |
cp asymmetry measurements in nonleptonic @xmath0 decays have been receiving considerable attention over the past several years since the recent experimental measurements in some decay channels are in disagreement with naive estimates of the standard model ( sm ) .
one of the important examples is the direct cp asymmetries in @xmath1 decays @xcite .
current world averages of the cp asymmetries in @xmath1 decays are given by @xcite : [ acp_exp ] a_cp(b^0 k^^ ) & = & -0.086 0.007 , a_cp(b^k^^0 ) & = & 0.040 0.021 , a_cp(b^k^0 ^ ) & = & -0.015 0.012 .
however , the naive factorization assumption predicts @xmath4 @xcite , which is inconsistent with the current data in eq .
( [ acp_exp ] ) .
this discrepancy can be explained by enhancing the smaller diagrams such as @xmath5 and @xmath6 with a sizable strong phase through the sm fit to the @xmath7 data @xcite , where @xmath5 and @xmath6 stand for the color - suppressed tree and electroweak penguin amplitudes , respectively , in the topological decomposition @xcite .
nonetheless , such enhancement of subdominant diagrams in the sm is not fully understood theoretically and also may not be sufficient to resolve other puzzles simultaneously in the @xmath8 meson system @xcite . alternatively to the sm fit , since the prediction given in ref .
@xcite did not incorporate all possible hadronic uncertainties , these decay modes have been also studied within the sm in the framework of different factorization approaches such as qcd factorization @xcite , perturbative qcd ( pqcd ) @xcite , and soft - collinear effective theory @xcite . even under such factorization assumptions , however , the above data have not been fully explained as well . in the sm
, the sizes and patterns of cp violation in various decay modes are governed by a single complex phase which resides in the cabibbo - kobayashi - maskawa ( ckm ) matrix , but such large cp violation effects have not been simply explained with this single parameter in any of those factorization methods in various decay modes simultaneously .
therefore , there has been several efforts to understand such large cp asymmetries beyond the sm with additional cp odd parameters @xcite .
similarly in this paper , we study the new physics ( np ) contributions to the direct cp asymmetries in @xmath1 decays as well as to ( semi-)leptonic @xmath8 decays and @xmath8 mixing where np effects could be sizable . in order to minimize hadronic uncertainties
, we consider all relevant tree and penguin contributions even including annihilation types by adopting pqcd approach , and estimate the possible np contributions .
one of the simplest extensions of the sm corresponding to such a scenario with additional cp phases is the nonmanifest ( @xmath9 ) left - right model ( lrm ) with gauge group @xmath10 where @xmath11 is the left(right)-handed quark mixing matrix @xcite . since the lrm has the extended group @xmath12 , there are new parameters such as a right - handed gauge coupling @xmath13 , new charged ( neutral ) gauge bosons @xmath14 ( @xmath15 ) , and the @xmath16 ( @xmath17 ) mixing angle @xmath18 ( @xmath19 ) .
after spontaneous symmetry breaking , the gauge eigenstates @xmath14 mix with @xmath20 to form the mass eigenstates @xmath21 and @xmath22 with masses @xmath23 and @xmath24 , respectively .
similarly , the neutral gauge bosons mix each other @xcite , but we do not present them here because @xmath15 contribution to flavor - changing b decays is negligible .
although tree - level flavor - changing neutral higgs bosons with masses @xmath25 enter into our theory due to gauge invariance , we also neglect their contributions by assuming @xmath26 @xcite . the mixing angle @xmath18 and the ratio @xmath27 of @xmath28 to @xmath29 are restricted by a number of low - energy phenomenological constraints @xcite .
one of the most stringent bounds on @xmath30 was obtained from @xmath31 mixing . if the model has manifest ( @xmath32 ) left - right symmetry ( @xmath33 ) where @xmath11 is the left(right)-handed quark mixing matrix , @xmath34 tev @xcite .
similar bounds were obtained recently by cms and atlas from direct searches for the decay channels of the extra gauge bosons @xmath35 under various assumptions on the right - handed neutrino masses and gauge couplings @xcite .
however , the form of @xmath36 is not necessarily restricted to manifest or pseudomanifest ( @xmath37 ) symmetric - type , where @xmath38 is a diagonal phase matrix @xcite .
if @xmath36 takes one of the following forms , the @xmath14 mass limit can be significantly lowered @xcite , and @xmath39 can be as large as @xmath40 ( for @xmath41 gev ) @xcite : [ vr3 ] ( ccc 1 & 0 & 0 + 0 & 0 & 1 + 0 & 1 & 0 ) , ( ccc 0 & 1 & 0 + 1 & 0 & 0 + 0 & 0 & 1 ) , ( ccc 0 & 1 & 0 + 0 & 0 & 1 + 1 & 0 & 0 ) . as well as the cp - violating observables in @xmath1 decays , we also accommodate a large cp - violating phase in @xmath42 mixing observed at tevatron and a disagreement emerged between the determination of @xmath2 from inclusive and exclusive b decays @xcite . in order to incorporate all of those considerations , we take the following form of @xmath36 as similarly done in ref .
@xcite : v^r = ( ccc ~0 & c_r e^i_1 & s_r e^i_2 + e^i & ~0 & ~0 + ~0 & -s_r e^i_3 & c_r e^i_4 ) , [ vr ] where @xmath43 @xmath44 . here
the matrix elements indicated as @xmath45 may be @xmath46 and unitarity requires @xmath47 .
especially , with this form , the present experimental measurement of the large branching fractions for @xmath48 decays can be explained @xcite , but the right - handed current effect in @xmath49 mixing is negligible so we only consider direct cp asymmetries in @xmath1 decays .
one can of course take different types of @xmath36 without taking into account of @xmath50 , and relevant studies were done earlier in refs .
@xcite .
this paper is organized as follows . in sec .
ii , we briefly discuss some of phenomenological constraints without assuming manifest ( or pseudomanifest ) left - right symmetry .
we present the effective hamiltonian describing @xmath51 and @xmath52 transition in sec .
iii , and obtain @xmath1 decay amplitudes including all relevant tree and penguin contributions in the general lrm in sec .
iv . in sec .
v , we explicitly show the allowed regions of np parameters satisfying the current experimental data , taking into account all the constraints obtained in sec .
finally , we conclude in sec .
we consider the case that @xmath53 masses are not too heavy so that it can be accessible at lhc . without assuming manifest or pseudomanifest left - right symmetry in the general lrm , @xmath53 masses are not highly constrained by low - energy electroweak measurements , but still @xmath53 exchange effects
could be seen in various decay modes , and the bound of its mass could be obtained independent of the form of @xmath36 .
for instance , we can obtain the lower bound on @xmath24 from global analysis of muon decay measurements as follows @xcite : _
g < 0.017 m_w^ > ( g_r / g_l ) 620 . [ bound_mwr ] where @xmath54 . in general , @xmath55 for ordinary higgs representations @xcite . as well as @xmath30 and @xmath56 , we have additional np parameters such as @xmath57 and @xmath58 in the quark sector as shown in eq .
( [ vr ] ) in the general lrm . among those new parameters ,
@xmath57 and @xmath59 can be constrained by the disagreement emerged between the determination of @xmath2 from inclusive and exclusive b decays .
@xmath2 determined in exclusive @xmath8 decays is related to @xmath60 in the lrm as |v_ub|_excl = |v^l_ub||1+_u| |v_ub|_incl|1+_u| , where @xmath61 and @xmath62 @xcite . from the mismatch between the values of @xmath2 extracted from inclusive and exclusive @xmath8 decays ,
we roughly obtain the following @xmath63 bound : [ bound_vub ] -1.55 < _ g s_r(_2 + ) 10 ^ 3 < 0.41 , where @xmath64 . in eq .
( [ vr ] ) , @xmath65 and @xmath66 are constrained by @xmath3 mixing measurements .
the dispersive part of the @xmath3 mixing matrix element in the lrm can be written as [ massmixing ] m_12^s = m_12^sm + m_12^lr = m_12^sm ( 1 + r^s_lr ) , where r^s_lr = , and the explicit form of the effective hamiltonians @xmath67 and
@xmath68 describing the @xmath69 transition in the lrm can be found in refs .
@xcite . following the factorization methods used in ref . @xcite with the given form of @xmath36 in eq .
( [ vr ] ) , we obtain the right - handed current contributions to @xmath3 mixing as [ rlrs ] r^s_lr & & 162 ( ) _ g s_rc_re^-i(_3 - _ 4 ) + 1.70_g s_r e^-i_3 .
the deviation of the present experimental data from the sm predictions on @xmath42 meson mixing gives the following @xmath63 bound @xcite : [ bound_mixing ] 0.86 < |1+r^s_lr| < 1.22 , and we will use this bound together with those in eqs .
( [ bound_mwr ] ) and ( [ bound_vub ] ) for our numerical analysis in sec v.
in order to include qcd effects systematically , we start from the following low - energy effective hamiltonian describing @xmath51 and @xmath52 transition as done similarly in ref .
@xcite : _ eff = + ( c_i o_i c_i o_i ) , where @xmath70 , @xmath71 are the standard current - current operators , @xmath72 are the standard penguin operators , and @xmath73 and @xmath74 are the standard photonic and gluonic magnetic operators , respectively , which can be found in ref . @xcite .
in addition to those sm operators , in the lrm , the operator basis is doubled by @xmath75 which are the chiral conjugates of @xmath76 . also new operators
@xmath77 and @xmath78 arise with mixed chiral structure of @xmath71 and @xmath79 as follows : o^q_11 = ( |s_q_)_v - a ( @xmath80 refers to the lorentz structure @xmath81
. these new operators may play an important role in tree - level dominated @xmath82 decays in the general lrm .
the low - energy effects of the full theory at an arbitrary low - energy scale @xmath83 can then be described by the linear combination of the given operators and the corresponding wilson coefficients ( wcs ) @xmath84 . in order to calculate @xmath84 ,
we first calculate them at @xmath85 scale . after performing a straightforward matching computation
, we find the wcs including the electromagnetic penguin contributions at @xmath21 scale neglecting the @xmath86-quark mass : [ coefmw ] c^q_2(m_w ) & = & 1 , c^q_2(m_w ) = _ g^rr_q/^ll_q ( q = u , c ) , c_3(m_w ) & = & , c_7(m_w ) & = & , c_9(m_w ) & = & , + c_7^(m_w ) & = & f(x_t ) + a^tb(x_t ) , c_7^(m_w ) = a^ts(x_t ) , c_8^g(m_w ) & = & g(x_t ) + a^tb(x_t ) , c_8^g(m_w ) = a^ts(x_t ) , c^u_12(m_w ) & = & a^ub , c^u_12(m_w ) = a^us , where [ ckmratio ] x_u = ( u = u , c , t ) , a^ud = _ g e^i _ ( d = b , s ) , and @xmath87 is a cp phase residing in the vacuum expectation values , which can be absorbed in @xmath58 in eq .
( [ vr ] ) by redefining @xmath88 .
all other coefficients are negligible or vanish . in eq .
( [ coefmw ] ) , the explicit forms of the functions @xmath89 , @xmath90 , and @xmath91 can be found in refs .
@xcite , and @xmath92 , @xmath93 , @xmath94 , and @xmath95 are given in ref .
@xcite . in the above magnetic coefficients ,
the terms proportional to @xmath56 and @xmath96 are neglected except the contribution coming from the virtual @xmath97 quark which gives @xmath98 enhancement .
also the term proportional to @xmath96 in the coefficient @xmath99 is not neglected because @xmath100 and there is possible enhancement by the ratio of ckm angles ( @xmath101 ) in the nonmanifest lrm .
note that the new coefficient @xmath102 can be important in some @xmath103 transitions because the @xmath56 suppression can be offset by the ratio @xmath104 in eq .
( [ ckmratio ] ) .
the coefficients @xmath84 at the scale @xmath83 below @xmath105 can be obtained by evolving the coefficients @xmath106 with the @xmath107 anomalous dimension matrix applying the usual renormalization group procedure in the following way : ( ) = u_4(,m_b)m(m_b)u_5(m_b , m_w)(m_w ) , where @xmath108 is the evolution matrix for @xmath109 active flavors and @xmath110 gives the matching corrections between @xmath111 and @xmath112 . in the leading logarithmic ( ll ) approximation @xmath113 and the evolution matrix @xmath114
is given by [ rgevol ] u(m_1,m_2 ) = vv^-1 , where @xmath115 diagonalizes the transposed of the anomalous dimension matrix @xmath116 and @xmath117 is the vector containing the eigenvalues of @xmath118 . in the right - hand side of eq .
( [ rgevol ] ) , the first term represents the pure qcd evolution and the second term describes the additional evolution in the presence of the electromagnetic interaction .
the leading order formula for the matrix @xmath119 can be found in refs . @xcite .
since the strong interaction preserves chirality , the @xmath107 anomalous dimensional matrix decomposes into two identical @xmath120 blocks .
the sm @xmath121 submatrix describing the mixing among @xmath122 , @xmath73 , and @xmath74 can be found in ref .
@xcite , and the explicit form of the remaining @xmath123 matrix describing the mixing among @xmath77 , @xmath73 , and @xmath74 , which partially overlaps with the sm @xmath121 submatrix , can be found in ref .
@xcite . in this paper , unlike the previous analysis in ref .
@xcite , we set the scale of weak wcs at @xmath124 1.5 gev to use the pqcd results for the hadronic matrix elements @xcite .
for 4 flavors , we have the following numerical values of @xmath125(1.5 gev ) in ll precision using the standard quark masses : are different from those in @xmath126 in general , we use an approximation @xmath127 for simplicity , which will not change our result . ]
c_1^q = -0.453 , c_1^q = c_1^q_g^rr_q/^ll_q , c_2^q = 1.231 , c_2^q = c_2^q_g^rr_q/^ll_q , c_3 = 0.024 , c_4 = -0.046 , c_5 = 0.012 , c_6 = -0.066 , [ coef - num ] c_7 = 0.014 , c_8 = 0.069 , c_9 = -1.436,c_10= 0.503 , c_7^= -0.389 - 17.86a^tb , c_7^ = -17.86a^ts , c_8^g = -0.177 - 7.858a^tb , c_8^g = -7.858a^ts .
c_11^u = 0.641a^ub , c_12^u = 0.879a^ub , c_11^u = 0.641a^us , c_12^u = 0.879a^us , where subdominant np terms are neglected .
note that @xmath128 are negligible comparing to @xmath129 and @xmath130 whereas @xmath131 and @xmath132 are not .
@xmath133 and @xmath132 can be important especially to the tree - dominated @xmath8 decays .
following the procedure of ref .
@xcite of including the penguin - type diagrams of the current - current operators @xmath71 and the tree - level diagrams associated with the magnetic operators @xmath73 and @xmath74 , the one - loop matrix elements of @xmath134 can be written in terms of the tree - level matrix elements of the effective operators : = -^ll_t _ i=1 ^ 12c_i^eff < sq|q|o_i|b>^tree + ( c_i^eff o_i c^eff_i o_i ) , with the effective wcs c_i^eff ( ) = c_i^ ( ) & & ( i=1,2,8,10,11,12 ) , c_3^eff ( ) = c_3^ ( ) - c_g^ ( ) , & & c_4^eff ( ) = c_4^ ( ) + c_g^ ( ) , c_5^eff ( ) = c_5^ ( ) - c_g^ ( ) , & & c_6^eff ( ) = c_6^ ( ) + c_g^ ( ) , + c_7^eff ( ) = c_7^ ( ) + c_^ ( ) , & & c_9^eff ( ) = c_9^ ( ) + c_^ ( ) , where c_g^ ( ) & = & - , + c_^ ( ) & = & -,and ( m , k , ) = 4_0 ^ 1 dx x(1-x ) . and where @xmath135 is the momentum transferred by the photon or the gluon to the ( @xmath136 ) pair . here
@xmath137 is expected to be typically in the range @xmath138 @xcite , and we will use @xmath139 for our numerical analysis .
the expression of the decay amplitudes can be further simplified by combining the effective wcs in the following way : a_2i-1^ ( ) = c^eff()_2i-1+c^eff()_2i , a_2i^ ( ) = c^eff()_2i+c^eff()_2i-1 ( i=1,2,3 ) , where the factor @xmath140 originates from fierzing the operators @xmath141 after adopting the factorization assumption , and @xmath142 is simply equal to the number of colors in the naive factorization approximation based on the vacuum - insertion method @xcite .
also , in the pqcd approach , @xmath143 as well because of the cancellation between the nonfactorizable contributions
. the matrix amplitudes for @xmath1 decays can then be written in terms of the effective wcs as [ amplitudes ] ( |b^0 ^ 0 |k^0 ) & = & \ { x^(bk , ) & & + _ t^llx^(b , k ) & & + _ t^llx^(b , k ) } & & + ( a_i -a_i^ ) , + ( |b^0^+ k^- ) & = & \ { x^(b , k ) & & - _ t^llx^(b , k ) } & & + ( a_i -a_i^ ) , + ( b^-^0 k^- ) & = & \ { x^(bk , ) & & + x^(b , k ) & & + x^(b , k ) } & & + ( a_i -a_i^ ) , + ( b^-^- |k^0 ) & = & \ { -_t^llx^(b , k ) & & + x^(b , k ) } & & + ( a_i -a_i^ ) , where [ formfactor ] x^(bk , ) & = & -<^0||u^_5 u|0><|k^0||s_b||b^0 > & = & i f_f_0^bk(m_^2)(m_b^2-m_k^2 ) , x^(b , k ) & = & + < |k^0||s^_5 d|0><^0||d_b||b^0 > & = & i f_k f_0^b(m_k^2)(m_b^2-m_^2 ) , x^(b , k ) & = & + < ^0|k^0||s^d|0><0||d__5 b||b^0 > & = & i f_b f_0^k(m_b^2)(m_k^2-m_^2 ) , and where ^h_q ( h=,k , b , q = u , s ) . note from eq . ( [ formfactor ] ) that the form factors @xmath144 and @xmath145 can be determined by relevant semileptonic @xmath8 decays , but @xmath146 is not . because of the significant hadronic uncertainties in the factorization approximation of the matrix amplitudes , it is very difficult to separately determine the size of np contributions .
therefore , in this paper , instead of performing a complete analysis by varying all relevant independent np parameters ( @xmath147 ) in this model , we fix @xmath56 and @xmath148 for simple illustration of np effects . also ,
for numerical analysis , we use the following values of form factors obtained from pqcd calculation @xcite : f_0^bk(m_^2 ) = 0.37 , f_0^b(m_k^2 ) = 0.24 , f_0^k(m_b^2 ) = ( 0.39 + 8.16i)10 ^ -4 . the form factor @xmath149 which originates from annihilation contributions is complex due to the final state quark interactions , so that it could be important in cp observables .
also , due to the enhancement factor @xmath150 proportional to @xmath151 , the annihilation contributions are not negligible .
for illustration of allowed new model parameter spaces , we first fix the new gauge boson mass to be @xmath152 tev , and plot the allowed region of @xmath59 and @xmath57 at @xmath63 level for @xmath153 in fig .
[ phase1 ] and for @xmath154 in fig .
[ phase2 ] using the present experimental bounds of the cp asymmetries in eq .
( [ acp_exp ] ) . with the chosen np parameters , the branching fraction of each decay mode in eq .
( [ amplitudes ] ) agrees with the present experimental measurement as well . in fig .
[ phase ] , the region ( nearly all ) above the dotted line is allowed by the @xmath155 bound given in eq .
( [ bound_vub ] ) , and the shaded area inside the dashed lines is allowed by the @xmath3 mixing bound in eq .
( [ bound_mixing ] ) . from the overlapped allowed regions of the figures
, one can see that the value of @xmath57 could be either small or large , but nonzero value of @xmath59 near @xmath156 is preferred in both cases similarly . in order to clearly see @xmath53 mass dependence , we plot the allowed region of @xmath57 and @xmath96 at @xmath63 level for @xmath157 in fig . [ angle ] , taking into account the constraint from the muon decay measurements given in eq .
( [ bound_mwr ] ) . in fig .
[ angle ] , similarly , the shaded regions left of the dotted and dashed lines are allowed by the @xmath155 bound given in eq .
( [ bound_vub ] ) and by the @xmath3 mixing bound in eq .
( [ bound_mixing ] ) , respectively . with the given parameter sets , we estimate the size of the right - handed current contributions responsible for the present measurements of @xmath158 , @xmath159 , and @xmath160 as shown in fig .
[ angle ] , and obtain the lower bound of @xmath96 approximately given as @xmath161 which corresponds to the upper bound of @xmath53 mass @xmath162 tev .
we found that this mass bound could be somewhat higher for different values of @xmath59 , but not drastically different .
it should also be noted that we scanned other sets of np parameters , and have no better results ( no wider simultaneously allowed regions ) for different values of @xmath148 .
in this paper , we studied the right - handed current contributions to the direct cp asymmetries in @xmath1 decays taking into account all possible tree and penguin contributions including annihilation - type amplitudes by adopting pqcd approach in the nonmanifest lrm . without imposing manifest or pseudomanifest left - right symmetry , we parametrized @xmath36 as shown in eq .
( [ vr ] ) so that @xmath53 mass is not strongly constrained by the current direct and indirect search results @xcite , and showed that the cp asymmetries are sensitive to the phases and angles in @xmath36 as well as to the mass of @xmath53 .
we considered the constraints from the global analysis of muon decay parameters , the determination of @xmath2 in inclusive and exclusive @xmath8 decays , and @xmath3 mixing measurements . with the given phases ,
one can see from the figures that relatively large value of the mixing angle @xmath57 is preferred unless @xmath53 mass is as light as a few hundred gev .
this could also give simultaneous explanations to the large branching fractions of @xmath48 transitions due to a large fraction @xmath163 @xcite , and also to the large cp - violating like - sign dimuon charge asymmetry in semileptonic b decays @xcite due to a large cp - violating phase in @xmath42 mixing @xcite . also , with the given parameter sets , fig .
[ angle ] shows that it is favorable that the mass of @xmath53 is lighter than around @xmath164 tev in order to incorporate the current experimental measurements . in this way , cp asymmetries in other nonleptonic b decays such as @xmath165 and @xmath166 can be estimated systematically and similarly , and all of these analysis of possible np contributions can be tested once future experimental progress can further improve the bounds .
nam thanks c .- h .
chen and h - n .
li for useful communications , and c .- w .
chiang for collaboration at the beginning of this project .
the computation of this work was supported in part by the plsi supercomputing resources of kisti .
# 1#2#3 eur .
j. c * # 1 * , # 2 ( # 3 ) # 1#2#3 jhep * # 1 * , # 2 ( # 3 ) # 1#2#3 nucl .
b * # 1 * , # 2 ( # 3 ) # 1#2#3 mod .
lett . a * # 1 * , # 2 ( # 3 ) # 1#2#3 phys .
b * # 1 * , # 2 ( # 3 ) # 1#2#3 phys .
d * # 1 * , # 2 ( # 3 ) # 1#2#3 phys .
lett . * # 1 * , # 2 ( # 3 ) # 1#2#3 prog .
# 1 * , # 2 ( # 3 ) # 1#2#3 rev .
# 1 * , # 2 ( # 3 ) # 1#2#3 z. phys .
c * # 1 * , # 2 ( # 3 ) # 1#2#3 _ ibid . _ * # 1 * , # 2 ( # 3 ) # 1#2#3 * # 1 * , # 2 ( # 3 ) a.j .
buras , r. fleischer , s. recksiegel , and f. schwab , ; v. barger , c .- w .
chiang , p. langacker , and h .- s .
lee , ; r. arnowitt , b. dutta , b. hu , and s. oh , ; y .- d . yang , r. wang , and g.r .
lu , ; m. imbeault , s. baek , and d. london , ; s. baek _ et al .
_ , ; s. chang , c.s .
kim , and j. song , ; j.n .
ng and p.t .
winslow , . | we reexamine the right - handed current effects in @xmath0 transitions in nonmanifest left - right models . using the effective hamiltonian approach including all possible low - energy operators , we obtain especially the @xmath1 decay amplitudes including annihilation contributions , and investigate the right - handed current contributions to cp asymmetries in @xmath1 decays . taking into account the constraints from global analysis of muon decay measurements , @xmath2 measurements in inclusive and exclusive b decays , and @xmath3 mixing measurements , we find the allowed regions of new physics parameters satisfying the current experimental data . |
the hadronic contribution to the muon anomaly can be related to the hadronic cross sections via a dispersion relation @xmath6 where the integral is carried out over the invariant mass squared _ s _ of the hadronic system .
the kernel @xmath7 is a monotonously varying function that behaves approximately like @xmath8 .
the annihilation cross section also has an intrinsic @xmath8 behaviour and is largely enhanced around the mass of the @xmath3 meson .
data at low energies contribute therefore strongly to @xmath9 .
since perturbative qcd fails at energies below @xmath10 gev , one has to use experimentally measured cross sections in the dispersion integral for low energy regions .
+ the most recent experimental value for @xmath11 is @xcite + @xmath12x@xmath13 [ e821 ] , + while a recent theoretical evaluation gives @xcite @xcite + @xmath14x@xmath13 [ @xmath15 data ] + @xmath16x@xmath13 [ @xmath17 data ] + the first theoretical evaluation of @xmath11 is obtained with @xmath15 decay data as experimental input to the dispersion integral , using conservation of vector current and isospin symmetry to relate @xmath15 decays to annihilation cross sections .
the second theoretical evaluation uses data exclusively , including the reanalyzed data from novosibirsk ( , @xcite ) . comparing the theoretical evaluations with the experimental value ,
one finds a deviation of 2.0 @xmath18 for the -based theory value , while the @xmath15-based value shows only a 0.7 @xmath18 effect . in order to resolve this ambiguous situation and to be able to claim an eventual violation of the standard model ,
more and better information on hadronic cross sections is needed . for a more detailed discussion , see also @xcite .
the standard way to measure hadronic cross sections in the past was to perform an energy scan , in which the energy of the colliding beams was changed to the desired value .
the frascati collider da@xmath0ne , however , was designed as a particle factory operating at the fixed energy of the @xmath1 resonance ( 1020 mev ) with high luminosity and the current physics program does not allow to vary the beam energy away from the @xmath1 resonance . as a consequence of this , the idea has been worked out to obtain @xmath19 at da@xmath0ne using the radiative process hadrons + , where the photon has been radiated in the initial state by one of the electron or the positron , lowering the collision energy @xcite @xcite . by measuring this _ radiative return _ process
the hadronic cross sections become accessible from the @xmath1 mass down to the two - pion threshold .
the total cross section @xmath20 can be obtained by measuring @xmath21 , where @xmath22 from the radiative return .
the two quantities can be related by the radiator @xmath23 : @xmath24 @xmath23 depends on _ s _ and on the polar angle of the radiated photon @xmath25 .
an accuracy better than 1% is needed for @xmath23 in order to perform a precision measurement .
radiative corrections have been computed by different groups up to next - to - leading order for the exclusive final hadronic states @xmath26 @xcite @xcite @xcite and 4@xmath27 @xcite .
our analysis makes use of the theoretical montecarlo event generator phokhara @xcite @xcite @xcite @xcite , which also includes now final state radiation of the pions using scalar qed @xcite .
+ below 1 gev , the contribution of the process to the total hadronic contribution to @xmath11 is 67% @xcite .
the cross section for this process can be obtained using eq .
[ eq : h ] by measuring @xmath28 . in the following results obtained for this process with the kloe detector in frascati will be reported .
-.5 cm the kloe detector consists of a high resolution drift chamber ( @xmath30 , @xcite ) and an electromagnetic calorimeter ( @xmath31 , @xcite ) . in the current analysis
we have concentrated on events in which the pions are emitted at polar angles @xmath32 between @xmath33 and @xmath34 .
no explicit photon detection is required and an _ untagged _ measurement is performed . as a consequence , a cut on the di - pion production angle @xmath35 smaller than @xmath36 ( or greater than @xmath37 )
is performed instead of a cut on the photon angle @xmath25 .
@xmath35 is calculated from the momenta of the two pions and is exactly equal to @xmath38 if only one photon is emitted .
the acceptance regions are shown in fig .
[ fig : fiducial ] .
the reason to choose these specific acceptance cuts are reduced background contaminations and a low relative contribution of final state radiation from the pions .
it will be shown in the following that an efficient and almost background free signal selection can be performed without explicit photon tagging @xcite @xcite @xcite . + the selection of @xmath39 is performed with the following steps : * _ detection of two charged tracks connected to a vertex _ : two charged tracks with polar angles between @xmath33 and @xmath34 connected to a vertex in the fiducial volume @xmath40 cm , @xmath41 cm are required .
additional cuts on transverse momentum @xmath42 mev or on longitudinal momentum @xmath43 mev reject spiralizing tracks and ensure good reconstruction conditions . * _ identification of pion tracks _ : a @xmath44-@xmath45 separation is done by cutting on a likelihood function which parametrizes the interaction of charged particles with the electromagnetic calorimeter .
the function has been built using control samples from data of @xmath46 and @xmath47 in order to find the calorimeter response for pions and electrons .
the event is selected if at least one of the two tracks is identified to be a pion . in this way
, @xmath47 events are drastically removed from the signal sample while the efficiency for @xmath48 is still very high ( @xmath49 99% ) . *
_ cut on the track mass _ : the track mass is a kinematic variable corresponding to the mass of the charged tracks under the hypothesis that the final state consists of two particles with the same mass and one photon .
it is calculated from the reconstructed momenta @xmath50 , @xmath51 .
cutting at a value of 120 mev rejects @xmath52 events , while in order to reject @xmath46 events a cut in the plane of track mass and @xmath53 is performed ( see fig .
[ fig : trackmass ] ) : @xmath54 all units in eq .
[ eq : trackmass ] are in mev . *
_ cut on the di - pion angle @xmath35 _ : the aforementioned cut on the di - pion angle @xmath55 or @xmath56 is performed . /gev@xmath57 ,
title="fig:"]-.5 cm applying the selection steps mentioned in the previous section on a sample of @xmath58 pb@xmath59 of data taken in 2001 , ca .
events are selected , corresponding to 11000 events / pb@xmath59 .
[ fig : spectrum ] shows the number of @xmath48 events selected in bins of @xmath60 gev@xmath57 .
the @xmath61 interference is clearly visible , demonstrating the excellent momentum resolution of the kloe drift chamber .
events selected by the analysis in the region @xmath62 or @xmath63 , @xmath64.,title="fig:"]-.5 cm to obtain the effective @xmath48 cross section one has to subtract the residual background from this spectrum and divide by the selection efficiency and the integrated luminosity : @xmath65 * _ residual background @xmath66 _ : given the angular acceptance regions for pions and photon as described in fig .
[ fig : fiducial ] , the amount of residual background is considered to be small .
it has been evaluated by fitting the mc distributions for @xmath48 ( _ signal _ ) and @xmath52 , @xmath47 and @xmath67 ( _ background _ ) to the data .
the relative contribution of each background channel is found to be well below 5% . * _ the selection @xmath68 _ : the selection efficiency is the product of trigger , reconstruction , filtering , @xmath44-@xmath45 separation and track mass cut efficiencies .
apart from the last one , which has been evaluated from montecarlo , all efficiencies have been evaluated from data , using unbiased control samples matching the kinematics of the actual signal .
the total selection efficiency is about @xmath69 . *
_ luminosity _ : at kloe , the luminosity measurement is performed using bhabha events at large angles ( @xmath70 , @xmath71nb ) .
the number of bhabha candidates is counted and the luminosity is obtained from the effective cross section obtained from montecarlo .
kloe uses two independent bhabha event generators ( the berends / kleiss generator @xcite modified for da@xmath0ne @xcite and the babayaga generator @xcite ) , for both of which the authors quote a theoretical error of 0.5% .
both generators agree within @xmath72 .
the good agreement of the experimental distributions ( @xmath73 , @xmath74 ) with the event generators results in an experimental uncertainty for the luminosity measurement of 0.4% . .
the radiator @xmath75 obtained from mc is also shown.,title="fig:"]-.5 cm the effective cross section within the acceptance cuts described in fig .
[ fig : fiducial ] is shown in fig .
[ fig : pipig ] .
assuming that initial and final state radiation processes are independent ( cf . neglecting interference effects ) the cross section for @xmath77 can be extracted as a function of @xmath78 from the observed differential @xmath48 cross section as : @xmath79 where @xmath80 is the nlo cross section for @xmath39 ( initial state radiation only ) under the assumption of pointlike pions and corresponds to the quantity @xmath23 from eq .
[ eq : h ] .
we obtain @xmath23 bin - by - bin technically from the theoretical montecarlo generator phokhara by setting @xmath81 and switching off the vacuum polarisation of the intermediate photon ( see fig .
[ fig : pipig ] ) .
@xmath82 is the pion velocity defined as @xmath83 .
note that no explicit acceptance correction has been performed , but @xmath23 is evaluated within the acceptance cuts of fig .
[ fig : fiducial ] . dividing by the _ effective _
@xmath23 thus yields the total cross section for @xmath84 . in order to evaluate @xmath9 by inserting the cross section in the dispersion integral ( eq .
[ eq : amu ] ) , some radiative corrections to the cross section have to be considered . * _ final state radiation _ : events with final state radiation from the pions should be included in the cross section to be put in the dispersion integral @xcite . the leading order final state radiation
where there is no initial state radiation ( hence the @xmath85 and the @xmath86 collide at the energy of the @xmath1 meson ) creates a large tail , which is unwanted in our case and is suppressed very efficiently to the level of few permille using the acceptance cuts described in fig .
[ fig : fiducial ] via the separation of pion and photon fiducial volumes ( the photon emission by the pions is peaked in the pion direction ) .
considering the next - to - leading order process of simultaneous occurence of a photon radiated in the initial state and a photon radiated in the final state , events of this type are rejected by the aforementioned cut in trackmass .
the presence of two photons in the event moves the @xmath87 event towards higher values of trackmass in fig .
[ fig : trackmass ] , therefore especially at low @xmath53 they are moved into the region which is cut out by the trackmass cut . as of now
, we recover these events by evaluating the trackmass efficiency from the phokhara 3.0 montecarlo @xcite which includes the simulation of this kind of events within the model of scalar qed . dividing the raw spectrum by this efficiency according to eq .
[ eq : ana ] retains the @xmath87 events rejected by the trackmass cut . *
_ vacuum polarisation _ : the cross section has to be undressed from vacuum polarisation @xcite .
this is done by dividing for the running of the fine structure constant : @xmath88 the correction function to @xmath89 has been provided by f. jegerlehner @xcite . the resulting total cross section @xmath76
after these corrections is shown in fig .
[ fig : pipi ] . .,title="fig:"]-.5
cm the contribution of the several analysis items to the systematic error of the measurement is shown in table [ tab_ex ] . given the excellent resolution of the kloe drift chamber a dedicated unfolding from the detector resolution seems not necessary .
eventual systematic contributions from this omission are currently under study , but are expected to be small .
.list of exp .
systematic uncertainties [ cols="<,^",options="header " , ] additionally , we attribute a preliminary systematic error of 1.% due to the treatment of final state radiation in our analysis .
this covers the limited validity of the radiator @xmath23 due to eventual non - factorizing terms between initial and final state radiation , possible limitations of the model used to simulate final state radiation , the error arising due to the fact that in the presence of final state radiation the ( measured ) @xmath90 is slightly different from @xmath91 ( where @xmath92 is the virtual photon produced in the @xmath93-collision ) and additional acceptance corrections in case the acceptance changes in the presence of final state radiation events .
a method of measuring hadronic cross sections at particle factories has been presented using the radiative return .
this method is complementary to the energy scan used to perform this measurement up to now and has completely different systematical errors , making it the ideal tool to check and improve existing measurements .
such a crosscheck has become very important given the disagreement between the evaluations of @xmath11 obtained in @xcite .
the radiative return has been used at the da@xmath0ne @xmath1-factory in frascati in order to extract the total cross section for @xmath94 , making good use of the recently developed upgrade of the phokhara event generator by @xcite .
an interpretation of our cross section in terms of @xmath9 as well as future plans and prospects of this measurement at present and upcoming @xmath93-colliders in frascati can be found in a. denig s contribution to this workshop @xcite . in particular , the quoted result for the two - pion contribution to @xmath9 is ( in units of @xmath13 ) : @xmath95 our preliminary result agrees within the errors with the result found by cmd2 @xcite .
we thank the da@xmath0ne team for their efforts in maintaining low background running conditions and their collaboration during all data - taking . we want to thank our technical staff : g.f.fortugno for his dedicated work to ensure an efficient operation of the kloe computing center ; m.anelli for his continous support to the gas system and the safety of the detector ; a.balla , m.gatta , g.corradi and g.papalino for the maintenance of the electronics ; m.santoni , g.paoluzzi and r.rosellini for the general support to the detector ; c.pinto ( bari ) , c.pinto ( lecce ) , c.piscitelli and a.rossi for their help during shutdown periods . this work was supported in part by doe grant de - fg-02 - 97er41027 ; by eurodaphne , contract fmrx - ct98 - 0169 ; by the german federal ministry of education and research ( bmbf ) contract 06-ka-957 ; by graduiertenkolleg ` h.e .
phys . and part
. astrophys . ' of deutsche forschungsgemeinschaft , contract no .
gk 742 ; by intas , contracts 96 - 624 , 99 - 37 ; and by tari , contract hpri - ct-1999 - 00088 . | the measurement of hadronic cross sections at low centre - of - mass energies is currently attracting much interest because of the rle these measurements play in the theoretical evaluation of the anomalous magnetic moment of the muon .
the kloe experiment at the da@xmath0ne @xmath1-factory in frascati aims to determine the cross section of the dominant contribution @xmath2 at energies below 1 gev exploiting the radiative return from the @xmath1 to the @xmath3 and @xmath4 mesons .
this method has systematic errors completely different from previous measurements and allows a cross check of existing data in this energy region .
the method is presented in detail and results on both the differential cross section for @xmath5 and the total cross section for @xmath2 are presented . |
by analogy with the classical period matrix of an ordinary riemann surface , one can define the super period matrix ( sometimes just called the period matrix ) of a super riemann surface .
if @xmath4 is a super riemann surface of genus @xmath0 with even spin structure , then its super period matrix is a @xmath5 symmetric matrix of positive imaginary part .
actually , the super period matrix is only defined for the case that @xmath4 has an even spin structure , and even then it is only defined generically : it can acquire a pole , with nilpotent residue , when moduli of @xmath4 are varied . the first goal of the present paper is to extend the definition of the super period matrix to the case of a super riemann surface with ramond punctures .
( a neveu - schwarz puncture , which is simply a marked point , does not affect the definition of the super period matrix . )
it is conceivable that there is more than one reasonable definition .
the definition we give here is motivated by an application that we will explain shortly . in this definition ,
the super period matrix of a super riemann surface @xmath4 of genus @xmath0 with @xmath1 ramond punctures ( the number of ramond punctures is always even ) is a @xmath2 matrix , symmetric in the @xmath3-graded sense , whose @xmath5 bosonic block has positive definite imaginary part .
( the super period matrix is in general not an arbitrary matrix of this sort , since in general there are schottky relations . ) just as in the classical case , the definition of the period matrix depends on a choice of @xmath6-cycles ; when ( and only when ) ramond punctures are present , one has to define fermionic as well as bosonic @xmath6-cycles .
if one changes the @xmath6-cycles that are used in defining it , the super period matrix is transformed by an element of an integral form of the supergroup @xmath7 , generalizing the fact that in the classical case ( or for a super riemann surface without ramond punctures ) , the period matrix is defined up to the action of an integral form of @xmath8 . just as in the absence of ramond punctures ,
the super period matrix is only generically defined , and can acquire singularities as moduli are varied .
the application we have in mind involves superstring perturbation theory in genus 2 .
every @xmath9 matrix of positive imaginary part is the period matrix of an ordinary riemann surface of genus 2 , unique up to isomorphism ( schottky relations only exist in genus @xmath10 ) .
hence , to a super riemann surface @xmath4 of genus 2 with even spin structure , we can associate an ordinary riemann surface @xmath11 of the same period matrix .
@xmath11 also inherits a spin structure from the spin structure of @xmath4 , and the association @xmath12 gives a natural holomorphic projection @xmath13 from the moduli space @xmath14 of super riemann surfaces @xmath4 of genus 2 with even spin structure to its reduced space @xmath15 which parametrizes an ordinary riemann surface @xmath16 with even spin structure .
is everywhere defined ( on @xmath14 as opposed to its deligne - mumford compactification ) , in part because for @xmath17 ( unlike @xmath18 ) the super period matrix has no poles . since the odd dimension of @xmath14 is 2 , @xmath19 is actually a splitting of @xmath14 . ] by integrating over the fibers of @xmath19 , one can map the two - loop vacuum amplitude of superstring theory , which is naturally a measure @xmath20 on @xmath14 , to a measure @xmath21 on @xmath15 .
this procedure was the starting point in the celebrated analysis of the two - loop vacuum amplitude by dhoker and phong .
( for a review with further references , see @xcite .
dhoker and phong also went on to calculate scattering amplitudes in genus 2 , a much more difficult computation that is beyond the scope of the present paper . ) to analyze the integral @xmath22 , it makes sense to first sum over even spin structures before performing any integration . in this way
, one projects @xmath21 from @xmath15 to @xmath23 , with the spin structure forgotten . in their original work , dhoker and phong analyzed this sum over spin structures for superstring theory on @xmath24 and for certain supersymmetric orbifold compactifications to six dimensions .
they showed that the sum over spin structures vanishes in those models , analogous to the familiar gso cancellation in genus 1 .
something new happens in general in the case of a compactification to _ four _ dimensions that at tree level has @xmath25 supersymmetry .
( the most simple examples are provided by compactification of the heterotic string on a calabi - yau three - fold . ) in this case , it is possible for a 1-loop effect to generate a fayet - iliopoulos @xmath26-term , triggering the spontaneous breaking of supersymmetry @xcite .
when this happens , one expects the genus 2 vacuum amplitude to be non - zero and proportional to @xmath27 .
how does this occur in the context of the dhoker - phong procedure for computing the genus 2 vacuum amplitude ? in general @xcite , the dhoker - phong procedure must be supplemented with a boundary correction ( a contribution supported on the divisor at infinity in the compactified moduli space ) .
the boundary contribution to the genus 2 vacuum amplitude vanishes in supersymmetric compactifications above four dimensions , but in a four - dimensional model with @xmath25 supersymmetry , it is proportional to @xmath27 .
this raises the possibility that in such models , the full answer comes from this boundary correction , and that the bulk contribution , computed with the procedure of dhoker and phong , always vanishes .
something similar happens in the same models in one - loop computations of certain supersymmetry - violating mass splittings @xcite .
in fact , in examples of orbifold compactifications to four dimensions with @xmath25 supersymmetry @xcite , the same behavior has been found that was found earlier in supersymmetric models above four dimensions : the bulk contribution @xmath21 to the genus 2 vacuum amplitude vanishes after summing over spin structures , without any integration over bosonic moduli . in the present paper , we will use the theory of the super period matrix with ramond punctures to demonstrate that this very striking behavior will occur in all supersymmetric compactifications to four or more dimensions .
perhaps we should remark that general arguments based on supersymmetric ward identities ( see for example section 4 of @xcite ) can be used to determine the _ integrated _ genus 2 vacuum amplitude , but do not explain the `` _ pointwise _ '' vanishing that occurs in the dhoker - phong procedure .
our goal here is to explain this more detailed phenomenon .
the arguments governing the integrated behavior are completely general and apply for all values of the genus .
the procedure that leads to pointwise vanishing is defined only for genus @xmath28 or at most ( as we discuss in section [ twop ] ) @xmath29 .
we define the super period matrix with ramond punctures and explain some of its simplest properties in sections [ odd]-[dimform ] of this paper .
( a parallel treatment of some of these issues from the point of view of supergravity will appear elsewhere @xcite . )
the application to the two - loop vacuum amplitude is in sections [ twop]-[lowgenus ] .
the general strategy to constrain the vacuum amplitude via supersymmetry is familiar @xcite , and involves comparing the genus @xmath0 vacuum amplitude to an amplitude computed on will denote the moduli space of super riemann surfaces of genus @xmath0 with @xmath30 neveu - schwarz punctures and @xmath1 ramond punctures .
we write @xmath31 if we wish to indicate the type of spin structure .
similarly , @xmath32 is the moduli space of ordinary riemann surfaces of genus @xmath0 with @xmath30 punctures , while @xmath33 is the corresponding moduli space with a choice of even or odd spin structure . ] @xmath34 .
since we specifically want to constrain the genus 2 vacuum amplitude computed with a procedure that uses the super period matrix , we have to begin with an understanding of the super period matrix of a genus 2 super riemann surface with 2 ramond punctures . the super period matrix in this situation
is only generically defined , with singularities on a certain locus in the moduli space .
the trickiest part of our analysis is to show that these singularities do not ruin the argument ; see section [ lowgenus ] .
some technical issues are treated in appendices . in appendix
a , we explain in detail why the super period matrix has a pole ; in appendix b , we give an alternative explanation of the fact that the dhoker - phong procedure for integration over genus 2 supermoduli space requires a correction at infinity ; and in appendix c , we describe some properties of the genus 3 analog of the dhoker - phong procedure .
we will begin by recalling the definition of a super riemann surface with or without ramond punctures .
( the reader may want to consult a more detailed reference such as @xcite or @xcite . )
then we go on to discuss periods .
a super riemann surface @xmath35 is a complex supermanifold of dimension @xmath36 whose tangent bundle @xmath37 is endowed with a subbundle @xmath38 of rank @xmath39 that is completely unintegrable .
complete unintegrability means that if @xmath40 is a nonzero section of @xmath38 , then @xmath40 and @xmath41 are everywhere linearly independent , so that @xmath41 generates @xmath42 .
thus @xmath43 fits in an exact sequence @xmath44 dually , the cotangent bundle of @xmath4 fits in an exact sequence @xmath45 one can show that locally , one can pick coordinates @xmath46 on @xmath4 such that @xmath38 is generated by @xmath47 such coordinates are called local superconformal coordinates . dually
the subbundle @xmath48 of @xmath49 is generated by @xmath50 a super riemann surface @xmath4 with ramond punctures is a complex supermanifold of dimension @xmath36 whose tangent bundle is still endowed with a subbundle @xmath38 of rank @xmath39 , but now the condition of complete unintegrability fails along a certain divisor .
the local behavior is that @xmath38 is generated , in some coordinates @xmath46 , by @xmath51 since @xmath52 , we see that @xmath53 and @xmath54 fail to be linearly independent precisely along the divisor @xmath55 defined by @xmath56 .
thus , the exact sequence ( [ morz ] ) is replaced by @xmath57 dually , one has @xmath58 with @xmath59 generated by @xmath60 more globally , @xmath4 may have many divisors @xmath61 along which the local behavior of the subbundle @xmath62 is as just described .
we call the @xmath61 ramond divisors
. if @xmath4 is compact , the number of ramond divisors is always even .
what appears in the exact sequences ( [ oppo ] ) and ( [ noppo ] ) is @xmath63 , which we might call the total ramond divisor . away from @xmath55 , @xmath4 is an ordinary super riemann surface . along @xmath55 , @xmath4 remains smooth , but there is a singularity in its superconformal structure . a general holomorphic 1-form on @xmath4 can be written locally as @xmath64 .
in contrast to the case of an ordinary riemann surface , a holomorphic 1-form is not necessarily closed ; if @xmath65 is a holomorphic 1-form , then @xmath66 is a holomorphic 2-form , with an expansion @xmath67 . in defining differential forms on a supermanifold
, we define the exterior derivative to be odd , so @xmath68 is odd and anticommutes with @xmath69 , while @xmath70 is even and commutes with @xmath69 and @xmath68 .
we want to define periods of holomorphic 1-forms . just as in ordinary geometry ,
periods , as topological invariants , are only defined for 1-forms that are closed .
the most obvious periods are the analogs of classical periods .
let @xmath65 be a 1-form on @xmath35 .
if @xmath71 is an oriented circle and @xmath72 is any continuous map , then @xmath65 pulls back to an ordinary 1-form @xmath73 on @xmath71 so we define the integral @xmath74 .
just as in the classical case , if @xmath65 is closed , then @xmath74 only depends on the homology class determined by the map @xmath75 . for our purposes , the case
that @xmath75 is an embedding is sufficient , so we can just think of @xmath71 as a smooth submanifold .
see section 5 of @xcite .
] of @xmath4 , of real dimension 1 ( or @xmath76 ) . in this case , we write just @xmath77 rather than @xmath74 . an embedded circle in the reduced space @xmath11 of @xmath4 can be lifted ( not canonically , but in a way that is unique up to homology ) to an embedded circle in @xmath4 .
so , as in @xcite , one can define @xmath6-periods and @xmath78-periods for a closed holomorphic 1-form on @xmath4 that correspond precisely to the familiar @xmath6-periods and @xmath78-periods of a holomorphic differential on @xmath11 .
thus if @xmath65 is a closed holomorphic 1-form on @xmath4 , it has the usual @xmath0 @xmath6-periods and @xmath0 @xmath78-periods . however , in the presence of ramond punctures , such a @xmath65 also has what we might call odd periods . as explained above , near any ramond divisor @xmath79 , we can pick local coordinates @xmath46 in which @xmath79 is defined by @xmath56 and in which the superconformal structure is defined by the distribution generated by @xmath80 or dually by the subbundle of the contangent bundle generated by @xmath81 in what follows , it is important that @xmath69 is uniquely defined up to @xmath82 where @xmath83 is an odd constant .
preserves @xmath84 , it is obvious that the superconformal structure determines @xmath69 only up to sign . to see that @xmath69 is also only defined up to a shift @xmath85 with @xmath83 an odd constant
, one observes that the supergroup of dimension @xmath39 that acts by @xmath86 preserves the superconformal structure of @xmath4 and acts transitively on @xmath79 .
one can verify that along @xmath79 , any superconformal transformation is equivalent to @xmath87 . ] in particular , at @xmath56 , we are not free to rescale @xmath69 except by @xmath88 , since this would disturb the relation between the two terms in @xmath84 or in @xmath53 . accordingly , the 1-form @xmath70 on @xmath79 is well - defined up to sign .
we consider a choice of what we mean by @xmath70 as opposed to @xmath89 to represent an `` orientation '' of @xmath79 .
the odd periods of a closed holomorphic 1-form @xmath65 are now defined as follows .
since @xmath79 is defined by @xmath56 , when restricted to @xmath79 , we have @xmath90 here @xmath91 is a constant independent of @xmath69 since @xmath92 .
( the odd - looking factor @xmath93 in the denominator will be convenient later . )
we simply define @xmath91 to be the odd period of @xmath65 associated to the ramond divisor @xmath79 .
thus , an odd `` period '' is not defined by an integral but by `` evaluation '' of a 1-form along a ramond divisor .
the sign of the odd period depends on the orientation of the ramond divisor , somewhat analogously to the fact that a 1-manifold @xmath94 in an ordinary riemann surface @xmath95 must be oriented if one wishes to define the sign of the period @xmath96 of a closed 1-form @xmath65 . with @xmath1 ramond divisors
@xmath97 , we define in this way @xmath1 odd periods @xmath98 .
thus in all , on a super riemann surface @xmath4 of genus @xmath0 with @xmath1 ramond punctures , we define @xmath99 even periods , just as in the classical theory , and @xmath1 odd periods .
what is the natural symmetry group acting on these periods ?
on the bosonic periods , we have the usual group @xmath100 of automorphisms of @xmath101 ( which coincides with @xmath102 ) .
we can think of @xmath100 as the automorphism group of the skew form @xmath103 defined over @xmath104 , where @xmath105 , @xmath106 are even variables so @xmath107 @xmath108 are odd .
the natural symmetries acting on the fermionic periods are as follows : we could permute the ramond divisors , thus permuting the fermionic periods , or we could reverse the orientations of the ramond divisors and thereby reverse the signs of the fermionic periods .
the permutations and sign changes make up a finite group with @xmath109 elements .
we can think of this group as a form of @xmath110 , since it is the automorphism group of the quadratic form @xmath111 , defined over @xmath104 , where the @xmath112 are the even differentials of odd variables @xmath113 , @xmath114 .
there are no obvious symmetries between even and odd periods .
if we combine the two constructions , the full group @xmath115 that acts on the even and odd periods is the automorphism group of a form @xmath116 that is symmetric in the @xmath3 graded sense in @xmath99 odd variables @xmath117 , @xmath118 , and @xmath1 even variables @xmath112 .
( for now the sign of the @xmath119 term is an arbitrary choice . )
we think of this as a sort of superanalog of the intersection form of an ordinary riemann surface . over @xmath104 , there are no symmetries that exchange even and odd variables , so @xmath115 can be interpreted as @xmath120 , that is , as a form over @xmath104 of the orthosymplectic group .
if we work over @xmath121 , of course the supergroup @xmath122 of symmetries of the form @xmath123 does mix even and odd variables .
typically , not all of @xmath120 is realized as symmetries in string theory .
a super riemann surface has a spin structure , so usually one has to consider only the subgroup of @xmath124 that preserves a spin structure .
also it usually is more useful to consider the ramond divisors to be labeled ( or distinguishable ) , in which case one considers only the sign changes rather than permutations in @xmath110 .
finally , in a certain sense , the space spanned by @xmath125 has a natural orientation , as we explain in section [ anex ] , so one can replace @xmath110 by @xmath126 .
let us recall how one proves the symmetry of the period matrix of an ordinary riemann surface @xmath95 . if @xmath65 and @xmath127 are closed 1-forms on @xmath95 , one has the topological fact @xmath128 if @xmath65 and @xmath127 are holomorphic 1-forms , then @xmath129 , which would be a holomorphic 2-form , vanishes identically .
so the left hand side of ( [ zelmo ] ) vanishes , and this leads to the symmetry of the period matrix . to imitate this argument on a super riemann surface @xmath4 , we face two difficulties : _ ( i ) _
if we view @xmath4 as a smooth supermanifold ( this can be done , though not quite in a canonical way ) , then it has dimension @xmath130 , but a two - form @xmath129 can only be integrated on a manifold of dimension @xmath131 ; _ ( ii ) _ on a super riemann surface , a holomorphic 2-form is not necessarily 0 . the resolution of the first point is simply that , as explained in @xcite , if @xmath4 is a super riemann surface , with reduced space @xmath16 , then @xmath16 can be embedded in @xmath4 in a way that is not canonical ( or holomorphic ) but is unique up to homology .
the image of the embedding is a smooth submanifold @xmath132 of dimension @xmath131 , and the proof of symmetry of the period matrix is made using integrals over @xmath133 rather than @xmath4 .
the resolution of the second point was also explained in @xcite ( see also section 8.2 of @xcite ) . on a super riemann surface ( without ramond punctures ) , a closed holomorphic 2-form is exact , so if @xmath65 and @xmath127 are holomorphic 1-forms , then @xmath134 for some @xmath135 .
this implies vanishing of @xmath136 , just as if @xmath129 were 0 , and leads to the proof of symmetry of the super period matrix . before describing these arguments , and explaining how they must be modified in the presence of ramond punctures
, we give a more elementary example .
let @xmath137 be the supermanifold of dimension @xmath36 defined as the quotient of @xmath138 by @xmath139 here @xmath75 is an odd constant .
@xmath137 is a complex supermanifold of dimension @xmath36 , but not a super riemann surface since the identifications in eqn .
( [ torg ] ) do not preserve a superconformal structure .
the expression @xmath140 defines a closed holomorphic 2-form on @xmath137 that is globally - defined and is not exact ( we can write it locally as @xmath141 , but @xmath142 is not globally - defined , since it is not invariant under @xmath143 )
. the reduced space @xmath144 of @xmath137 is the ordinary riemann surface of genus 1 defined by @xmath145 .
it can be embedded in @xmath137 as the submanifold @xmath146 defined by the equation @xmath147 .
then one finds that @xmath148 the nonzero value of the integral is one way to prove that @xmath140 is not exact .
( since @xmath140 is closed , the value of this integral does not depend on the precise choice of @xmath146 . ) on a super riemann surface @xmath35 without ramond punctures , the proof that a closed holomorphic 2-form @xmath149 is exact proceeds as follows . in local superconformal
coordinates @xmath46 , @xmath149 can be expanded @xmath150 where as usual @xmath151 .
the condition @xmath152 gives @xmath153 and then one finds that @xmath154 even though we have computed in local superconformal coordinates , the object @xmath155 that we have defined does not depend on this choice .
this statement can be explained in the following way ( as in footnote 34 of @xcite ) , using the exact sequence @xmath156 .
the projection @xmath157 , tensored with itself , gives a holomorphic map @xmath158 .
when this is composed with the inclusion @xmath159 , we get a natural map @xmath160 which in local superconformal coordinates is the map @xmath161 .
let us now see how these considerations are modified in the presence of a ramond divisor . as usual , we consider a local model with coordinates @xmath46 and a superconformal structure defined by @xmath162 .
the ramond divisor @xmath55 is defined by @xmath56 .
a holomorphic 2-form @xmath149 can be expanded @xmath163 the condition @xmath152 implies that @xmath164 is independent of @xmath69 at @xmath56 , so we define the constant @xmath165 note that @xmath166 is completely well - defined , as it is not affected by the indeterminacy ( [ ffoo ] ) of @xmath69 . away from @xmath56
, we can replace @xmath46 by local superconformal coordinates @xmath167 ( these are superconformal coordinates , because in these coordinates the usual expression @xmath168 near a ramond puncture takes the standard superconformal form @xmath169 ) . in these coordinates
, @xmath149 has a pole at @xmath56 , with @xmath170 , and hence @xmath155 as determined in eqn .
( [ lug ] ) behaves as @xmath171 going back to the coordinates @xmath172 that behave well near the ramond divisor , since @xmath168 , we have simply @xmath173 away from @xmath56 , it is still true that @xmath174 .
but because of the pole of @xmath155 at @xmath56 , there is actually a delta - function contribution to @xmath175 at @xmath56 : @xmath176 , where @xmath177 is a two - form delta function that is poincar dual to the ramond divisor @xmath55 at @xmath56 . including many ramond divisors @xmath178 , we get @xmath179 where @xmath180 is dual to @xmath178 and @xmath181 is the residue of the corresponding pole in @xmath155 . integrating this formula over a submanifold @xmath182 that is isomorphic to @xmath11
, we get @xmath183
if @xmath184 , where @xmath65 and @xmath127 are two closed holomorphic 1-forms on a super riemann surface @xmath35 with ramond punctures , then we can combine ( [ norog ] ) with the topological formula ( [ zelmo ] ) , with the result that @xmath185 to make use of this , we must express the constants @xmath186 in terms of the odd periods @xmath187 and @xmath188 of @xmath65 and @xmath127 . recalling that the odd period is simply the constant @xmath91 in eqn .
( [ doddo ] ) , we see that near the ramond divisor @xmath178 , we have @xmath189 , @xmath190 , and hence @xmath191 . thus @xmath192 .
if then we denote the @xmath6- and @xmath78-periods as @xmath193 , @xmath194 and similarly @xmath195 , @xmath196 , then we arrive at the analog of the riemann bilinear relations for a super riemann surface with ramond punctures : @xmath197 on the right hand side of eqn .
( [ bilrel ] ) , we see the `` intersection form '' @xmath123 that was introduced in eqn .
( [ tomega ] ) .
thus , it is natural to introduce a space @xmath198 that is endowed with this quadratic form ( tensored with @xmath199 ) .
we denote the intersection form on @xmath200 as @xmath201 .
we combine the whole collection of even and odd periods of @xmath65 to a vector @xmath202 : @xmath203 similarly , the periods of @xmath127 combine to @xmath204 .
( [ bilrel ] ) is equivalent to @xmath205 . in other words , by analogy with the classical case
, the bilinear relations assert that the subspace @xmath206 that is spanned by the periods of holomorphic 1-forms is an isotropic subspace : the bilinear form @xmath201 vanishes when restricted to @xmath207 . in section [ dimform ] , we will show that generically , for @xmath208 ( or for @xmath209 with even spin structure ) the space of closed holomorphic 1-forms on @xmath4 has dimension @xmath210 . in this case , @xmath207 is middle - dimensional in @xmath200 , and thus it is a maximal isotropic subspace of @xmath200 , again by analogy with the classical case .
just as in the more familiar case @xmath209 , the information about a maximal isotropic subspace of @xmath200 can generically be encoded by a super period matrix .
the super period matrix will now be a @xmath2 matrix that will be symmetric in the @xmath3-graded sense , as described more concretely below . to define the classical period matrix of an ordinary riemann surface @xmath95 ,
one starts by picking a set of @xmath6-periods .
this amounts to picking a maximal isotropic subspace of a particularly simple and convenient sort for the intersection form on @xmath211 . to generalize this for a super riemann surface with ramond punctures
, we similarly must first pick a simple maximal isotropic subspace for the form @xmath123 , which we regard as the superanalog of the classical intersection form .
we again use a set of @xmath6-periods as a maximal set of even null vectors , but what is a natural set of odd null vectors ?
the simplest choice seems to be to order the fermionic periods as @xmath212 and then form the complex linear combinations @xmath213 the complementary fermionic periods are @xmath214 the @xmath215 and @xmath216 will be the fermionic analogs of @xmath6-periods and @xmath78-periods .
now we define a basis of closed holomorphic 1-forms @xmath217 by requiring @xmath218 and @xmath219 for this definition to make sense , the choice of fermionic @xmath6-periods must be generic enough so that closed holomorphic 1-forms obeying the conditions ( [ zorb ] ) and ( [ omorb ] ) exist and are unique .
the condition for this is that if @xmath220 is the subspace of @xmath200 defined by @xmath221 , then we must have @xmath222 .
we return to this condition in sections [ nonperiods ] and [ anex ] , and for now just remark that it places a non - trivial constraint on the choices of ordering and signs in the definition of the @xmath223 .
finally , we define the super period matrix @xmath224 by specifying its matrix elements @xmath225 as in the classical theory , the bilinear relations ( [ bilrel ] ) imply that @xmath226 is symmetric in the @xmath3-graded sense : @xmath227 , @xmath228 , @xmath229 .
if we write the super period matrix in blocks @xmath230 then the upper left @xmath5 block , which we will call @xmath231 , corresponds most closely to the ordinary period matrix in the classical theory of riemann surfaces .
we will call this the pseudoclassical block .
if reduced modulo odd variables , it coincides with the ordinary period matrix of the reduced space @xmath16 .
this will be clear in section [ anex ] .
so in particular @xmath231 has positive - definite imaginary part .
note that the pseudoclassical block depends on the choice of fermionic @xmath6-periods ( though this dependence disappears if we reduce modulo the odd variables ) , since the definition of the @xmath232 depends on that choice .
to count the closed holomorphic 1-forms on a genus @xmath0 super riemann surface @xmath4 , first assume that @xmath35 is split , with a reduced space @xmath11 whose canonical bundle and tangent bundle we denote as @xmath233 and @xmath234 . to begin with ,
assume there are no ramond punctures , then @xmath35 can be constructed as the total space of an odd line bundle @xmath235 .
here @xmath236 is a square root of @xmath234 corresponding to a choice of spin structure , the inverse of @xmath236 will be denoted @xmath237 , and for a line bundle @xmath238 , @xmath239 is @xmath238 with the fiber understood to be odd .
there is a natural projection @xmath240 ( and an embedding of @xmath11 in @xmath35 as the zero - section of @xmath241 ) .
there is always a @xmath0-dimensional space of holomorphic 1-forms @xmath242 on @xmath16 .
these can be pulled back via @xmath19 to closed holomorphic 1-forms on @xmath35 .
there actually are additional odd holomorphic 1-forms @xmath243 , but they are not closed .
the situation for even closed holomorphic 1-forms is more interesting .
an even holomorphic 1-form is in general @xmath244 , but for @xmath135 to be closed , this expression must reduce to @xmath245 . here in classical geometry
@xmath246 represents a holomorphic section of @xmath237 , that is , an element of @xmath247 .
generically , if the spin structure of @xmath4 is even , @xmath248 . in this case
the space of closed holomorphic 1-forms has dimension @xmath249 ( we will reverse the parity in writing dimension formulas ) .
their periods are used to define a @xmath5 super period matrix @xmath250 , which is symmetric and has positive definite imaginary part , just as in the classical case . again assuming that @xmath4 has even spin structure , in genus @xmath251 , there is a divisor @xmath252 along which @xmath253 .
the space of closed holomorphic 1-forms is then of dimension @xmath254 , for some ( even ) @xmath255 .
there are more closed holomorphic 1-forms than periods so some closed holomorphic 1-forms must have vanishing periods .
in fact , as explained in the last paragraph , the even closed holomorphic 1-forms are exact ( @xmath245 ) , so their periods vanish . in defining a super period matrix , one can take the quotient of the space of closed holomorphic 1-forms by the subspace consisting of those whose periods vanish . for @xmath4 split
, the quotient space always has dimension @xmath249 .
so as long as @xmath4 is split , the condition @xmath256 does not lead to trouble in defining the super period matrix . to define the super period matrix without assuming that @xmath4 is split , we need to know that ( away from @xmath257 ) the space of closed holomorphic 1-forms is still of dimension @xmath249 when the odd moduli of @xmath35 are introduced .
this is true but not completely trivial .
one elegant proof uses the fact that there is a natural 1 - 1 correspondence between closed holomorphic 1-forms and holomorphic sections of @xmath258 , the berezinian line bundle of @xmath4 .
the correspondence is given by an explicit formula ; in local superconformal coordinates , a holomorphic section @xmath259 $ ] of @xmath260 corresponds to the closed holomorphic 1-form @xmath261 . to show that the space of holomorphic 1-forms on @xmath35 is still of dimension @xmath249 when @xmath4 is not split
, one must show the analogous statement for @xmath262 : its dimension should not jump when odd moduli are turned on . on general grounds ,
this is true if and only if @xmath263 varies as the fiber of a locally - free sheaf ( or vector bundle ) .
but @xmath264 is serre - dual to @xmath265 , where @xmath266 is the sheaf of holomorphic sections on @xmath4 .
away from the divisor @xmath257 , one has @xmath267 , generated by the constant function 1 . in particular , @xmath265 is locally - free , and hence also are @xmath264 and @xmath268 .
this reasoning fails along the divisor @xmath252 , because given @xmath269 , there is an odd holomorphic function @xmath270 on @xmath4 .
thus for a split super riemann surface @xmath4 , vanishing of @xmath247 is a necessary and sufficient condition for @xmath271 .
what actually happens near @xmath257 is that , although the super period matrix is well - defined and holomorphic as long as the odd moduli vanish , or in other words along the split locus @xmath272 , it develops a pole ( with nilpotent residue ) as soon as one varies away from that locus .
this follows from the formula of dhoker and phong @xcite for the dependence of the super period matrix on odd moduli .
( see section 8.3 of @xcite , or appendix [ perturb ] below . )
now let us consider the case that @xmath35 is a super riemann surface with ramond punctures .
again we start with the split case .
we pick in the reduced space @xmath273 of @xmath35 a collection of distinct points @xmath274 that will represent the ramond punctures , and a line bundle @xmath275 endowed with an isomorphism such a line bundle defines what we call a generalized spin structure .
one is free to tensor @xmath275 with a line bundle of order 2 , so for any @xmath276 and any points @xmath277 , there are @xmath278 generalized spin structures . for @xmath209 ,
the choice of @xmath275 is tantamount to an ordinary spin structure on @xmath4 ; the spin structures on @xmath4 can be naturally divided into odd and even ones . for @xmath208 , the @xmath278
generalized spin structures are permuted transitively under monodromy of the points @xmath279 , so there is no notion of an even or odd generalized spin structure . ] @xmath280 or equivalently @xmath281 the line bundle @xmath275 has degree @xmath282 .
@xmath4 is then defined to be the total space of the line bundle @xmath283 . as before , there are projections @xmath284 and an embedding @xmath285 . away from the points @xmath286 ,
the line bundle @xmath275 is a square root of @xmath234 and @xmath4 is an ordinary super riemann surface , which can be described by local superconformal coordinates @xmath46 and superconformal structure generated by @xmath287 .
however , because the isomorphism @xmath288 is only valid away from the points @xmath279 , the superconformal structure breaks down along the divisors @xmath289 .
those divisors are ramond divisors , representing singularities in the superconformal structure of @xmath4 . closed holomorphic 1-forms can be described as before .
holomorphic 1-forms on @xmath11 can be pulled back to give a @xmath249-dimensional space of closed holomorphic 1-forms on @xmath4 .
as before , these are the only odd closed holomorphic 1-forms , and even ones are of the form @xmath290 , where now geometrically @xmath291 is a holomorphic section of @xmath292 .
this line bundle is of degree @xmath293 , so generically @xmath294 is of dimension @xmath276 and @xmath295 .
this fails on a locus @xmath296 characterized by @xmath297 .
this locus , which will be studied in section [ badset ] , is a rough analog of the theta - null divisor @xmath257 for @xmath209 .
overall , away from @xmath296 , the space of closed holomorphic 1-forms on a split super riemann surface @xmath4 has dimension @xmath210 , as assumed in section [ superp ] in defining the super period matrix . to show that this remains so
if @xmath35 is not assumed to be split , one can adapt the arguments that are used in the absence of ramond punctures . in the presence of ramond punctures ,
closed holomorphic 1-forms correspond ( see appendix d.1 of @xcite ) not to elements of @xmath298 , but to elements of @xmath299 , where a section of @xmath300 is a section of @xmath258 that is allowed to have a simple pole , with @xmath69-independent residue , along a ramond divisor .
the space of closed holomorphic 1-forms is locally - free if @xmath301 is locally - free . by serre duality , this is equivalent to @xmath302 being locally - free , where @xmath303 is the sheaf of holomorphic functions on @xmath35 that are constant when restricted to a ramond divisor .
equivalently , @xmath304 , where @xmath305 is a complex supermanifold obtained from @xmath4 by blowing down the ramond divisors ( this blowdown operation is discussed in @xcite , section 3.4.2 ) .
the locally - free condition @xmath306 is equivalent to the familiar condition @xmath307 .
the last claim is shown as follows .
it suffices to assume that @xmath4 is split , in which case @xmath308 is the total space of the line bundle @xmath309 , so that an odd holomorphic function on @xmath308 ( which would obstruct the claim that @xmath310 ) corresponds to an element of @xmath311 . as a consequence of ( [ zelmonk ] )
, we have @xmath312 by serre duality , this vanishes if and only if @xmath307 .
as long as this is true , the space of closed holomorphic 1-forms has the expected dimension .. in defining the super period matrix in section [ superp ] , we assumed that the space @xmath207 of periods of closed holomorphic 1-forms is middle - dimensional in the space @xmath200 of periods .
we will now show that this is true for any split super riemann surface @xmath4 .
as long as the locally - free condition @xmath313 is satisfied , @xmath207 automatically remains middle - dimensional when @xmath4 is deformed away from the split locus .
( since odd moduli are infinitesimal , turning them on will not cause a nonzero period to become zero , so it will not reduce the dimension of @xmath207 . on the other hand , the bilinear relations ( [ bilrel ] )
ensure that the dimension of @xmath207 can not increase . ) for @xmath4 split , the following analysis will show that @xmath207 is middle - dimensional even for @xmath314 , but in that case , we can not say anything simple about what happens away from the split locus .
middle - dimensionality of the even periods on a split super riemann surface @xmath4 just amounts to the classical fact that on an ordinary riemann surface @xmath16 , the periods of holomorphic differentials are middle - dimensional in the space of all @xmath6- and @xmath78-periods .
we will now show that the same is true for the odd periods . for a line bundle @xmath315
, we write @xmath316 for the dimension of @xmath317 .
since @xmath318 has degree @xmath293 , the riemann - roch theorem gives @xmath319 . via serre duality , this is equivalent to @xmath320 . in view of ( [ welbo ] ) , this is equivalent to @xmath321 a section of @xmath322 is simply a section of @xmath318 that vanishes at @xmath277 , so @xmath323 is a subspace of @xmath324 .
( [ welmonk ] ) says that the quotient space has dimension @xmath276 : @xmath325 now , @xmath324 is the space of even closed holomorphic 1-forms , and @xmath326 is its subspace consisting of those that vanish when restricted to ramond divisors or in other words whose odd periods vanish .
so eqn . ( [ molfonk ] )
says that for any split super riemann surface @xmath4 , the space of even closed holomorphic 1-forms modulo those with vanishing periods is of dimension @xmath276 . in other words , for @xmath4 split or for @xmath327
, the periods of closed holomorphic 1-forms always span a middle - dimensional subspace @xmath206 .
next we will look more closely at the super period matrix of a split super riemann surface @xmath35 .
odd closed holomorphic 1-forms on @xmath35 are simply pullbacks of holomorphic 1-forms on @xmath16 .
their periods are just the corresponding periods on @xmath16 .
so the pseudoclassical block @xmath231 of the period matrix of @xmath35 is just the classical period matrix of @xmath11 .
a form on @xmath4 that is a pullback from @xmath16 has no @xmath70 component .
so its odd periods vanish , and hence @xmath328 .
thus the super period matrix of @xmath4 is @xmath329 where only @xmath330 remains to be understood . for this , we first recall that an even holomorphic 1-form on @xmath4 is exact , @xmath331 for some @xmath332 , so its ordinary @xmath6- and @xmath78-periods that is , its even periods vanish .
( this gives another explanation of why @xmath328 . )
now suppose that @xmath333 and @xmath334 are two even closed holomorphic 1-forms on @xmath4 , with respective odd periods @xmath187 and @xmath188 .
we combine the odd periods of @xmath333 and @xmath334 into vectors @xmath335 , which take values in a vector space @xmath336 of dimension @xmath1 that has a basis corresponding to the oriented ramond divisors @xmath178 .
specialized to the case that the even periods are zero , the bilinear relation of eqn .
( [ bilrel ] ) reduces to @xmath337 we can see very directly why this is true .
suppose that @xmath338 , @xmath339 , with @xmath340 .
the product @xmath341 is then a section of @xmath342 , but the isomorphism in ( [ elmonk ] ) identifies this with @xmath343 . a section of @xmath343 is a meromorphic 1-form that may have simple poles at the points @xmath277 ; the residue of the pole of @xmath341 at @xmath344 is the product @xmath345 . thus eqn .
( [ mondo ] ) asserts the vanishing of the sum of residues of a certain meromorphic 1-form . for an even more explicit example
, we consider a super riemann surface @xmath4 of genus 0 with two ramond punctures .
we parametrize @xmath4 or more precisely , the complement of a divisor in @xmath4 by coordinates @xmath46 with superconformal structure defined by @xmath346 the ramond punctures are at @xmath56 and @xmath347 . to understand what is happening at @xmath347
, we introduce new coordinates via @xmath348 , @xmath349 , whence @xmath350 the factor of @xmath351 is not important here , since we only care about the subbundle of the cotangent bundle of @xmath4 that is generated by @xmath84 .
thus the superconformal structure near @xmath352 is generated by @xmath353 , showing that @xmath352 is a ramond divisor and that the coordinate system @xmath354 puts the superconformal structure in a standard form near this divisor .
notice that the factor of @xmath355 in the formula @xmath349 is necessary for this result .
there are no nonzero odd closed holomorphic 1-forms on @xmath4 , and the space of even ones is one - dimensional , generated by @xmath356 .
the odd periods of @xmath333 are 1 at @xmath56 and @xmath355 at @xmath352 , so the sum of squares of the odd periods is zero , as expected .
of course , we could also describe the same example more globally in projective coordinates . for this
, we simply take @xmath4 to be a weighted projective superspace @xmath357 with even homogeneous coordinates @xmath358 of weight 1 and an odd homogeneous coordinate @xmath69 of weight 0 .
the superconformal structure of @xmath4 can be defined by the following section of @xmath359 : @xmath360 more generally , we can describe a split genus 0 super riemann surface @xmath4 with @xmath1 ramond punctures in affine coordinates @xmath46 by the superconformal structure @xmath361 alternatively , @xmath35 is the weighted projective space @xmath362 with homogeneous coordinates @xmath363 of weights
@xmath364 and superconformal structure defined by @xmath365 with @xmath366 a homogeneous polynomial of degree @xmath1 .
it is instructive to consider this last example more carefully to see what is involved in orienting the ramond divisors . to orient the ramond divisor @xmath367 at , say , @xmath368
, we should pick new coordinates @xmath369 such that @xmath370 takes the standard form @xmath371 near @xmath368 .
then we orient @xmath367 by choosing the differential @xmath372 along @xmath367 .
clearly , we need @xmath373 now let us ask what happens to the orientations of the ramond divisors @xmath367 when the moduli @xmath374 are varied . to keep things simple , we consider the ramond divisors to be labeled ( as is usually most natural in string theory ) , so we do not allow permutations of the @xmath375 ; we consider only what happens when the @xmath375 are braided around each other .
such a process is made by composing elementary moves in which one of the @xmath376 s , say @xmath377 , makes a small loop around another , say @xmath378 . in the process , @xmath379 changes sign , but there are no sign changes in @xmath380 for any other pair .
so the orientations of @xmath381 and @xmath382 are reversed , and no others . combining any number of operations of this kind
, we see that the only constraint is that there are an even number of orientation reversals . in other words ,
the only constraint is that the monodromies preserve the orientation of the space @xmath336 that parametrizes the odd periods .
this has a simple explanation .
@xmath336 is an even - dimensional vector space with a non - degenerate quadratic form .
in such a vector space , there are two families of maximal isotropic subspaces , associated to a choice of orientation .
@xmath336 has a distinguished middle - dimensional isotropic subspace , the space @xmath383 of odd periods of odd holomorphic differentials .
so it has a preferred orientation that is preserved under any monodromies .
our application in the remainder of this paper will involve the case of a super riemann surface @xmath4 with precisely 2 ramond punctures .
so let us point out some particularly nice things that happen in this case . with two ramond punctures ,
there are precisely 2 odd periods , say @xmath384 and @xmath385 , so the space @xmath336 of odd periods only has two null subspaces , generated by @xmath386 . the choice of a fermionic @xmath6-period in this case ( eqn .
( [ melz ] ) ) is particularly simple .
there is only one fermionic @xmath6-period @xmath91 , and up to an integer power of @xmath355 ( which will arise if we exchange @xmath384 and @xmath385 , or reverse their orientations ) , it must be either @xmath387 or @xmath388 .
however , only one of the two choices is viable in the definition of the super period matrix . according to eqn .
( [ omorb ] ) , we are supposed to find an even closed holomorphic 1-form @xmath333 with @xmath389 .
however , for @xmath4 split , @xmath333 is a null vector in @xmath336 , so its periods obey @xmath390 , with one choice of the sign or the other .
this means that if we choose the wrong sign in the definition of the fermionic @xmath6-period @xmath91 , then we will get @xmath391 and will be unable to satisfy @xmath389 .
so a unique definition of @xmath392 is forced upon us , up to an integer power of @xmath355 . moreover , with this choice , as long as the space of closed holomorphic 1-forms has the expected dimension @xmath393 , a unique set of forms obeying ( [ zorb ] ) and ( [ omorb ] ) always exists , hence , the super period matrix is always defined away from the usual locus @xmath296 along which @xmath394 .
if we do multiply @xmath91 by @xmath395 , for some integer @xmath396 , what happens to @xmath226 ? to compensate for the change in @xmath91 , we will have to multiply @xmath397 and @xmath333 by @xmath398 .
@xmath330 is multiplied by @xmath399 and @xmath400 and @xmath401 are multiplied by @xmath398 .
@xmath231 is unchanged . with more than 2 ramond punctures ,
there are more choices in defining the fermionic @xmath6-periods .
the nondegeneracy condition in the definition of the super period matrix is more complicated , and the super period matrix has poles when this condition fails .
we consider this next . in the definition of the super period matrix
, we needed to know that closed holomorphic 1-forms obeying the conditions ( [ zorb ] ) and ( [ omorb ] ) exist and are unique . saying that a system of linear equations ( with the same number of variables and unknowns ) has a unique solution is an open condition , so it suffices to consider the case that @xmath4 is split .
then the condition is simply that it should be possible to find an even differential @xmath333 with prescribed values of half of its odd periods @xmath402 , @xmath403 ( and no condition on the other odd periods @xmath404 ) .
( this is clearly equivalent to the existence of differentials @xmath405 with @xmath406 . ) on dimensional grounds , an equivalent statement is the following .
let @xmath407 have a basis corresponding to oriented ramond divisors , let @xmath383 be the middle - dimensional isotropic subspace of @xmath336 that parametrizes odd periods of closed holomorphic 1-forms , and let @xmath408 be the middle - dimensional isotropic subspace of @xmath336 characterized by @xmath409 . then the desired condition is @xmath410 .
as remarked at the end of section [ anex ] , middle - dimensional isotropic subspaces of @xmath336 come in two families associated with a choice of orientation of @xmath336 .
a necessary condition for @xmath411 is that @xmath383 and @xmath408 should be oppositely oriented ( that is , associated to opposite orientations of @xmath336 ) if @xmath276 is odd , or oriented the same way if @xmath276 is even .
conversely , if the orientations of two middle - dimensional isotropic subspaces @xmath383 and @xmath408 make this possible , then generically @xmath410 .
one may therefore expect that as long as the right orientation is used in defining the fermionic @xmath6-periods , the nondegeneracy condition @xmath411 will be satisfied generically , on the complement of a divisor in the reduced space of @xmath412 . to show that this is true , it suffices to show it for @xmath413 , since a super riemann surface of any genus with @xmath1 ramond punctures can degenerate to several components one of which is a genus 0 surface containing all of the ramond punctures . using the explicit description ( [ efinit ] ) of a super riemann surface of genus @xmath0 with @xmath1 ramond punctures , one can show directly that , with the right definition of the fermionic @xmath6-periods , the nondegeneracy condition is obeyed generically .
pick @xmath276 distinct points @xmath414 and consider a limit with @xmath415 near @xmath416 for @xmath417 .
explicitly we find that in this limit , with fermionic @xmath6-periods @xmath418 defined as in eqn .
( [ melz ] ) , the differentials @xmath419 that satisfy @xmath420 are @xmath421 with @xmath422 we do not know a useful characterization of the divisor on which @xmath423 , producing additional poles in the super period matrix . here
we will make some observations about the bad set @xmath296 in moduli space along which @xmath424 .
let @xmath425 be a line bundle of degree @xmath426 , determined by a point in @xmath427 , the component of the jacobian that parametrizes line bundles of that degree .
generically , the condition @xmath428 is satisfied only in codimension @xmath429 in @xmath427 . for example , if @xmath430 , then @xmath431 if and only if @xmath432 , which determines a unique point in @xmath427 , of codimension @xmath0 . however , we are interested in a line bundle @xmath318 with an isomorphism @xmath433 this case is somewhat exceptional for small @xmath276 .
we will examine this in detail , since our application later in this paper involves @xmath434 . for @xmath209 ,
@xmath318 is simply a square root of @xmath233 .
the codimension along which @xmath435 is 0 or 1 depending on whether @xmath318 defines an odd or even spin structure .
the case @xmath434 is somewhat similar to @xmath209 with odd spin structure : the condition @xmath436 is satisfied in codimension 1 , not in the `` expected '' codimension @xmath437 . before explaining this in detail
, we first consider some small genus cases that are relevant to our applications .
our main application later in this paper involves genus 2 . a genus 2 riemann surface @xmath11 is hyperellipic , and is a two - fold cover @xmath438 . for @xmath434 ,
the line bundle @xmath318 has degree 2 .
the riemann - roch formula gives @xmath439 , and serre duality gives @xmath440 .
so @xmath441 is equivalent to @xmath442 .
but @xmath443 is of degree 0 .
a line bundle of degree 0 with a holomorphic section must be trivial , so @xmath444 and @xmath445 . since also @xmath446
, we must have @xmath447 . as explained in section [ specialcase ] , this is so if and only if the two points @xmath448 are exchanged by the hyperelliptic involution of @xmath35 .
there is a one - parameter family of such pairs .
this family is of codimension 1 in the space of all pairs @xmath448 , so the exceptional set @xmath296 is a divisor @xmath257 in this case
. we will also consider in section [ genthree ] the case of genus 3 .
a generic genus 3 riemann surface @xmath11 , in affine coordinates , is described as a plane curve @xmath449 , where @xmath450 is a quartic polynomial in two variables .
a canonical divisor is the intersection of @xmath11 with a line @xmath451 in the plane .
a line is of course defined by a linear equation @xmath452 .
pick any point @xmath453 and let @xmath451 be the line tangent to @xmath11 at @xmath91 .
the equations @xmath454 , which describe intersections of @xmath4 and @xmath451 , will be satisfied at four points in @xmath455 , counted with multiplicity .
the point @xmath91 of tangency has multiplicity 2 , so @xmath11 and @xmath451 intersect at two other points @xmath448 .
generically these are distinct points with multiplicity 1 each .
so @xmath456 .
thus the line bundle @xmath457 admits an isomorphism @xmath446 .
the line bundle @xmath458 has a non - zero holomorphic section ( the section `` 1 '' that vanishes precisely at @xmath91 ) . since @xmath459 has degree 1 , we have @xmath460 ( a degree 1 line bundle @xmath238 over a curve of positive genus always has @xmath461 ) and the riemann - roch formula @xmath462 implies that @xmath463 . but by serre duality @xmath464 .
so @xmath465 .
thus we have found a 1-parameter family of pairs @xmath448 , parametrized by @xmath466 , such that @xmath436 , showing that again the exceptional set @xmath296 is of codimension 1 .
now let @xmath11 have arbitrary genus @xmath0 . for @xmath434 ,
the line bundle @xmath467 has degree @xmath468 , so one expects that there is a divisor @xmath257 in the moduli space along which @xmath469 .
along @xmath257 , let @xmath40 be a nonzero holomorphic section of @xmath467 .
then @xmath470 is a holomorphic section of @xmath471 . in other words
, @xmath470 is a meromorphic section of @xmath233 that is holomorphic except possibly for a single pole at @xmath472 .
since the sum of residues of a meromorphic section of @xmath233 must vanish , @xmath470 is actually holomorphic at @xmath472 .
this means that @xmath40 must vanish at @xmath472 , so @xmath40 is a holomorphic section of @xmath473 .
thus along @xmath257 , @xmath474 . since @xmath473 has degree @xmath475 , riemann - roch implies that @xmath476 , so @xmath477 . by serre duality ,
this is equivalent to @xmath478 . finally , using the isomorphism ( [ monx ] ) , this is equivalent to @xmath441 . in short , along the divisor @xmath257 ,
one has @xmath441 . along this divisor , the super period matrix has a pole with nilpotent residue , as will be described in appendix [ perturb ] . for @xmath479 ,
it is likely that the exceptional set @xmath296 has codimension greater than 1 , but we will not analyze this case in detail .
we should note that although the choice of fermionic @xmath6-periods is essentially unique for @xmath434 , as described in section [ tworp ] , for @xmath479 , one requires a somewhat arbitrary choice of fermionic @xmath6-periods and this introduces artificial singularities in codimension 1 . thus describing the periods by a period matrix
is less natural for @xmath479 than it is for @xmath434 .
it is perhaps more natural for @xmath479 to simply study the lagrangian submanifold spanned by the periods , rather than to define a period matrix .
in the rest of this paper , we navigate toward an application of the super period matrix with ramond punctures that was described in the introduction .
the application mainly involves super riemann surfaces of genus 2 , so we begin by explaining some special facts about the super period matrix for small values of the genus . in genus
@xmath29 , any @xmath5 complex symmetric matrix with positive imaginary part is the period matrix of an ordinary riemann surface @xmath95 , which is unique up to isomorphism .
( this is not true for @xmath480 ; for a symmetric matrix of positive imaginary part to be a period matrix , it must obey the schottky relations .
that is why the following construction is limited to @xmath29 . )
if @xmath95 is a riemann surface of genus @xmath28 with even spin structure , then @xmath481 is zero always this statement is true for a smooth riemann surface of genus 2 .
however , a smooth curve of genus 2 with even spin structure can degenerate to a pair of genus 1 components each with odd spin structure , meeting at a point ; for such a singular curve , the appropriate analog of @xmath481 is non - zero .
thus the divisor @xmath257 has a component at infinity in the deligne - mumford compactification of @xmath482 . ] ( and not just generically , as is the case for @xmath251 ) , so any super riemann surface of genus 2 with even spin structure has a super period matrix @xmath226 . by mapping a super riemann surface @xmath4 to the ordinary riemann surface with the same period matrix ,
we get for @xmath483 a natural holomorphic map , we generically use the symbol @xmath19 to denote a projection defined using the period matrix , or its pseudoclassical block in the presence of ramond punctures . ]
@xmath484 , where @xmath485 parametrizes super riemann surfaces of genus @xmath0 with even spin structure , and its reduced space @xmath486 parametrizes an ordinary riemann surface of genus @xmath0 also with even spin structure . for @xmath487 ,
this construction is trivial , as a genus 1 super riemann surface with even spin structure has no odd moduli and @xmath488 .
however , for @xmath17 , the statement is non - trivial and has been exploited by dhoker and phong in computing superstring scattering amplitudes , as summarized in @xcite .
we will ultimately focus mostly on this case . for @xmath489 ,
since every @xmath490 complex matrix of positive definite imaginary part is a period matrix , we can use the same construction to define a meromorphic projection @xmath491 , but now @xmath19 has poles ( with nilpotent residue ) . as explained in appendix
[ hyperelliptic ] , @xmath19 has for genus 3 a fairly obvious pole along the divisor @xmath257 where the super period matrix has a pole , and a somewhat less obvious pole along a second divisor @xmath492 .
we can do something somewhat similar for @xmath412 , which parametrizes a genus @xmath0 super riemann surface @xmath4 with @xmath1 ramond punctures and no ns punctures .
once we pick a set of fermionic @xmath6-periods , we can define a super period matrix @xmath226 , which in particular has the @xmath5 pseudoclassical block @xmath231 . for @xmath29 , mapping @xmath4 to an ordinary riemann surface @xmath95 whose period matrix @xmath493
coincides with @xmath231 gives a map @xmath494 the fiber @xmath495 parametrizes all moduli of @xmath35 other than its super period matrix .
note that in ( [ moxo ] ) , the base space is simply @xmath496 , with no memory of the generalized spin structure . in the absence of ramond punctures , the analogous projection @xmath497 can be defined to remember the spin structure .
we can not do something analogous in the case of a projection from @xmath412 to @xmath496 ; since the definition of a generalized spin structure ( a line bundle @xmath275 with the isomorphism in eqn .
( [ elmonk ] ) ) depends on the positions of the ramond punctures , there is no way to forget the ramond punctures while remembering the generalized spin structure .
that is why anything that one can deduce from the fibration @xmath19 for @xmath208 notably the vanishing under certain conditions of the dilaton tadpole will involve a sum over spin structures .
the nicest case of the fibration @xmath498 and the case that we will use in our application is for @xmath434 , for then the choice of a fermionic @xmath6-period is essentially unique ( and @xmath231 is entirely unique ) as we saw in section [ tworp ] . for @xmath479 ,
the projection @xmath19 does depend on the choice of fermionic @xmath6-periods and moreover it has unphysical singularities that depend on that choice .
keeping @xmath29 , there is no trouble to include ns punctures in this picture , since an ns puncture is just a marked point .
there is therefore a forgetful map for ns punctures , and an ns puncture does not affect the definition of the super period matrix . composing the map @xmath499 which forgets the ns punctures with the projection @xmath19 , we get a projection @xmath500 where now the fiber @xmath501 parametrizes also the positions of the ns punctures .
our application will involve the case @xmath434 ( precisely 2 ramond punctures ) and @xmath502 ( no ns punctures ) .
we will also encounter the case @xmath209 , @xmath503 .
the projection @xmath504 that we will use has poles associated to poles of the super period matrix .
the complications associated to those poles can be overcome , at least for certain purposes .
we will analyze this in section [ lowgenus ] , but there are a number of things to explain first .
all statements in this section have been made for the uncompactified moduli spaces .
some care is required to extend these statements over the corresponding deligne - mumford compactifications .
for example , in the absence of ramond punctures , the projection @xmath505 used by dhoker and phong develops a pole if one attempts to extend it over the deligne - mumford compactifications of these spaces .
this has been explained in @xcite ( the underlying reason was explained in footnote [ zelf ] above ) , and is important in understanding the behavior near infinity of the measure on @xmath15 computed by dhoker and phong .
a certain subtlety of superstring theory is important in our application . before explaining it , we begin by recalling what happens in bosonic string theory .
consider a closed oriented bosonic string vacuum such that at tree level the matter system has a continuous symmetry ,-
branes and/or an orientifold projection can be included in the following discussion if the @xmath26-brane boundary condition and/or the orientifold projection are chosen to preserve the symmetry . ] with a conserved charge @xmath506 associated to a conserved worldsheet current @xmath507 .
such a symmetry constrains the worldsheet correlation functions on any worldsheet .
let @xmath508 be an arbitrary set of matter vertex operators that are eigenstates of @xmath506 ( and are conformal primaries of the appropriate dimension ) .
then the correlation function @xmath509 on a worldsheet @xmath510 of any genus @xmath0 will vanish unless the sum of the charges of the operators @xmath511 vanishes .
thus , the existence of a conserved worldsheet current leads automatically to a conservation law for genus @xmath0 amplitudes , for any @xmath0 . as a result ,
in closed oriented bosonic string theory , a continuous symmetry at tree level remains valid as a continuous symmetry to all orders of perturbation theory .
this result holds for any given riemann surface @xmath95 ; it does not in any way involve integration over moduli space .
the same argument holds for continuous symmetries of superstring theory that arise from the ns sector .
thus , loop corrections do not trigger spontaneous breakdown of a continuous global symmetry that arises in closed oriented bosonic string theory or the ns sector of superstring theory .
however , we can not make such a simple argument for spacetime supersymmetry , which comes from the ramond sector .
the analog of the conserved current @xmath507 is the usual fermionic vertex operator @xmath512 ( here @xmath6 is a spinor index and @xmath512 is the combined spin field of matter and ghosts ) . the considerations of this paper apply equally to heterotic or type ii superstrings , but for simplicity , and also with a view toward our eventual application , we consider the heterotic string , in which spacetime supersymmetry comes from the holomorphic part of the worldsheet theory only . then @xmath512 is a holomorphic object , but it can not be understood as a conserved current on a fixed super riemann surface @xmath4 ; indeed , it is inserted at a ramond puncture , which is a singularity in the superconformal structure of @xmath4 , and there is no notion of moving a ramond puncture while otherwise leaving @xmath4 unchanged . instead , the proof of a supersymmetric ward identity proceeds essentially by constructing a conserved current on the super moduli space rather than on the super riemann surface @xmath4 . to do this , we combine the spin field with the ordinary @xmath83 ghost to make a dimension @xmath513 superconformal primary @xmath514 .
we similarly combine the matter vertex operators @xmath511 with ghosts in the usual way to make superconformal primaries @xmath515 of dimension @xmath513 .
( for example , if @xmath511 is a vertex operator of the ns sector , then we set @xmath516 , where @xmath517 are the holomorphic superconformal ghost fields and @xmath518 is the antiholomorphic ghost field . )
the correlation function @xmath519 does not define a measure on supermoduli space or more precisely in the product of holomorphic and antiholomorphic moduli spaces ( in the case of the heterotic string , the holomorphic moduli space is the super moduli space @xmath520 and the antiholomorphic space is its reduced space with complex structure reversed ; one can take the reduced space of @xmath521 to coincide with the reduced space of @xmath520 ) .
the distinction between @xmath520 and @xmath521 will not be very important in the present paper , but we express our statements in terms of integration over @xmath521 since this is more accurate . ] on the appropriate integration cycle @xmath521 of superstring perturbation theory because the operator @xmath522 has ghost number less by 1 than the ghost number of a standard ramond - sector vertex operator . as a result
, @xmath523 represents not a measure on @xmath521 that could be integrated to compute a scattering amplitude , but an integral form of codimension 1 the supermanifold analog of a conserved current ( see for example @xcite or @xcite ) .
brst - invariance of @xmath522 and of the @xmath515 implies that this form is closed , @xmath524 .
using this fact and the supermanifold version of stokes s theorem , we derive a supersymmetric ward identity : @xmath525 on the right hand side , @xmath526 is a sum of components `` at infinity '' in @xmath521 ; they correspond to different ways that @xmath4 might degenerate .
although in general @xmath4 has various separating and nonseparating degenerations , there are , as explained in @xcite ( and more fully in section 8 of @xcite ) , only two types of degeneration that contribute to the ward identity .
one such degeneration is the case that @xmath4 degenerates to a union of two components , one of which contains @xmath522 and just one of the @xmath515 .
such a degeneration is sketched in fig .
[ supersymmetry ] . as is explained in @xcite ,
if only such components of @xmath526 contribute on the right hand side of ( [ melf ] ) , then ( [ melf ] ) becomes a standard ward identity of unbroken supersymmetry and the genus @xmath0 contribution to the @xmath71-matrix is spacetime supersymmetric . in general
, there may also be a goldstone fermion contribution to the ward identity ; it can arise from a degeneration in which @xmath4 splits off a positive genus contribution @xmath527 that contains @xmath522 but no other vertex operator , as sketched in fig .
[ goldstone ] .
just as in field theory , if the genus @xmath0 ward identity receives a goldstone fermion contribution , then the genus @xmath0 contribution to the @xmath71-matrix is not spacetime supersymmetric . contains a supersymmetry generator @xmath528 and precisely one more vertex operator . if these are the only contributions , then the ward identity expresses the invariance of the @xmath71-matrix under spacetime supersymmetry .
the usual case , as sketched here , is that @xmath527 has genus 0 .
this leads to the familiar tree - level expressions for the supercharges .
, width=240 ] is positive.,width=240 ] the importance of the fibrations that were described in section [ twop ] is that in favorable cases , they can be used to establish supersymmetric ward identities in which one knows _ a priori _ that there can be no goldstone fermion contribution . rather than try to explain abstractly how that can happen , we will first describe the problem that we have in mind .
we turn to the two - loop vacuum amplitude of heterotic string theory .
potential contributions come from worldsheets with even spin structure only ( worldsheets with odd spin structure contribute to parity - violating amplitudes ) , so the vacuum amplitude is found by integration over @xmath14 .
an effective procedure @xcite has been to first integrate over the fibers of the projection @xmath529 , so as to reduce the vacuum amplitude to an integral over the bosonic moduli space @xmath15 . at that point
, it makes sense to sum over spin structures without integrating over any additional moduli .
( this does not make sense before reducing to the bosonic moduli space , since there is no notion of changing the spin structure of a super riemann surface while otherwise leaving the surface unchanged . )
quite a few supersymmetric compactifications to @xmath10 dimensions have been studied this way . in each case , after reducing to @xmath15 , the two - loop vacuum amplitude vanishes upon summing over spin structures , even without any integral over the bosonic moduli .
our goal here is to explain this and show that it is true in general .
as we recalled in the introduction , in general , in a supersymmetric compactification to four dimensions , the procedure of integrating over @xmath15 by integrating first over the fibers of the projection @xmath530 misses a contribution at infinity .
so our result means that in such compactifications , the full two - loop vacuum amplitude is given by the contribution at infinity . in supersymmetric compactifications above four dimensions , there is no such contribution at infinity , and the whole two - loop vacuum amplitude vanishes . before discussing the two - loop case , let us recall one way to understand what happens in genus 1 . on a superstring worldsheet @xmath4 of genus 1
, we consider a two point function @xmath531 here as before @xmath532 is the spacetime supersymmetry generator with the ghost field @xmath83 included . on the other hand , we take @xmath533 to be the vertex operator for the dilatino the spin 1/2 partner of the dilaton at zero momentum . for superstring theory in @xmath24 , @xmath534 where @xmath535 are matter superfields representing the motion of the strings in @xmath24 , @xmath536 is once again the holomorphic spin field , and @xmath537 , @xmath538 , @xmath539 are gamma matrices . in compactified models , the definition of @xmath533 is changed slightly , but the details will not be important . the notation @xmath540 and @xmath541 is just meant to remind us that @xmath522 is holomorphic while @xmath533 is neither holomorphic nor antiholomorphic .
these two vertex operators are both in the ramond sector , so @xmath4 is a genus 1 surface with 2 ramond punctures .
such a surface has precisely 1 odd modulus .
the characteristic subtleties of superstring perturbation theory result from the possibility of changes of variables such as @xmath542 , where @xmath543 is an even modulus and @xmath544 are odd moduli .
so they do not come into play until there are at least two odd moduli . accordingly , none of these subtleties are relevant to the example under discussion .
except for one important detail , we can think of @xmath545 and @xmath546 as points on the ordinary genus 1 riemann surface @xmath11 , and use the translation symmetries of @xmath11 to set @xmath547 .
the detail in question is as follows : the generalized spin structure of @xmath11 changes when @xmath545 moves around a non - contractible loop in @xmath11 , and hence we should think of @xmath545 as a point in a @xmath548-fold unramified cover of @xmath11 , which we will call @xmath549 , that parametrizes a point in @xmath11 together with a generalized spin structure .
the condition @xmath56 defines a single point on @xmath11 , but on @xmath549 it corresponds to four possible points @xmath550 , which are labeled by the four possible spin structures on @xmath4 or equivalently on @xmath11 .
( the generalized spin structure of @xmath4 reduces to an ordinary spin structure for @xmath551 , where the two ramond punctures coincide . )
the correlation function @xmath552 is a holomorphic 1-form on @xmath549 that has poles only at the points vanishes because of fermion zero modes in the matter system , so there are really only three poles , not four . ]
because of the operator product relation @xmath553 where @xmath554 is the dilaton vertex operator at zero momentum , the residue of each pole is the dilaton tadpole corresponding to the given spin structure .
the vanishing of the sum of the residues of the holomorphic 1-form @xmath552 on @xmath549 means that the dilaton tadpole vanishes , after summing over spin structures and before integration over bosonic moduli .
a standard argument shows that in genus @xmath0 , the dilaton tadpole is @xmath99 times the vacuum amplitude .
so the genus 1 vacuum amplitude likewise vanishes after summing over spin structures , but without integration over bosonic moduli .
the same argument does not work in genus @xmath555 , because there is more than 1 odd modulus , and the subtleties of superstring perturbation theory do come into play .
we can not think of the fermion vertex operator @xmath512 as a conserved current on a fixed worldsheet ; instead , as was summarized in section [ app ] , a proper general argument proceeds by applying the identity ( [ melf ] ) to the correlation function @xmath556 and analyzing all possible boundary contributions , including contributions that involve degenerations of @xmath4 . by such reasoning
one can show that ( in a supersymmetric compactification with no goldstone fermion contribution ) , the genus @xmath0 vacuum amplitude vanishes after integration over all moduli .
there is in general no simpler version of this statement that involves integrating or summing over only some of the moduli .
however , for @xmath17 we can in fact imitate the classical genus 1 argument , with some more care , using the map @xmath557 described in eqn .
( [ moxo ] ) . of course
, we will have to take into account the fact that this map is only generically defined , but this turns out to cause no problem .
the obvious singularities of the correlation function @xmath558 arise from a collision of the two vertex operators ; such a collision gives a pole whose residue is the amplitude on @xmath559 associated to the dilaton tadpole .
we will reduce from @xmath560 to @xmath559 by extracting this residue . by adapting the classical argument that was explained above in genus 1 ,
we will show that if one follows the procedure of dhoker and phong to compute genus 2 amplitudes , then the bulk contribution to the dilaton tadpole and hence to the vacuum amplitude vanishes after integrating over the fiber of @xmath561 , even before integrating over @xmath23 . concretely , integrating over the fiber of @xmath562 means integrating over the position at which the dilaton vertex operator is inserted , and summing over spin structures , but not integrating over the moduli of the underlying genus 2 surface @xmath11 .
the bulk contribution to the genus 2 vacuum amplitude , computed with the dhoker - phong procedure , vanishes before that last integration . to make the analysis , we write @xmath563 for a generic fiber of the projection @xmath504 , and we look at the behavior of @xmath556 near @xmath563 .
we recall that , as discussed in section [ app ] , @xmath564 is not a volume form on @xmath560 but a form of codimension 1 ( technically an integral form of codimension 1 ) . near @xmath563 , we can factorize @xmath556 as @xmath565 , where @xmath65 is a volume form on @xmath23 and @xmath566 is a form of codimension 1 along the fibers of @xmath19 .
the reason that this factorization exists is that the missing `` index '' of the codimension 1 form @xmath556 is tangent to the fibers of @xmath19 , not the base .
this index is missing because @xmath522 is missing a factor of the antiholomorphic ghost field @xmath518 . from an antiholomorphic point of view
, the map @xmath19 is just the forgetful map @xmath567 , and @xmath518 represents a 1-form dual to the motion of the operator @xmath522 along the fiber of this map .
so the missing index is tangent to the `` fiber '' of the fibration @xmath19 , not to the base , and that is why @xmath556 can be factored as the product of the pullback @xmath568 of a full volume form @xmath65 on the base times a form @xmath566 of codimension 1 in the fiber direction .
this factorization is not completely canonical , since @xmath65 could be multiplied by a nonzero function on the base of the fibration , but once we restrict to a particular fiber @xmath563 , @xmath566 depends on the choice of @xmath65 only by an overall multiplicative constant , which will not affect what follows .
we can now make the same argument as in ( [ melf ] ) , but with the fiber @xmath563 replacing the full integration cycle @xmath521 , and @xmath566 replacing @xmath556 .
since @xmath569 , we have @xmath570 here @xmath571 is defined by throwing away a small neighborhood of the singularities of @xmath566 in @xmath563 , and @xmath572 is a union of the boundaries of those small neighborhoods . ) come from collisions of the two vertex operators . from a conformal point of view , these involve the splitting of @xmath4 into two components , of which one is a genus 0 surface with 2 ramond punctures , one ns punctures , and no even or odd moduli , and the second is a genus 2 surface with 1 ns puncture .
the contribution of such a splitting to the identity is the bulk contribution to the genus 2 dilaton tadpole , computed with the procedure of dhoker and phong .
( the dilaton vertex operator is inserted on the genus 2 component at the point where it meets the second component.),width=336 ] the obvious singularities of @xmath573 come from a collision of the two vertex operators @xmath522 and @xmath533 . via the operator product ( [ mondox ] ) , such a collision gives a pole whose residue is a dilaton tadpole evaluated on the surface @xmath4 , endowed with one of its possible spin structures .
all possible spin structures occur because the fiber @xmath563 parametrizes among other things the generalized spin structure of the super riemann surface @xmath4 ( just as in genus 1 , the possible choices of this generalized spin structure are permuted when @xmath522 or @xmath533 is taken around a noncontractible loop in @xmath35 , and the generalized spin structure reduces to an ordinary one when the two operators meet ) .
the locus in @xmath560 along which the two vertex operators meet is a divisor @xmath574 that is a copy of @xmath575 ( fig .
[ splitting ] ) .
the projection @xmath557 which we use to formulate eqn .
( [ mono ] ) restricts along @xmath574 to the projection @xmath576 that dhoker and phong would use to compute the bulk contribution to the dilaton tadpole .
so if the singularities that correspond to dilaton tadpoles are the only ones that contribute in the identity ( [ mono ] ) , then we learn that with the dhoker - phong procedure , the sum over spin structures of the genus 2 dilaton tadpole vanishes .
( the full answer therefore comes from the correction at infinity that is described in @xcite . )
it remains to consider the possibility of additional singularities contributing to the identity ( [ mono ] ) , resulting from the fact that the super period matrix is only generically defined .
we will show in section [ lowgenus ] that although there is indeed a locus on @xmath563 on which the super period matrix has a pole ( with nilpotent residue ) , this leads to no contribution in the identity ( [ mono ] ) because the fermions of the matter system acquire zero - modes on just the dangerous locus .
in this section , we complete the argument of section [ twol ] , by analyzing the singularities associated to poles of the super period matrix , and showing that they do not affect the vanishing of the bulk contribution to the two - loop vacuum amplitude . as described in section [ badset ] , in the case of a genus 2 surface @xmath4 with ramond punctures @xmath448 , the condition @xmath577 is equivalent to @xmath578 .
moreover , this is also equivalent to @xmath579 . to make this condition more explicit , we recall that a genus 2 riemann surface @xmath11 is hyperelliptic ; it admits a holomorphic map @xmath580 that is a double covering , branched over 6 points .
@xmath11 has a @xmath3 symmetry group , generated by a symmetry @xmath581 that exchanges the two sheets of the covering .
concretely , @xmath11 can be described by a hyperelliptic equation @xmath582 where @xmath583 is a homogeneous polynomial of degree 6 in homogeneous coordinates @xmath358 of @xmath584 . in this description , @xmath585 is defined by forgetting @xmath546 and @xmath586 acts by @xmath587 . for @xmath588 a point in @xmath11 , we sometimes write @xmath589 for @xmath590 and say that @xmath588 and @xmath589 are conjugate . the condition that @xmath578 means that there is a holomorphic differential on @xmath11 whose zeroes are @xmath591 and @xmath472 .
a holomorphic differential on @xmath11 has the form @xmath592 , with constants @xmath593 .
it vanishes precisely when @xmath594 , a condition that defines a single point in @xmath584 , but a conjugate pair of points in @xmath11 . since
@xmath591 can be chosen arbitrarily and @xmath472 is then determined to be @xmath595 , the condition that @xmath591 and @xmath472 are conjugate defines a divisor @xmath257 in the space of all pairs @xmath448 . for the case
@xmath596 , this is the bad set in the definition of the super period matrix .
as explained in detail in appendix [ perturb ] , the super period matrix has a pole with nilpotent residue along @xmath257 .
this pole does not lead to additional contributions in the ward identity ( [ mono ] ) because when the ramond divisors are associated to conjugate points @xmath448 , the fermions of the matter system have many zero - modes that cancel the potential singularity due to the pole in the period matrix . to understand this
, we must recall some facts about the superstring spin field @xmath512 and its coupling to the matter fermions of the rns model .
first we consider the case of uncompactified ten - dimensional spacetime .
the rotation group is @xmath597 , and its maximal torus is @xmath598 .
the weights of the spinor representation take the form @xmath599 , where @xmath600 is the charge under the @xmath601 @xmath602 and each @xmath600 is @xmath603 . for a spinor of positive chirality ,
the number of weights @xmath604 is even ; we will assume that the spin field @xmath512 is a spinor of positive chirality .
the matter fermions of the heterotic string can be grouped as five complex fermions @xmath605 and their charge conjugates @xmath606 , all of which are holomorphic fields . @xmath605 and @xmath606 have charge 1 and @xmath607 , respectively for the @xmath601 @xmath602 in the maximal torus , and zero charge for the other @xmath602 s . the fields @xmath605 and @xmath606 appear in a dirac action @xmath608 , and this tells us that if @xmath605 is a section of a line bundle @xmath238 , then @xmath606 is a section of a conjugate line bundle @xmath609 .
suppose that these fermions interact with spin fields placed at points @xmath610 , and let the weights of the spin field at @xmath186 be @xmath611 , @xmath612 .
in the most traditional formulation , one says that as @xmath613 , @xmath614 has a half - order zero or pole at @xmath615 , depending on @xmath616 , with @xmath617 . in terms of complex geometry , this means roughly that @xmath605 is a section of @xmath618 but what precisely is meant by the half - integral powers @xmath619 ? if ( for some @xmath620 ) the @xmath616 are all @xmath621 , the meaning is simply that @xmath605 is a section of @xmath318 ( which has an isomorphism @xmath622 , so it is informally @xmath623 ) . in general , @xmath605 is a section of @xmath624 , and dually @xmath606 is a section of @xmath625 .
these statements incorporate the usual assertions about zeroes and poles of half - order , but in a way that is more natural in algebraic geometry . by including in @xmath626 a factor of @xmath627 whenever @xmath628
, we ensure that in this case ( in the more informal language ) , @xmath605 has a half - order zero at @xmath186 rather than a half - order pole . in our application , there are just two ramond insertions the operator @xmath522 at one point @xmath166 and the operator @xmath629 at another point @xmath588 .
( to reduce clutter , in the following discussion we write @xmath166 and @xmath588 rather than @xmath591 and @xmath472 for the points with ramond insertions . )
we first discuss the case of superstring theory in @xmath24 .
it is convenient to pick @xmath522 to be the particular spin field whose weights are all @xmath630 , and to pick @xmath533 to have all weights @xmath604 .
this choice ensures that the dilaton vertex operator @xmath631 does appear as a pole in the product @xmath632 .
acting with a matrix @xmath633 ( for any value of @xmath634 ) will always flip precisely one weight .
so ( looking at the definition ( [ olor ] ) of @xmath533 ) the component of @xmath533 with all weights @xmath604 is a sum of terms , each proportional to a spin field @xmath635 that has four weights @xmath604 and one weight @xmath621 .
as long as we are on @xmath24 , for counting fermion zero - modes , the different components are equivalent .
we may as well look at the term with @xmath636 and @xmath637 , @xmath638 . with those weights , @xmath639 is a section of @xmath318 . as we have seen above , the critical case that might give an unwanted contribution to the identity ( [ mono ] ) is that @xmath640 ( that is , @xmath588 is conjugate to @xmath166 ) , and moreover in this case , @xmath641 .
so in the critical case , @xmath639 is a section of @xmath642 and dually @xmath643 is a section of @xmath444 . on the other hand , @xmath605 for @xmath638
is a section of @xmath644 .
dually , @xmath606 for @xmath638 is a section of @xmath645 .
now we can count fermion zero modes .
the line bundle @xmath233 has a two - dimensional space of holomorphic sections , while @xmath266 , @xmath646 , and @xmath647 all have one - dimensional spaces of holomorphic sections .
so in the critical case @xmath640 , all @xmath605 and @xmath648 have 1 zero - mode , except that @xmath639 has 2 , making a total of 11 fermion zero - modes in all .
this is far too many zero modes for the locus @xmath640 to be dangerous . before discussing what happens to the identity ( [ mono ] ) ,
let us simply discuss how the correlation function @xmath556 , understood as a form on @xmath560 of codimension 1 , behaves for @xmath649 .
@xmath560 has odd dimension 3 , so in conventional language , to compute @xmath650 , one inserts three picture - changing operators on the worldsheet @xmath4 .
each picture - changing operator ( pco ) contains a factor of the worldsheet supercurrent , which ( for strings in @xmath24 ) is linear in the rns fermions @xmath605 and @xmath648 and so can absorb one zero - mode .
so the pco s remove 3 zero - modes , leaving 8 .
because of the 8 remaining zero - modes , @xmath556 vanishes on the divisor @xmath640 . to determine the order of its vanishing , we let @xmath651 be a local parameter on @xmath560 with a simple zero at @xmath640 . at @xmath652 , there are eight fermion zero - modes that are not lifted by the pco s .
all of these modes are lifted away from zero for @xmath653 and generically they are lifted to first order in @xmath651 .
so if we write @xmath654 for the relevant modes that become zero - modes at @xmath652 , the integral over those modes near @xmath652 looks like @xmath655 where @xmath543 is an antisymmetric form that generically is nondegenerate and @xmath656 is its pfaffian .
thus @xmath657 for @xmath658 .
the analysis of the identity ( [ mono ] ) is more tricky than this , because in defining @xmath566 , we remove a factor @xmath568 that has a pole at @xmath652 . as we explain in section [ poleffect ] , the effect of this is to cancel one power of @xmath651 , so that @xmath659 for @xmath658 . in particular
, @xmath566 has no singularity that would contribute to the identity ( [ mono ] ) .
now let us consider compactification to four dimensions .
the main difference is that the fields @xmath660 , @xmath648 , @xmath661 are not free fields on the worldsheet ; we can not talk about their zero - modes or use them in a simple way to predict a zero of an amplitude .
also , we need to slightly modify the definition of the dilatino vertex operator @xmath533 that has all weights @xmath604 ; it is now the sum of two terms , each proportional to a spin field @xmath536 whose weights @xmath600 are all @xmath604 except for either @xmath662 or @xmath663 ( this ensures that the part of @xmath533 in the internal compact manifold is actually a holomorphic primary field of the appropriate dimension ) . in the counting of zero - modes , we should consider only @xmath605 and @xmath606 for @xmath664 , and the number of such modes , counted the same way as before , is now 5 instead of 11 .
repeating the previous reasoning but with the smaller number of zero - modes , we find that @xmath556 is of order @xmath651 for @xmath658 and @xmath566 is of order 1 .
in particular , @xmath566 has no singularity at @xmath652 that could contribute in the identity ( [ mono ] ) .
hence in general , with the dhoker - phong procedure , the `` bulk '' contribution to the superstring measure vanishes in genus 2 after integrating over odd moduli and summing over spin structures , but before any bosonic integrations .
our goal here is to justify some of these statements by describing the behavior of @xmath566 along the divisor @xmath665 where the super period matrix has a pole .
it is helpful first to recall ( following @xcite , and using the language explained in section 3.6 of @xcite ) the origin of the picture - changing operators in worldsheet path integrals . for each odd variable @xmath666 , it is convenient to introduce also its differential @xmath667 .
then the insertion of a pco at a point @xmath668 comes from an integral @xmath669 where @xmath670 is the usual commuting antighost field of superstring perturbation theory and @xmath671 is the worldsheet supercurrent .
the integral over @xmath667 gives a factor @xmath672 , and naively the integral over @xmath666 gives a factor of @xmath671 .
the product is the pco @xmath673 . however , although this treatment of the integral over @xmath667 is correct , there is more subtlety in the integral over @xmath666 .
because of the lack of a natural separation between even and odd variables on supermoduli space , there are in general , depending on precisely how one parametrizes supermoduli space , additional @xmath666-dependent contributions hidden in other parts of the worldsheet path integral , beyond those written in ( [ zong ] )
. we can expand @xmath674 .
the @xmath675 term leads after integrating over @xmath666 to an insertion at @xmath166 of the usual pco @xmath673 , but the `` 1 '' term leads to insertion at @xmath166 of the operator @xmath672 , which we might call an incomplete pco .
locally , one can parametrize supermoduli space in such a way that the incomplete pco s can be ignored , but in general , globally either one can not do this or doing this introduces other complications .
so it is best to keep track of the contributions involving incomplete pco s .
the only general constraint is that the worldsheet path integral with insertion of an odd number of incomplete pco s vanishes because of fermi statistics , so one can always assume the number of incomplete pco s to be even .
now we consider our problem of the influence of the poles of the super period matrix on @xmath566 .
those poles will only affect the dependence of @xmath566 on holomorphic variables , and in what follows we may as well ignore the antiholomorphic variables .
@xmath560 is a smooth supermanifold ( or rather orbifold ) of dimension @xmath676 , say with even and odd coordinates @xmath677 .
@xmath556 is , from a holomorphic point of view , a smooth top form , so ignoring its dependence on antiholomorphic variables , it is @xmath678.\ ] ] the function @xmath679 can be expanded in powers of the @xmath680 .
when @xmath556 is computed in superconformal field theory , there is a contribution , proportional to @xmath681 , that can be evaluated using complete pco s only . as there are 3 odd moduli , this contribution involves 3 insertions of @xmath671 .
these insertions can absorb 3 of the 11 fermion zero - modes ( for strings in @xmath24 ) described in section [ fzero ] , leading to a contribution to @xmath155 that , as explained there , is of order @xmath682 near @xmath652 .
@xmath556 also has a contribution that is linear in the @xmath680 , arising from a contribution evaluated with 2 incomplete pco s and only one insertion of @xmath671 .
this insertion lifts only one fermion zero - mode , giving a contribution to @xmath556 that is of order @xmath683
. we can thus rewrite ( [ tombo ] ) in the more detailed form @xmath684,\ ] ] with @xmath685 naively , the term in ( [ wombo ] ) that is linear in the @xmath666 s is irrelevant both because it vanishes more rapidly for @xmath658 and because , with some of the @xmath666 s missing , one might think that this term would not contribute in the integration over odd moduli that goes into evaluating the boundary contributions in eqn .
( [ mono ] ) . however , this reasoning is not valid , for reasons that are related in part to some of the usual subtleties of superstring perturbation theory . to define @xmath566 ,
we are supposed to split off from @xmath556 the pullback of a volume form on @xmath23 .
concretely , let @xmath686 be the three independent matrix elements of the pseudoclassical block of the super period matrix .
then we can define @xmath566 by @xmath687\cdot f^*_{\s_a\v^b},\ ] ] where @xmath566 is a relative volume form along the fibers of @xmath557 .
the reason that pulling out the factor of @xmath688 $ ] changes the order of vanishing at @xmath652 is that the @xmath689 have poles at @xmath652 .
with only three odd variables @xmath680 , since the @xmath689 are even , the residue of such a pole can only be a bilinear expression in the @xmath680 , and the products of the residues of the poles in different @xmath689 would vanish .
accordingly , we lose nothing essential if we assume that only @xmath690 has a pole , and that its residue is proportional to @xmath691 .
we can pick coordinates so that when the @xmath680 vanish , the @xmath689 coincide with @xmath692 , @xmath693 .
then the general form of the pole is @xmath694 for some function @xmath0 .
so we see that @xmath695 where we write only terms proportional to @xmath696 on the right hand side ; we have also dropped a term @xmath697 , as it is less singular than the @xmath698 term that we have kept . if in defining @xmath566 in eqn .
( [ mofto ] ) , we were to split off a factor of @xmath699 $ ] , then @xmath566 would have the same behavior for @xmath658 as @xmath556 .
however , instead we are supposed to split off a factor of @xmath688\sim [ \d h_1\d
h_2\d h_3](1-\eta_1\eta_2 g\partial_{h_1}u / u^2+\dots)$ ] , and this gives @xmath700,\ ] ] where we omit some terms that do not affect the argument .
since @xmath701 , it follows that @xmath566 has a contribution proportional to @xmath702 .
since this term is proportional to the product of all three @xmath666 s , it can contribute to the surface integrals on the right hand side of eqn .
( [ mono ] ) .
the effect of the poles of the super period matrix together with the use of incomplete pco s has been to reduce by 1 the expected order of vanishing of @xmath566 near @xmath652 : it vanishes as @xmath703 rather than @xmath682 .
with more odd variables , we can repeat this process . for every pair of odd variables , the order of vanishing of @xmath566 along @xmath652 is reduced by 1 ( or the order of a pole
is increased by 1 ) compared to what one would expect from counting fermion zero - modes while ignoring the pole of the super period matrix .
much less is known about superstring scattering amplitudes for @xmath18 .
however , for @xmath489 , one can still define a meromorphic projection @xmath704 using the super period matrix .
this map behaves badly along hyperelliptic divisors in @xmath705 that are described in appendix [ hyperelliptic ] , and this may ultimately prevent it from being very useful . in the following , we simply avoid these issues by assuming that @xmath11 ( which is kept fixed in the whole analysis ) is not hyperelliptic .
the natural superstring measure on @xmath706 can be pushed forward to a measure on @xmath705 by integrating over the fibers of @xmath19 . here
we will show , generalizing the above arguments for @xmath17 , that in the case of superstring theory in @xmath24 , the resulting measure on @xmath705 vanishes when summed over spin structures , without any integration over bosonic moduli . for @xmath489 , the condition on a pair of ramond insertions such that @xmath577 was determined in section [ badset ] .
in particular , this condition defines a divisor @xmath257 in the space of all pairs @xmath707 .
it actually turns out that the counting of fermion zero - modes can be done without using explicit knowledge of @xmath257 . for strings in @xmath24 , with the same configuration of spin fields as in section [ specialcase ] , @xmath639 is a section of @xmath318 , @xmath643 is a section of @xmath443 , @xmath605 for @xmath638 is a section of @xmath708 , and @xmath606 for @xmath638 is a section of @xmath709
. we have to count the number of zero - modes of these fermion fields .
the riemann - roch theorem and serre duality tell us that @xmath710 . along the divisor @xmath574
, @xmath711 is at least 2 , so @xmath712 is at least 1 .
hence along @xmath574 , the fields @xmath713 and @xmath714 have together at least @xmath715 zero - modes . for the fields @xmath716 @xmath638 , we have to replace @xmath318 by @xmath708 .
this replacement imposes 1 condition on a holomorphic section of @xmath318 ( it must vanish at @xmath588 ) , so it reduces the number of zero - modes by at most 1 ( and generically by 1 ) .
so along @xmath574 , the number of @xmath605 zero - modes for @xmath638 is always at least 1 . for @xmath318 of degree @xmath489 , so that @xmath708 has degree 2 , the riemann - roch theorem and serre duality imply that @xmath717 , or @xmath718 , so that @xmath719 , which is the number of @xmath648 zero modes for @xmath720 , is also at least 1 . in sum , along @xmath574 , we always have for @xmath489 at least the same 11 fermion zero - modes that we found for @xmath17 ( and generically the number is precisely 11 )
. we can then use the same reasoning as in section [ specialcase ] to analyze the behavior of @xmath556 and @xmath566 along @xmath574 .
the only difference is that the number of odd moduli of @xmath721 is 5 instead of 3 .
so to compute @xmath556 , we would use 5 pco s instead of 3 , with the result that @xmath556 has a zero of order 3 along @xmath574 , rather than the zero of order 4 that we found for genus 2 . in going from @xmath556 to @xmath566
, we can now lose two orders of vanishing ( one for each pair of incomplete pco s ) , so @xmath566 has a simple zero along @xmath574 . as before ,
supersymmetric compactification to four dimensions reduces the number of zero - modes from 11 to 5 .
but now , with 5 pco s , @xmath556 need not vanish along @xmath574 , and @xmath566 may have a double pole .
so it appears that in a general supersymmetric compactification to four dimensions , the bulk contribution to the genus 3 vacuum amplitude , defined using the projection @xmath722 derived from the super period matrix , does not necessarily vanish pointwise after summing over spin structures . for any @xmath0 , with precisely 2 ramond punctures , riemann - roch and serre duality can be used in the same way just described , to show that if @xmath441 , then the rns fermions associated to strings in @xmath24 have at least 11 zero - modes .
but for @xmath480 , this can not be used in any obvious way to study the superstring measure .
the most basic problem is that for @xmath480 , the super period matrix generically does not obey the schottky relations that are satisfied by the period matrix of an ordinary riemann surface , and consequently can not be used to define a projection @xmath723 . also , as @xmath0 increases , the odd dimension of @xmath34 increases . for @xmath724 ,
even if we did have a holomorphic projection @xmath723 with the properties that we have exploited , the above reasoning would allow a pole of @xmath566 along @xmath574 , leading to no simple conclusion from the identity ( [ mono ] ) .
for the case of two ramond punctures , that is @xmath434 , we have found in section [ badset ] that the exceptional set along which @xmath441 is of codimension 1 , and thus defines a divisor @xmath257 in the reduced space of @xmath34 . in section [ poleffect ] , it was important to know how the super period matrix @xmath226 behaves along @xmath257 .
we claim that the pseudoclassical block @xmath231 of the super period matrix has a pole along @xmath257 with nilpotent residue . to be more precise
, the claim is that if @xmath651 is a local parameter on @xmath725 with a simple zero along @xmath257 , and @xmath726 are the odd moduli , then @xmath231 has an expansion in which the leading term at @xmath727 is the classical period matrix @xmath493 and the corrections are a series in @xmath728 : @xmath729 here @xmath730 are functions on @xmath257 , and less singular terms are omitted . to deduce this formula
, we simply use the dhoker - phong expansion of the super period matrix as a function of odd variables @xcite , slightly adapted to take into account the presence of ramond punctures . for each pair of odd variables ,
the dhoker - phong formula contains a fermion propagator ( called @xmath731 below ) that behaves as @xmath732 because of the presence of zero - modes at @xmath652 .
this accounts for the form of the expansion in eqn .
( [ holfo ] ) .
we will assume that the reader is familiar with the derivation of the dhoker - phong expansion given in section 8.3 of @xcite .
we will essentially repeat this derivation , with some minor modifications to account for ramond punctures . starting with a split super riemann surface @xmath4
, we want to give a smooth model for its deformations associated to odd moduli . in the absence of ramond punctures ,
this is done as follows ( eqn .
( 8.17 ) of @xcite ) .
locally @xmath4 can be described by holomorphic superconformal coordinates @xmath46 and a local antiholomorphic coordinate @xmath733 ; in the case that @xmath4 is split , we can take @xmath733 to be the complex conjugate of @xmath545 .
a holomorphic function on @xmath4 is a function annihilated by @xmath734 . to deform the complex structure of @xmath4
, we replace @xmath734 by , where @xmath735 is a @xmath736-form on @xmath11 valued in @xmath234 .
this is not necessary for extracting the singular behavior claimed in eqn .
( [ holfo ] ) . ] @xmath737 where @xmath738 is a @xmath736-form on @xmath11 valued in @xmath236 .
( @xmath234 and @xmath233 will denote the holomorphic tangent and cotangent bundles of @xmath739 ) the expression @xmath740 is an odd @xmath736-form on @xmath35 valued in odd superconformal vector fields .
we need to find an analogous construction in the presence of ramond punctures .
the first step is to reinterpret @xmath738 as a @xmath736-form on @xmath11 that is valued in @xmath275 , which we characterize as in eqn .
( [ elmonk ] ) as a line bundle on @xmath11 with an isomorphism @xmath741 , where @xmath277 are the locations of ramond punctures .
equivalently , there is an isomorphism @xmath742 .
yet another equivalent statement is that there is a homomorphism @xmath743 which maps a section @xmath40 of @xmath275 to a section @xmath744 of @xmath745 that vanishes at @xmath277 .
if @xmath40 is a section of @xmath746 , then @xmath747 is an odd superconformal vector field on @xmath4 .
( if the superconformal structure of @xmath4 is defined locally by @xmath168 , then @xmath748 . ) with @xmath40 replaced by a @xmath736-form @xmath738 valued in @xmath275 , we get the appropriate generalization of eqn .
( [ mortz ] ) in the presence of ramond punctures : @xmath749 we expand @xmath738 in a basis @xmath750 of @xmath751 , with coefficients the odd moduli @xmath752 : @xmath753 in this way , the odd parameters @xmath754 @xmath755 are used to deform the complex structure of @xmath4 . with this description of the perturbation that we are trying to make , it is actually straightforward to repeat the derivation in section 8.3 of @xcite .
one basically just needs to replace @xmath738 in some places by @xmath756 .
a holomorphic 1-form @xmath242 on @xmath16 corresponds to a section @xmath757 $ ] of @xmath260
. if @xmath4 is split , the pseudoclassical block of its super period matrix coincides with the classical period matrix of @xmath11 . to compute the dependence of the super period matrix on odd moduli
, we have to analyze how a section @xmath758 of @xmath260 changes when the odd moduli are turned on . by analogy with eqn .
( 8.19 ) of @xcite , a general section of @xmath258 in the presence of the perturbation is @xmath759.\ ] ] if @xmath760 is understood as an integral form , then as in eqn .
( 8.22 ) of @xcite , the condition for @xmath760 to be holomorphic is @xmath761 , where @xmath762 we expand @xmath763 for @xmath764 to be a section of @xmath258 , @xmath765 should be a section of @xmath766 , and @xmath767 should be a section of @xmath233 .
the condition @xmath768 becomes a pair of equations , generalizing eqn .
( 8.23 ) of @xcite : @xmath769 define an ordinary 1-form on @xmath11 : @xmath770 the second equation in ( [ zebbo ] ) says that @xmath771 .
the @xmath6- and @xmath78-periods of the closed holomorphic 1-form @xmath65 on @xmath4 that corresponds to @xmath760 are simply the ordinary @xmath6- and @xmath78-periods of the ordinary 1-form @xmath772 .
( the proof of this statement can be found in section 8.3 of @xcite and is unaffected by the existence of ramond punctures . )
so to compute the super period matrix of @xmath4 as a function of odd moduli , we simply have to compute the periods of @xmath772 .
to do this , we have to solve the equations ( [ zebbo ] ) as a function of the odd parameters @xmath752 . on considering the first of these equations
, we immediately run into a problem .
generically , this equation has no solution .
the obstruction lies in @xmath773 , which generically ( in the presence of @xmath1 ramond punctures ) is of dimension @xmath276 .
this obstruction reflects something that was explained in appendix d.1 of @xcite , and that was important in section [ counting ] above .
in the presence of ramond punctures , closed holomorphic 1-forms on @xmath35 correspond to sections not of @xmath260 , but of @xmath300 , where @xmath300 is the sheaf whose sections are sections of @xmath260 that may have poles , with zero residue , along a ramond divisor . for our purposes , this means that in solving eqn .
( [ zebbo ] ) , we should allow @xmath774 to have simple poles at @xmath277 .
thus , it is a section not of @xmath443 but of @xmath775 . with @xmath774
understood in this way , the first of eqns .
( [ zebbo ] ) can be solved , but the solution is not unique : it is unique only modulo the possibility of adding to @xmath774 an element of @xmath324 , which generically is of dimension @xmath276 .
this non - uniqueness was to be expected ; it reflects the fact that @xmath776 has both odd and even holomorphic sections , with the odd sections having the form @xmath777 $ ] for @xmath778 .
when we deform a section @xmath758 of @xmath300 as a function of the odd moduli @xmath752 , we are free to add to @xmath758 a linear combination of the odd sections with @xmath752-dependent coefficients .
the procedure in defining the super period matrix is to consider closed holomorphic 1-forms , or equivalently sections of @xmath776 , that have specified @xmath6-periods .
we start at @xmath779 with holomorphic 1-forms @xmath780 such that , on the ordinary riemann surface @xmath11 , @xmath781 .
the section @xmath782 $ ] of @xmath776 then automatically has vanishing fermionic @xmath6- and @xmath78-periods .
we deform @xmath783 as a function of the odd moduli @xmath752 to get a section @xmath784 of @xmath776 such that @xmath785 , and @xmath784 has vanishing fermionic @xmath6-periods . as in the classical theory
, it is not possible to constrain the bosonic or fermionic @xmath78-periods of @xmath786 ; these make up the @xmath5 and @xmath787 blocks of the super period matrix .
in particular , the vanishing of the fermionic @xmath6-periods of @xmath784 will make the solution for @xmath774 unique .
this condition means that @xmath774 is not an arbitrary section of @xmath318 but ( in the notation of eqn .
( [ melz ] ) ) obeys @xmath788 away from a bad locus in the reduced space of @xmath412 on which the super period matrix develops a singularity , there is no holomorphic section of @xmath318 that satisfies ( [ refo ] ) , but there is such a section if we allow a pole at some point @xmath789 .
this means that there is a unique solution @xmath731 of the equation @xmath790 ( the delta function is defined by @xmath791 , where @xmath792 . )
@xmath71 plays the role in the presence of ramond punctures that the ordinary dirac propagator plays in their absence .
we will describe it more precisely momentarily .
we can express @xmath774 in terms of @xmath767 : @xmath793 and hence we can write an equation for @xmath767 only : @xmath794 however , we should explain better what sort of geometric object is @xmath731 .
in its dependence on @xmath545 , it is a section of @xmath318 that is constrained by eqn .
( [ refo ] ) , but what about its dependence on @xmath795 ?
since the answer is a little tricky and involves a slightly exotic use of duality , we first give an example . we take @xmath4 to be the complex @xmath545-plane with ramond divisors at @xmath56 and @xmath347 , as in eqn .
( [ zeldo ] ) .
if as in that discussion we write @xmath70 for a section of @xmath318 that has fermionic periods 1 and @xmath355 at @xmath56 and @xmath347 , then @xmath796 ( the symbol @xmath797 just denotes a tensor product of forms on two different factors of @xmath4 , one parametrized by @xmath46 and one by @xmath798 . )
the point of this formula is that the function @xmath799 equals @xmath800 or @xmath801 at @xmath56 or @xmath347 , and hence @xmath731 has fermionic periods @xmath800 and @xmath802 at @xmath56 and @xmath347 .
the relative minus sign means that while @xmath70 has a vanishing fermionic @xmath78-period , @xmath731 has a vanishing fermionic @xmath6-period in its dependence on @xmath545 , as desired .
we note from eqn .
( [ monkey ] ) that @xmath731 is odd under @xmath803 , and in particular that it is the same sort of geometric object in each variable .
this is what we want to explain in general .
we will use a physical language . if we are given a pair of fermi fields @xmath804 that are sections of serre dual line bundles @xmath238 and @xmath609 , where @xmath238 has degree @xmath468 and @xmath805
, then we would have a dirac action @xmath806 , with a fermion propagator @xmath731 obeying eqn .
( [ morfo ] ) .
it would be a section of @xmath807 ( with a simple pole on the diagonal of residue 1 ) . in our case , @xmath731 as a function of @xmath545 is a section of the line bundle @xmath318 of degree @xmath293 , but it obeys the @xmath276 constraints ( [ refo ] ) .
to write an action describing this situation , we take @xmath808 and @xmath809 to be sections of @xmath318 and @xmath443 , respectively , but with lagrange multipliers that enforce the constraints : @xmath810 varying with respect to @xmath808 , the equation of motion for @xmath809 is @xmath811 thus , @xmath809 has poles at @xmath277 , and we can view it as a section of @xmath812 .
but the @xmath1 poles are not independent ; their residues are determined by the @xmath276 lagrange multipliers @xmath813 .
the resulting relations between the residues mean precisely that as a section of @xmath318 , @xmath809 obeys the constraints ( [ refo ] ) and thus that it is the same sort of geometric object as @xmath808 .
thus we can view @xmath731 in eqn .
( [ morfo ] ) as a section of @xmath814 that obeys the conditions ( [ refo ] ) in each variable and has a simple pole on the diagonal .
@xmath815 obeys all of the same conditions , but with an opposite residue for the pole on the diagonal , so @xmath816 this is important in the symmetry of the formula ( [ golly ] ) below for the first correction to the super period matrix
. however , if we want to view @xmath731 as a section of @xmath817 rather than of @xmath818 , we have to replace @xmath819 in eqns .
( [ ubeklin ] ) and ( [ ueklin ] ) with @xmath820 , to give @xmath821 and @xmath822 starting with a choice of @xmath767 at @xmath779 , eqn .
( [ eklin ] ) can be solved iteratively to determine @xmath767 as a function of the @xmath752 . to make the solution unique
, we need a condition on the @xmath6-periods of the ordinary 1-form @xmath772 . to find the right condition , recall that to compute the super period matrix , we start with holomorphic 1-forms @xmath780 on @xmath11 with canonical @xmath6-periods , @xmath823 . we promote them to @xmath752-dependent holomorphic sections @xmath784 on @xmath4 , such that , with @xmath824 defined by eqns .
( [ morx ] ) , ( [ durfo ] ) and ( [ ebbo ] ) , @xmath825 this condition on bosonic @xmath6-periods makes the solution of eqn .
( [ eklin ] ) unique . the final computation of the super period matrix @xmath226 proceeds rather as in section 8.3 of @xcite .
the @xmath5 block of @xmath226 is defined as @xmath826 the difference between @xmath250 and the classical period matrix @xmath827 is @xmath828 where @xmath829 .
riemann s bilinear relations say that if @xmath830 are closed 1-forms on the ordinary riemann surface @xmath11 , then @xmath831 taking @xmath832 , @xmath833 , and remembering that @xmath834 is of type @xmath835 on the ordinary riemann surface @xmath11 , so that the @xmath835 part of @xmath836 does not contribute to @xmath837 , we learn that @xmath838 in turn , we can use ( [ beklin ] ) to eliminate @xmath839 : @xmath840 when @xmath841 is expanded in powers of the @xmath752 s , the lowest order term , which we call @xmath842 , is quadratic .
it can found by just replacing @xmath843 in the last formula by @xmath844 and writing @xmath845 : @xmath846 this is the analog of the dhoker - phong formula in the absence of ramond punctures .
higher order terms can be evaluated by using ( [ eklin ] ) to express @xmath847 as a polynomial in the @xmath666 s .
now we can explain the form of the expansion in eqn .
( [ holfo ] ) .
the kernel @xmath731 develops a pole as a function of the moduli of @xmath4 when the @xmath734 operator , acting on sections of @xmath318 that satisfy eqn .
( [ refo ] ) , has a non - trivial kernel .
this is the bad set discussed in section [ badset ] . if @xmath651 is a local parameter with a simple zero on the bad set , then @xmath848 near @xmath652 . when @xmath849 is evaluated by solving eqn .
( [ eklin ] ) iteratively for @xmath850 and substituting in eqn .
( [ delfom ] ) , each pair of odd moduli is accompanied by a factor of @xmath731 .
this leads to the behavior claimed in eqn .
( [ holfo ] ) .
the meaning of the poles at @xmath652 is , however , quite different for @xmath434 or @xmath479 . for @xmath434 ,
the choice of fermionic @xmath6-periods is essentially unique , as explained in section [ tworp ] ; the condition @xmath424 defines a divisor @xmath257 ; and the poles of @xmath851 along this divisor do not depend on any arbitrary choices .
by contrast , for @xmath479 , the definition of @xmath851 does depend on a somewhat arbitrary choice of fermionic @xmath6-periods , and @xmath851 develops poles along a divisor on which this choice breaks down .
these are the poles described in the above formula .
the condition @xmath424 for @xmath479 defines a set of complex codimension greater than 1 , and is not the locus of poles in @xmath851 .
as explained in the introduction , in general the procedure of dhoker and phong to compute the two - loop vacuum amplitude must be supplemented by a correction at infinity . the explanation given in @xcite for the need for this correction
was the following .
the correction is associated to the splitting of a genus 2 super riemann surface @xmath4 with even spin structure to two genus 1 components each with even spin structure .
so consider a partial compactification does have a pole at infinity if one tries to extend it over the full deligne - mumford compactification of @xmath14 .
the pole arises on the degeneration in which @xmath4 decomposes to two genus 1 components each with odd spin structure . because of fermion zero - modes
, this pole in @xmath19 does not lead to any correction to the dhoker - phong procedure for analyzing the vacuum amplitude . ]
in which one allows this type of degeneration , thus adding a divisor @xmath852 to @xmath14 .
although the holomorphic splitting @xmath853 used by dhoker and phong does extend to a holomorphic map between the corresponding partially compactified spaces , the extended map does not restrict to a splitting of the divisor @xmath852 .
this leads to a slight mismatch between what one gets by first integrating over the fibers of @xmath19 and the general formalism of superstring perturbation theory
. we will here give an alternative explanation in the language of picture - changing operators rather than super riemann surface theory .
in appendix [ genone ] , we recall some elementary facts about riemann surfaces of genus 1
. then in appendix [ dph ] , we recall what the dhoker - phong procedure means when expressed in terms of picture - changing operators ( pco s ) and analyze how this procedure behaves near the relevant separating degeneration .
let @xmath854 be a riemann surface of genus 1 . picking a point @xmath855 as the `` origin , ''
@xmath854 becomes an elliptic curve .
although the canonical bundle @xmath233 of @xmath854 is trivial , a square root @xmath237 of @xmath233 may not be trivial .
an even spin structure on @xmath854 corresponds to the case that @xmath237 is a non - trivial line bundle of order 2 , namely @xmath856 , where @xmath588 is one of the three non - zero points of order 2 on @xmath854 .
the dirac propagator @xmath857 is a section of @xmath858 with a simple pole of residue 1 on the diagonal .
we will need to understand the propagator for the case that @xmath859 .
@xmath860 is , as a function of @xmath545 , a nonvanishing section of @xmath856 that has a simple pole at @xmath166 ( and no other singularities ) .
differently put , it is a nonzero section of @xmath861 . but
a nonzero section of @xmath861 vanishes at @xmath588 and nowhere else .
thus , @xmath860 vanishes precisely at @xmath862 .
suppose that @xmath854 is the reduced space of a super riemann surface @xmath35 with one ns puncture .
in the picture - changing formalism , we represent an ns puncture on @xmath35 as a point in @xmath863 , which because of the translation symmetries of @xmath854 we may as well take to be @xmath166 , and we represent the odd modulus of @xmath35 by including a pco at some point @xmath864 .
the only constraint on @xmath651 is that to avoid a `` spurious singularity , '' we want @xmath865 .
( here @xmath866 is the sheaf of odd superconformal vector fields that vanish at @xmath166 , and the condition to avoid a spurious singularity is that this sheaf has no section whose only singularity is a simple pole at the position @xmath651 of the pco . )
but @xmath867 so we want @xmath868 .
this is true if and only if @xmath869 . in other words
, we may place the pco anywhere except at @xmath588 without running into a spurious singularity .
now let @xmath870 be a riemann surface of genus 2 with even spin structure .
we view it as the reduced space of a genus 2 super riemann surface @xmath35 .
a family of @xmath35 s depending on a full set of odd parameters
namey @xmath871 of them can be constructed by inserting pco s at two points @xmath872 . in this language
, the dhoker - phong procedure amounts to the following : one should choose the points @xmath358 so that the dirac propagator connecting them vanishes : @xmath873 .
( to obey @xmath873 , we may pick any @xmath651 and for given @xmath651 there are then two choices of @xmath874 . ) the dhoker - phong formula for the dependence of the super period matrix on odd moduli ( see @xcite and also @xcite , section 8.3 , or eqn .
( [ golly ] ) above ) shows that we get a family of super riemann surfaces whose super period matrix does not depend on the odd moduli if we include odd moduli by placing pco s at any points @xmath358 satisying @xmath873 , and avoiding spurious singularities .
( generic points satisfying @xmath873 do avoid spurious singularities . )
this means in particular that as long as we require @xmath873 and avoid spurious singularities , the pco formalism will give an answer that does not depend on the specific choice of @xmath651 and @xmath874 .
this is the dhoker - phong procedure expressed in terms of pco s .
following this procedure will certainly give a unique , globally - defined answer .
there can not be any global obstruction to following this procedure , because @xmath651 and @xmath874 do not have to vary continuously ; we can use different pairs @xmath358 in different regions of moduli space . now consider what happens when @xmath870 degenerates to a pair of genus 1 components @xmath854 and @xmath875 , each with even spin structure .
@xmath870 is constructed by gluing a point @xmath876 to a point @xmath877 ( fig .
[ twosplit ] ) .
the spin bundle of @xmath854 is @xmath878 for some @xmath879 , and the spin bundle of @xmath875 is @xmath880 for some @xmath881 . the general formalism of superstring perturbation theory , when expressed in terms of pco s , tells us that in the limit that @xmath870 degenerates as described in the last paragraph , we must ensure that the pco s are in opposite branches , say @xmath882 , @xmath883 .
( for example , see section 6.3.6 of @xcite . )
moreover , to avoid a spurious singularity , we need @xmath869 and @xmath884 . to @xmath885 .
pco s are inserted at @xmath882 and @xmath883.,width=288 ]
we will now see that there is a conflict between the dhoker - phong procedure and the conditions stated in the last paragraph .
that is why the dhoker - phong procedure requires a correction at infinity .
if @xmath870 is a singular surface made by gluing together @xmath854 and @xmath875 , then the dirac propagator @xmath886 is identically 0 for @xmath887 , @xmath883 .
however , we really want to know what happens if we impose the dhoker - phong condition @xmath873 away from the degeneration limit , and then let @xmath870 degenerate . if we deform a singular surface @xmath870 made by gluing at @xmath166 and @xmath888 into a smooth surface with a small deformation parameter @xmath889 , then to lowest order in @xmath889 , the propagator @xmath886 with @xmath651 and @xmath874 on opposite branches is @xmath890 ( for example , see eqn .
( 3.49 ) of @xcite . )
as long as @xmath870 is smooth , there is no difficulty in ensuring that @xmath873 and avoiding spurious singularities .
now let us consider when happens as @xmath891 . trying to avoid a spurious singularitiy , we take @xmath651 to not approach the point @xmath892 as @xmath891 .
but then as we know from appendix [ genone ] , @xmath893 for @xmath891 .
( [ dolfo ] ) then implies that to make @xmath886 identically 0 for all @xmath889 , we will have to have @xmath894 for @xmath891 in order to make @xmath895 vanish .
( otherwise , @xmath886 is of order @xmath889 for small @xmath889 and can not vanish identically . ) but this means that for @xmath891 , we will meet a spurious singularity after all . in short ,
if we follow the dhoker - phong procedure , then the divisor in which @xmath4 splits to a pair of genus 1 components with even spin structure is a locus of spurious singularities .
that is why in general this procedure needs to be supplemented by adding a correction term supported on this divisor . in this analysis ,
since we were not able to avoid a spurious singularity at infinity , we did not gain much by placing @xmath651 and @xmath874 on opposite branches of @xmath870 in the degeneration limit .
we could just as well have chosen @xmath651 and @xmath874 to be both in , say , @xmath854 , in the limit that @xmath870 degenerates .
for this , we could take @xmath651 to be a generic point in @xmath854 and then pick @xmath896 such that @xmath873 .
the correction at infinity can be computed by comparing this procedure to the general formalism of superstring perturbation theory .
this can possibly be the basis for justifying the procedure of @xcite .
a hyperelliptic riemann surface @xmath897 of genus 3 can be described in affine coordinates by an equation @xmath898 thus @xmath897 is a double cover of @xmath584 ( parametrized by @xmath899 with a point at infinity added ) with 8 branch points @xmath900 .
the hyperelliptic involution acts by @xmath901 , and acts as @xmath607 on the three linearly independent holomorphic differentials @xmath902 dually , @xmath586 acts as @xmath607 on @xmath903 and hence on all @xmath6- and @xmath78-periods .
accordingly the @xmath904 period matrix is even under @xmath586 .
modulo the action of @xmath905 on @xmath584 , this family of genus 3 hyperelliptic curves depends on 5 complex parameters ( the action of @xmath905 can be used to fix 3 of the 8 branch points @xmath906 ) .
these 5 moduli are clearly @xmath586-invariant .
a genus 3 riemann surface has altogether @xmath907 moduli , so there is 1 additional modulus @xmath908 that is odd under @xmath586 .
we can confirm this by examining the quadratic differentials , which are dual to the infinitesimal deformations of @xmath897 .
there are 5 even ones @xmath909 and 1 odd one @xmath910 a genus 3 riemann surface is parametrized locally by its @xmath490 symmetric period matrix @xmath827 .
since hyperelliptic curves of genus 3 are a five - parameter family , they span a codimension 1 subspace of period matrices .
let @xmath588 be a holomorphic function of the matrix elements @xmath827 with a simple zero along the subspace that parametrizes hyperelliptic riemann surfaces .
since all matrix elements of @xmath827 are @xmath586-invariant , @xmath588 in particular is an even function of @xmath908 , and we can normalize @xmath588 and/or @xmath908 so that the relation between them is simply @xmath911 now suppose that @xmath897 is the reduced space @xmath11 of a split super riemann surface @xmath4 , and in particular is endowed with a spin structure .
a genus 3 surface has @xmath912 odd spin structures and @xmath913 even ones .
a spin structure is a line bundle @xmath238 with an isomorphism @xmath914 .
such a line bundle can be characterized by specifying @xmath915 for some meromorphic section @xmath40 of @xmath238 . here
@xmath915 will have precisely the same zeroes and poles as @xmath40 , but with twice the multiplicity , so in particular it has zeroes and poles of even multiplicity only
. often we write @xmath237 for @xmath238 .
for example , to define an odd spin structure , we pick a pair of branch points @xmath906 , @xmath916 , and characterize @xmath238 by saying that it has a section @xmath40 with @xmath917 the differential @xmath918 has double zeroes at @xmath919 and at @xmath920 , and no other zeroes or poles .
so @xmath40 has simple zeroes at those two points and no poles .
in particular , @xmath40 is holomorphic .
we can describe this more intuitively by writing @xmath921 up to a constant multiple , @xmath40 is the only holomorphic section of @xmath238 .
so @xmath922 and @xmath238 is an odd spin structure .
the 28 odd spin structures on @xmath11 are all obtained by this construction , with the @xmath923 possible choices of the pair @xmath924 .
obviously , these 28 choices are permuted transitively by permutations of the branch points @xmath906 .
so in the moduli space @xmath925 that parametrizes a genus 3 surface with an odd spin structure , the hyperelliptic surfaces form an irreducible divisor @xmath926 .
by contrast , there are two essentially different types of even spin structure on @xmath11 , corresponding to whether @xmath927 equals 0 or 2 .
there is only 1 spin structure with @xmath928 .
the space @xmath929 is spanned by sections @xmath40 , @xmath930 that obey @xmath931 , @xmath932 ( @xmath40 has simple zeroes at two points in @xmath11 lying above @xmath933 and @xmath930 has simple zeroes at two points lying above @xmath934 ) .
more informally , we write @xmath935 since there is just one spin structure with @xmath928 , genus 3 surfaces with such a spin structure are parametrized by an irreducible divisor @xmath936 .
an example of an even spin structure with @xmath805 is given by assuming that @xmath238 has a meromorphic section @xmath40 with @xmath937 informally , @xmath938 such an @xmath238 has no holomorphic section .
how many choices are there of such @xmath238 s ? we get an isomorphic line bundle if we exchange @xmath939 with , say , @xmath940 , since this can be accomplished by replacing @xmath40 by @xmath941 : @xmath942 likewise , we do not get anything new if we replace @xmath943 by @xmath944 , since this exchange is equivalent to replacing @xmath40 by @xmath945 .
so the choice of @xmath238 depends only on the partition of the set @xmath946 as the union of two subsets @xmath947 and @xmath948 , each with 4 elements .
there are 35 such partitions , and thus 35 even spin structures with @xmath805 .
they are obviously permuted transitively by permutations of the @xmath906 .
so genus 3 surfaces with an even spin structure of this type are parametrized by an irreducible divisor @xmath949 . as described in section [ twop ] , mapping a super riemann surface to the riemann surface with the same period matrix
gives a meromorphic projection @xmath491 .
it is fairly obvious that @xmath19 has a pole along @xmath257 , since the super period matrix has a pole there .
it is less obvious that @xmath19 has a pole along @xmath492 ; this was essentially explained to the author by p. deligne . in the remainder of this appendix
, we have two goals .
the first is to describe the origin of the pole of @xmath19 along @xmath492 .
the second goal involves the following application to superstring perturbation theory .
let @xmath20 be the holomorphic measure on @xmath706 determined by superstring theory on @xmath24 .
integration over odd moduli of @xmath4 keeping fixing its super period matrix @xmath226 generates a natural measure @xmath21 on @xmath705 .
our second goal is to describe the behavior of @xmath21 along @xmath257 and along @xmath492 .
we start by considering @xmath257 .
we will need to understand how the odd moduli of @xmath4 transform under the hyperelliptic `` involution '' @xmath586 .
we have put the words `` involution '' in quotes because in acting on fermions , @xmath950 , not @xmath951 .
this is clear from the action of @xmath586 on @xmath952 .
since @xmath586 acts by @xmath953 , it multiplies @xmath954 by @xmath955 , where the choice of sign is arbitrary : it is up to us with which sign we want to take @xmath586 to act on the spin bundle .
let us make a choice of @xmath355 and declare that @xmath956 .
having made this choice , it is now meaningful to ask how @xmath586 acts on the odd moduli of @xmath4 .
it is convenient to write @xmath237 for what we have called @xmath238 in appendix [ spinstr ] and @xmath236 for @xmath957 .
we want to know how @xmath586 acts on @xmath958 , which parametrizes the odd moduli along the split locus .
it is slightly more convenient to determine the action of @xmath586 on the dual space @xmath959 . along @xmath257 ,
the four sections of @xmath960 are @xmath961 , @xmath962 .
differently put , a general element of @xmath963 is @xmath964 where @xmath965 and @xmath966 is a cubic polynomial .
@xmath586 acts as @xmath967 on @xmath964 , so dually it acts as @xmath968 on @xmath958 .
in other words , @xmath4 has four odd moduli @xmath969 , @xmath970 all transforming as @xmath355 under @xmath586 .
along @xmath492 , the four sections of @xmath960 are @xmath971 transforming under @xmath586 as @xmath967 , and @xmath972 transforming as @xmath968 .
here @xmath973 is an abbreviation for @xmath974 , where @xmath40 was characterized in eqns .
( [ dundo ] ) and ( [ pundo ] ) and is assumed to transform as @xmath355 .
( note that @xmath974 has simple zeroes at @xmath975 , while @xmath976 has simple poles at those points , so that the two sections listed in eqn .
( [ secondtwo ] ) are regular and nonzero at @xmath975 , while the sections in ( [ firsttwo ] ) have simple zeroes there .
the behavior is reversed at @xmath977 .
also , @xmath978 is of order @xmath979 at infinity , while @xmath974 is of order @xmath899 at infinity and @xmath976 is nonzero and bounded there ; this is why we take @xmath980 in eqns .
( [ firsttwo ] ) and ( [ secondtwo ] ) . )
dual to this , along @xmath492 , @xmath4 has two odd moduli @xmath981 that transform as @xmath355 under @xmath586 and two odd moduli @xmath982 that transform as @xmath967 .
as we essentially learned in appendix [ bosonic ] , near either @xmath257 or @xmath492 , @xmath706 has 5 bosonic moduli @xmath983 that are even under @xmath586 and 1 bosonic modulus @xmath984 that is odd .
we define @xmath984 so that on the reduced space of @xmath706 , it restricts to the parameter called @xmath908 in appendix [ bosonic ] .
if we write @xmath588 for the same function of a @xmath490 period matrix that was introduced in section [ bosonic ] , then along the reduced space of @xmath706 , we have @xmath985 , precisely in parallel with the purely bosonic formula @xmath986 .
however , the relation @xmath985 has corrections when the odd moduli are turned on .
corrections to this relation that vanish at @xmath987 could be eliminated by redefining @xmath984 , so we are only interested in corrections that are nonvanishing at @xmath987 , or even have a pole there .
first we consider the behavior along @xmath492 .
since @xmath588 is a matrix element of the super period matrix , we can use the dhoker - phong formula for the dependence of the super period matrix on odd moduli to compute its dependence on the @xmath988 and @xmath989 . the leading correction to @xmath588 due to
the odd moduli is a function bilinear in odd moduli .
this function is constrained by @xmath586-invariance , but is otherwise fairly generic , and in particular has no reason to vanish at @xmath987 .
thus , the bilinear correction to @xmath588 is @xmath990 for some @xmath9 matrix - valued function @xmath991 . generically along
@xmath492 , this matrix is nondegenerate and we can pick the odd moduli @xmath981 and @xmath982 so that @xmath992 where the @xmath993 term comes from the terms in the dhoker - phong formula for the super period matrix that are quartic in the odd variables . since we have only explained what we mean by @xmath984 along the reduced space of @xmath706 , we are free to redefine @xmath984 by adding holomorphic terms that are bilinear in the odd variables @xmath994 .
but it is not possible in this way to eliminate the nilpotent terms on the right hand side of eqn .
( [ trefo ] ) . along @xmath257
, we have @xmath256 , and this means that the super period matrix has a pole at @xmath987 ; the dirac propagator , which enters the dhoker - phong formula , is proportional to @xmath995 .
so the analog of eqn .
( [ trefo ] ) is @xmath996 ( in the dhoker - phong expansion of the super period matrix , the term quartic in odd variables multiplies the product of two dirac propagators and so is @xmath997 . )
( [ prefo ] ) makes clear that @xmath491 has a pole along @xmath257 .
less obviously , however , this is also true along @xmath492 .
this is because the natural variable that we use in expanding a string or superstring theory measure about the hyperelliptic locus @xmath257 or @xmath492 is not the matrix element @xmath588 of the period matrix , but its square root @xmath908 , which is associated to a deformation of the complex structure of @xmath11 ( or its metric ) ; that is , it is associated to an element of @xmath998 . rather than give an abstract explanation of this statement
, we refer the reader to appendix [ benear ] , where the point will hopefully become clear .
regarding @xmath908 rather than @xmath986 as a local parameter along @xmath949 amounts to treating @xmath705 as a `` moduli stack '' rather than a `` moduli space . ''
let us see what happens if eqn .
( [ trefo ] ) is written in terms of not @xmath588 but @xmath999 .
we get @xmath1000 the ellipses in eqn .
( [ zrefo ] ) comes both from expanding the square root to higher orders and from the term quartic in odd variables that was omitted in eqn .
( [ trefo ] ) .
we should interpret this formula as giving the pullback of the function @xmath908 on @xmath705 to @xmath706 : @xmath1001 clearly , this pullback has a pole . if we are supposed to take @xmath908 seriously as a function on @xmath705 , then the fact that it pulls back under @xmath19 to a function on @xmath706 with a pole along @xmath492 shows that the projection @xmath491 has a pole along @xmath492 . to describe the superstring measure @xmath20 near @xmath492
, we use @xmath586-invariant local bosonic coordinates @xmath1003 , along with bosonic and fermionic coordinates @xmath984 and @xmath1004 that are not @xmath586-invariant .
the super mumford isomorphism ( see @xcite for original references and @xcite for an introduction ) says that . to make eqn .
( [ uerg ] ) simple and concrete , we have written in the denominator a section of this line bundle over @xmath705 . to write an accurate formula , one should replace @xmath1005 , @xmath1006 , with corresponding differentials on the super riemann surface @xmath4 .
the choice of these differentials will affect the function @xmath1007 , but has no essential bearing on our discussion below . ]
@xmath1008 } { \left(\frac{\d x}{y}\wedge x\,\frac{\d x}{y}\wedge x^2\,\frac{\d x}{y}\right)^5},\ ] ] where the function @xmath1007 is holomorphic , nonzero , and @xmath586-invariant .
notice that @xmath20 is @xmath586-invariant : @xmath1009 is odd under @xmath586 , the denominator @xmath1010 is also odd , and the rest of the formula is @xmath586-invariant .
it is crucial that , despite being @xmath586-invariant , @xmath20 can not be written in terms of @xmath1011 and @xmath1012 ; as we will see momentarily , this fact leads to a pole in @xmath21 along @xmath492 . to compute the measure @xmath21 on the reduced space @xmath705 , the first step is to re - express @xmath20 in terms of @xmath908 rather than @xmath984
we do this by solving eqn .
( [ zrefo ] ) for @xmath984 : @xmath1013 the function @xmath1014 receives a contribution from the quartic terms that were omitted in eqn .
( [ trefo ] ) , and also from the expansion of the square root in eqn .
( [ zrefo ] ) . from ( [ zormo ] ) , we have @xmath1015 where we have written only the most singular term for @xmath1016 . making this substitution in ( [ uerg ] ) and integrating over the odd variables , we find the singular behavior of @xmath21 along @xmath492 : @xmath1017 thus @xmath21 is of order @xmath1018 or @xmath1019 along @xmath492 .
the literature actually contains a proposal @xcite for a holomorphic measure on @xmath705 that is supposed to arise by integrating over the odd variables in some fashion that has not been specified .
this formula is holomorphic along @xmath492 , so it does not coincide with @xmath21
. we can similarly determine the behavior of @xmath21 along @xmath257 .
the difference between @xmath257 and @xmath492 is that , if @xmath4 is split , each of the 10 rns fermions has a pair of zero - modes along @xmath257 and hence the path integral of the rns fermions is proportional to @xmath1020 . along the split locus
, therefore , one has near @xmath1021 @xmath1022 } { \left(\frac{\d x}{y}\wedge x\,\frac{\d x}{y}\wedge x^2\,\frac{\d x}{y}\right)^5},\ ] ] with @xmath1023 non - zero . , the cohomology @xmath1024 jumps , since @xmath247 is nonzero along @xmath257 .
the denominator in ( [ verg ] ) therefore trivializes the appropriate line bundle only for @xmath1025 . as a result
the formula ( [ verg ] ) , which vanishes at @xmath1021 , does not exhibit the super mumford isomorphism at @xmath1021 . to do so
, one would have to write the formula in a more sophisticated way , taking into account the jumping of the cohomology , but this is not necessary for our purposes , basically because there is no jumping in the cohomology of @xmath11 . ] to generalize this formula away from the split locus , we have to take into account that , in conventional language , pco insertions can absorb some zero - modes of the matter fermions , as discussed in section [ poleffect ] .
however , the most singular behavior comes from expressing @xmath1026 in terms of @xmath1009 .
the analog of eqn .
( [ zormo ] ) is @xmath1027 ( the @xmath1028 term comes from solving eqn .
( [ merox ] ) for @xmath984 and does not depend on the @xmath1029 term in eqn .
( [ trefo ] ) . )
hence @xmath1030 where less singular terms are omitted . combining this factor of @xmath1031 with the @xmath1020 in eqn .
( [ verg ] ) , and integrating over the odd variables , we expect @xmath21 to vanish as @xmath1032 along @xmath257 . _ acknowledgments _ research was partly supported by nsf grant phy-1314311 .
i would like to thank e. dhoker , r. donagi , and d. phong for discussions , and dhoker and phong for help in reconciling some formulas here with their results .
i also thank p. deligne for detailed comments on an earlier version and for several helpful suggestions .
p. deligne and j. w. morgan , `` notes on supersymmetry ( following joseph bernstein ) , '' in p. deligne et .
, eds . , _ quantum fields and strings : a course for mathematicians , vol . 1 _
( american mathematical society , 1999 ) .
a. m. baranov , yu .
i. manin , i. v. frolov , and a. s. schwarz , `` a superanalog of the selberg trace formula and multiloop contributions for fermionic strings , '' commun .
* 111 * ( 1987 ) 373 - 392 . | we generalize the super period matrix of a super riemann surface to the case that ramond punctures are present . for a super riemann surface of genus
@xmath0 with @xmath1 ramond punctures , we define , modulo certain choices that generalize those in the classical theory ( and assuming a certain generic condition is satisfied ) , a @xmath2 period matrix that is symmetric in the @xmath3-graded sense . as an application , we analyze the genus 2 vacuum amplitude in string theory compactifications to four dimensions that are supersymmetric at tree level .
we find an explanation for a result that has been found in orbifold examples in explicit computations by dhoker and phong : with their integration procedure , the genus 2 vacuum amplitude always vanishes `` pointwise '' after summing over spin structures , and hence is given entirely by a boundary contribution . *
* .15 cm * * .5 cm |
the mass of a central black hole ( bh ) in a galaxy can be directly weighed using gravitational test particles moving around it ( see e.g. , kormendy & richstone 1995 ) .
commonly used test particles are stars , optical emission - line gas and maser clouds .
these move around in the combined potential well of the stellar mass and the central black hole mass .
a secure direct dynamical measurement of the bh mass requires that the distribution and kinematics of the test particles are measured in the immediate vicinity of the bh , i.e. , the black hole s sphere of influence , so that its gravitational potential has a measurable effect on the kinematics in addition to the effect of the stellar mass potential .
ground - based optical telescopes can resolve the bh s sphere of influence only for very nearby galaxies , with our milky way being a spectacular nearby example ( e.g. , ghez et al .
2003 ; schdel et al .
the hubble space telescope ( hst ) can provide the required spatial resolution for galaxies with distances up to several tens of mpc ( e.g. , kormendy & gebhardt 2001 ) .
the very long baseline interferometer ( vlbi ) can in principle probe bh masses out to even larger distances provided that the galaxies display nuclear maser emission . together
these methods have led to direct dynamical measurements of bh masses @xmath3 in the range @xmath4 in several tens of galaxies in the nearby universe ( see e.g. , tremaine et al . 2002 ; marconi et al .
2003 for listings of bh mass measurements ) .
the bh masses are mostly based on dynamical modeling of kinematics of either stars or emission - line gas .
the detected black hole masses @xmath3 correlate well with both host spheroid luminosity @xmath5 , global stellar velocity dispersion @xmath6 and spheroid mass ( gebhardt et al .
, 2000 ; ferrarese & merritt , 2000 ; tremaine et al . 2002 ; marconi et al . 2003 ; hring & rix 2004 ) .
these correlations roughly represent our current knowledge of the local bh demography . to explore these correlations further , i.e. , their scatter and extent in bh mass , dynamical modeling using emission - line gas kinematics in early - type galaxies is convenient and in some respects even crucial .
first of all , more than 50% of the early - type galaxies in the nearby universe contain gas at their centers ( e.g. , goudfrooij et al .
1994 ; ho , filippenko & sargent 1997 ) .
second of all , the most massive black holes currently found ( i.e. , @xmath7 ) are often in giant ellipticals .
the central stellar surface brightness of these galaxies is typically too low to obtain accurate stellar kinematics with hst and maser emission has not been detected ( e.g. , barth et al .
2004 ) . at the centers of galaxies ,
the collisional gas is expected to settle quickly into a disk , if unperturbed by forces other than gravity ( e.g. , habe & ikeuchi 1985 ) .
the frequent detection with hst of disk - like structures in dust and gas in early - type galaxies suggests that settling actually takes place ( e.g. , van dokkum & franx 1995 ; tran et al .
2001 ; verdoes kleijn et al .
1999 ; laine et al .
therefore the extended gas velocities in galactic nuclei have been modeled assuming a thin disk in circular rotation .
such models have been successful in explaining the observed gas velocities .
however , in the few cases where both gas velocities and gas dispersions are modeled , it is often found that the nuclear velocity dispersion exceeds the prediction for a thin disk ( van der marel & van den bosch 1998 ; barth et al .
2001 ; maciejewski & binney 2001 ; cappellari et al . 2002 ; verdoes kleijn et al .
many of these targets are agn and display regular disks of dust and gas .
the origin of this excess velocity dispersion is unknown .
two common ad - hoc assumptions are that : ( a ) the excess dispersion is due to non - gravitational turbulent forces of unknown origin which do not affect the mean circular rotation of the gas ( e.g. , van der marel & van den bosch 1998 ; verdoes kleijn et al .
2002 ) ; or ( b ) the excess dispersion is purely gravitational and affects the rotation of the gas through the asymmetric drift equation ( e.g. , barth et al .
2001 ) .
a main aim of this paper is to determine which early - type galaxies do and which do not display an excess in nuclear gas dispersion .
if there are galaxies for which the disk remains thin all the way to the nucleus , then any velocity dispersion observed through a nuclear aperture is caused by differential rotation over this aperture . a nuclear aperture samples gas which is much closer to the bh than a series of apertures which samples the extended rotation curve .
hence , the nuclear gas dispersion can be sensitive to much lower black hole masses than the rotation curve .
this is relevant as the sensitivity in terms of the minimum detectable bh mass in early - type galaxies with stellar dynamical methods or with gas rotation velocities is typically not much below the values predicted by the correlations between bh mass and spheroid properties .
a relevant example in this case is the claim that active spiral galaxies classified as narrow line seyfert 1s harbor bhs with masses smaller than those predicted by the @xmath8 relation ( e.g. , grupe & mathur 2004 ) . for the galaxies with nuclear gas dispersions in excess of that expected from a thin disk , we want to determine the origin of this excess .
on the one hand , if it has a gravitational origin , this implies that the gas should rotate less fast than the circular speed and the thin - disk approximation does not hold .
the gas might then have a distribution somewhere in the range from a thin disk to a purely spherical distribution .
previous bh mass estimates based on the assumption of thin disks would then be underestimates , and this would affect the observed correlations between black hole mass and large - scale spheroid properties .
on the other hand , if the excess dispersion has a non - gravitational origin , we would like to quantify how it depends on galaxy parameters , such as nuclear activity .
this in turn could shed light on the accretion process in these nearby , typically low - luminosity , agn . in summary
, this paper studies the nature of the line widths of emission - gas in nearby early - type galaxy nuclei and its implications for the bh demography and bh accretion processes .
a similar study into the line - widths of gas for late - type galaxies has been performed by sarzi et al .
the outline of the paper is as follows .
section [ s : sampledata ] describes the sample selection , the data and the determination of the central gas velocity dispersions .
section [ s : excess ] compares the observed dispersions to those expected from a thin circular disk .
section [ s : spheroid ] compares the observed dispersions to those expected from a gas distribution which is more spheroidal than a disk .
section [ s : nongrav ] constrains the direction of the non - gravitational dispersion .
section [ s : cave ] describes the caveats of our modeling .
finally , section [ s : conc ] provides a discussion of the results and lists the main conclusions of the paper . throughout the paper we use a hubble constant @xmath9= 75 kms@xmath10mpc@xmath10 .
we analyze a sample of 27 galaxies .
it consists of galaxies which meet the following requirements : ( a ) early - type host morphology ; ( b ) distance below 110 mpc ; ( c ) a large - scale stellar velocity dispersion measurement is available from the literature ; ( d ) hst imaging exists that shows an identifiable ( i.e. , relatively unobscured ) nucleus ; ( e ) hst emission line spectroscopy exists for which kinematics and flux profiles were published ( 20 galaxies ) or which are available in the hst archive ( 7 galaxies ) .
we excluded galaxies for which the signal - to - noise ratio of the spectrum was very low , complicating a reliable analysis .
sixteen of the galaxies are classified as fanaroff & riley type i radio galaxies with jets on scales of tens of kpc or larger .
the other 11 galaxies also have radio emission .
their radio luminosity is typically much lower ( cf .
table [ t : sample ] ) and the radio emission originates from a central compact component .
we reduced and analyzed the archival data for the 7 galaxies similarly as in our previous papers ( e.g. , noel - storr et al .
2003 ; verdoes kleijn et al .
the spectra for all galaxies except two were obtained with hst / stis in combination with the g750 m grating . for ic 1459 a hst / stis spectrum with the g430l grating
was obtained and for ngc 6251 a hst / fos spectrum with the @xmath11 pair b aperture . all spectra cover the h@xmath12+[nii ] and [ sii]6716,6731 lines , except for the g430l spectrum of ic 1459 . in that case
we use the h@xmath13 line for our analysis .
table [ t : sample ] lists basic data for the sample galaxies .
the first step is to determine the gas velocity dispersions in the central region of each galaxy . in most cases
( both in the literature and in our analysis of the archival spectra ) single gaussian fits were made to each of the lines of the h@xmath12+[nii ] composite and to the weaker [ sii]6716 , 6731 lines . only in ngc 3245 and ngc
4526 are the [ sii ] lines too weak to be fitted . for ngc 6251
we used the fit to the h@xmath12+[nii ] lines available in ferrarese & ford ( 1999 ) . for these narrow components ,
the [ nii]@xmath14 was the strongest line in @xmath15 of the galaxies . in four galaxies ( ngc 3245 , ngc 3998 , ngc 4278 and ngc 6251 )
a broad emission - line component is clearly present in addition to the narrow h@xmath12+[nii ] components .
this is most likely due to a broad h@xmath12 component because the broad component is not seen in the [ sii ] forbidden lines and the forbidden [ nii ] lines originate presumably in the same region as the [ sii ] lines .
this broad component was fitted with an additional gaussian .
for the two archival sources with a broad component ( ngc 3998 and ngc 4278 ) we established that the [ sii ] lines provide an important fitting constraint in addition to the h@xmath12+[nii ] region which is blended heavily due to the broad component .
excluding the [ sii ] doublet from the fit results in @xmath16 larger ( ngc 3998 ) or smaller ( ngc 4278 ) dispersion .
similarly , ferrarese & ford ( 1999 ) report for ngc 6251 a width of the [ sii ] lines which is @xmath17 larger than the h@xmath12+[nii ] lines at the nuclear aperture . for the archival cases with only narrow lines
, we determined that the inferred gas dispersion changes by @xmath18 by excluding the [ sii ] doublet from the fit .
the next step is to find for each galaxy the aperture closest to the nucleus which we will refer to as the central aperture .
we assume that the dynamical nucleus coincides with the peak in emission - line and continuum flux .
the typical dimension of the stis apertures is @xmath19 in the spectral direction and @xmath20 in the spatial direction ( see table [ t : data ] ) .
the exact sub - pixel location along the slit of the nucleus was determined by fitting a gaussian to the central few emission - line fluxes . in most cases
the peak in velocity dispersion coincides with the flux peak .
we indicate in table [ t : sample ] the few cases in which the maximum dispersion does not peak at the central aperture but just outside it .
the exact location of the nucleus ( i.e. , the flux peak under our assumptions ) in the direction of the slit width is not known .
we assume that it is centrally located in this direction .
this seems plausible given that the target acquisition procedure is designed to have the center of the slit located exactly on the brightest continuum point in the galaxy .
that this peak - up procedure actually succeeded is supported by the steep decline in flux in parallel slits which are available for many targets .
nevertheless , the effect on our results of a potential off - set from this location is discussed in section [ s : cave ] .
the gas velocity dispersion @xmath21 at the central aperture is listed in table [ t : sample ] .
the typical relative formal measurement error for @xmath21 is @xmath22 .
we also require the unconvolved flux profile for our analysis .
the emission - line surface brightness is represented by fitting a double exponential function , @xmath23 to the narrow emission - line fluxes taking into account the psf and disk inclination ( see e.g. , verdoes kleijn et al .
2002 for a detailed description ) .
the disk inclinations were taken from the literature or determined by us from the extended dust disk often present on hst imaging .
if dust is absent or the inclination could not be determined reliably , we assumed @xmath24 .
the hst observational characteristics , unconvolved flux profiles and disk inclinations are listed in table [ t : data ] .
the fluxes are typically fitted within the formal errors or within 10% .
a few galaxies have some asymmetry in their profiles over which the symmetric fit averages , namely ic 1459 , ngc 3245 , ngc 4278 , ngc 4459 and ngc 4486 . finally , we use published stellar dispersions ( see table [ t : sample ] ) .
the stellar dispersions @xmath6 are measured through an aperture of typically several arcsec@xmath25 .
this corresponds to a region much larger than the bh sphere of influence , and measures primarily the virial dispersion of the system as a whole .
we will make use of the stellar dispersions to estimate black hole masses using the @xmath26 relation as determined by tremaine et al .
the relation is calibrated to flux - weighted stellar dispersions @xmath27 inside an effective radius of the galaxy , which are not available for most of our sample galaxies . however , comparing @xmath27 and @xmath6 where possible shows a difference of typically @xmath28 between @xmath6 and @xmath27 .
we will use this as the typical relative error on our measurement of @xmath6 in approximation to @xmath27 .
lastly , table [ t : sample ] also lists the black hole masses obtained from dynamical modeling of central gas disk rotation velocities as published for eight galaxies in the sample .
figure [ f : sigmagassigmastar ] shows the central gas velocity dispersion at small scales ( i.e. , typical aperture size @xmath29 or @xmath30 for typical galaxy distance of @xmath31 ) versus the large - scale stellar dispersions ( tracing the flux - weighted dispersion at kpc scales ) .
the nuclear gas dispersions are almost always larger than the large - scale stellar dispersions , increasingly so for larger stellar dispersions .
given the success rate of black hole mass determinations in early - type galaxies , it now seems quite possible that all spheroids harbor a black hole at their nucleus .
there are two generic ways in which the presence of a bh can contribute to the nuclear gas dispersion .
first , gas motions increase at smaller distances from a central black hole due to the increase of the gravitational force .
integrated over a finite aperture this will lead to a larger observed gas dispersion .
we will call this the gravitational contribution .
an increase will occur always , but its magnitude will depend on the physical structure and orbital distribution of the gas , e.g. , a thin rotating disk or a more spheroidal distribution .
the effect is noticeable because for a typical @xmath32 in our sample , the typical central hst aperture is only a factor two larger than the rough measure of the black hole radius of influence @xmath33 adopting the @xmath26 relation from tremaine et al .
the relation finds @xmath34 with @xmath35 .
if the gas motions are dominated by gravitation it is expected that @xmath36 with @xmath37 and hence @xmath38 qualitatively consistent with the increasing ratio of @xmath21 and @xmath6 shown in figure [ f : sigmagassigmastar ] .
the @xmath26 relation also suggests that one should expect a smaller excess gas dispersion in galaxies with smaller @xmath6 for another reason : galaxies with small @xmath6 have a smaller bh mass and a sphere of influence so that a stis aperture of fixed size samples more of the gas outside of the bh sphere of influence .
a second generic way in which the presence of a bh can contribute to the nuclear gas dispersion is not through its gravitational force , but through input of kinetic energy which perturbs the collisional gas .
we will call this the hydrodynamical or non - gravitational contribution .
this energy could be released by processes related to an active black hole . as a side remark ,
it is unlikely that the kinetic energy might be provided by collisions with photons emerging from the active region around the black hole .
the reason is that the bolometric luminosity of the low - luminosity active nuclei discussed here is always orders of magnitudes below the eddington luminosity ( e.g. , ho 1999 ) .
shocks , e.g. , jet - gas interactions are a more plausible source of kinetic energy .
it could be that the gravitational contribution dominates ( e.g. , a thin disk of non - colliding gas particles in circular rotation ) or that the hydrodynamical contribution dominates ( e.g. , a fully collisionally driven outflow of gas from the active black hole ) .
the well - defined thin dust and gas disks often seen immediately outside the nucleus in these galaxies suggest that the gas might be settled all the way to the nucleus .
therefore , we will first determine the answer to the question : in which galaxies can the bh potential account for the observed nuclear gas dispersion assuming that the gas is located in a thin circular disk ?
the answer to this question is known already for five galaxies in our sample for which detailed modeling has been performed of both gas velocities and dispersions of the extended gas disk under the assumption of a thin circular disk . in the case of ngc 7052 ( van der marel & van den bosch 1998 ) , ic 1459 ( cappellari et al .
2002 ; verdoes kleijn et al .
1999 ) and ngc 4335 ( verdoes kleijn et al .
2002 ) , an excess of dispersion by a factor @xmath39 or more was observed . for ngc 3245 ( barth et al .
2001 ) , a smaller excess of @xmath16 was observed . in the case of m87 ( macchetto et al .
1997 ; harms et al .
1994 ) , no dispersion excess was observed .
the velocity dispersion for this galaxy was accounted for by a model in which the gas resides in an annulus instead of a disk .
( this model can also account well for the velocities and peculiarities in the flux distribution . )
it is these mixed results on excess gas dispersion that lead us to examine the presence of excess gas dispersion for a larger sample of galaxies . in the present and following sections ( section [ s : excess][s : nongrav ] ) we present results of dynamical models of the gas velocity dispersions
. caveats that result from the assumptions in our models are discussed in section [ s : cave ] . performing gas disk kinematical modeling for our complete sample in similar detail as done for the aforementioned few cases
is beyond the scope of this paper .
thus we constructed somewhat more simplified thin disk models .
the main simplification is to neglect the stellar mass contribution to the central gravitational potential .
the effect of this simplification on the predicted velocity dispersions for the galaxy sample can be estimated as follows .
assume a spherical stellar mass density @xmath40 in the nuclear region , where @xmath41 is the radius in pc and @xmath42 a scaling constant . according to the leda database , the absolute blue magnitude in our sample varies in the range
@xmath43 $ ] with a mean @xmath44 .
assuming a typical galaxy color @xmath45 , this indicates that a typical galaxy in our sample is about 1 magnitude brighter than the division magnitude at @xmath46 between shallow core and steep core ( `` power - law '' ) ) galaxies ( gebhardt et al .
galaxies brighter than this division magnitude typically have @xmath47 ( while below they have @xmath48 ) . from the study by gebhardt
( 1996 ) we infer a typical @xmath49 . with this information
we can compute the circular velocity @xmath50 due to the stellar mass at a given radius as @xmath51 . by comparison ,
the circular velocity due to the bh is @xmath52 .
we compute the ratio of these two at @xmath53 from the nucleus assuming a bh mass according to the @xmath26 relation .
the value @xmath53 is a typical distance from the galaxy center at which we have measured gas dispersions with hst .
the ratio of the circular velocities in quadrature ( i.e. , equivalent to the ratio in dynamical mass ) is always less than 20% .
this is much smaller than , e.g. , the scatter of @xmath54 dex in the @xmath26 relation ( tremaine et al .
this confirms that the stellar mass contribution in our analysis can be safely neglected , provided that : ( a ) the bh masses are not significantly below the @xmath26 relation ; and ( b ) we are not attempting to model the large scale rotation of the gas at radii much beyond @xmath0 . as an additional check we directly determined the difference in predicted dispersion with and without bh mass for ngc 4335 , ngc 7052 and ic 1459 . for these galaxies , we modeled the stellar mass in previous papers ( verdoes kleijn et al . 2002 ; van der marel & van den bosch 1998 ; verdoes kleijn et al .
these cases confirm the result from the general argument that the difference in predicted dynamical masses is always less than 20% ( see also figure [ f : mbhlit ] , discussed below ) .
for a thin circular disk , the velocity dispersion observed in a nuclear aperture is due entirely to differential rotation over the aperture . to calculate this dispersion we used the modeling software that is described in more detail in van der marel & van den bosch ( 1998 ) and
verdoes kleijn et al .
( 2000 ; 2002 ) .
this modeling takes into account the emission - line flux profile , slit width and psf convolution .
we first modeled the galaxies for which a bh mass estimate is available in the literature from detailed modeling of the gas rotation velocities .
figure [ f : mbhlit ] shows the predicted versus observed velocity dispersions for these eight galaxies .
seven galaxies have an observed dispersion in excess of the prediction .
the modeling nicely reproduces the aforementioned relative differences in excess dispersion for m87 , ngc 3245 , ngc 7052 , ngc 4335 and ic 1459 known from prior modeling of the rotation velocities in the extended gas disk . to put the complete sample in a similar diagram
, we need an estimate of the bh mass for every galaxy . for this
we use the @xmath8 relation as calibrated by tremaine et al .
they derive the relation @xmath55 = 8.13 + 4.02 \log ( \sigma/ 200 { \>{\rm km}\,{\rm s}^{-1}})$ ] .
the slope of this relation falls between those reported by gebhardt et al .
( 2000 ; slope=3.75 ) and ferrarese & merritt ( 2000 ; slope=5.27 ) . figure [ f : mbhdisk ] shows the predicted versus observed dispersion for a thin disk model , but now assuming for _ all _ galaxies a bh mass according to the @xmath26 relation .
the observed dispersions in non - radio galaxies scatter around the prediction from the model ( except for ic 989 : cf . section [ s : conc ] ) .
in fact , the number of non - radio galaxies and their range in observed gas dispersions is large enough that they independently confirm the @xmath8 relation for the thin disk model .
this is non - trivial because bh masses significantly below the @xmath8 relation could have been detected with the hst / stis observations . as a measure of the minimally detectable bh mass ,
we compute the bh mass which causes a gas dispersion which equals that predicted from the combined effect of instrumental broadening and the stellar mass model discussed above .
we use again the typical radius of @xmath11 and take into account the disk inclination and distance for each galaxy .
the resulting masses vary between @xmath56 and @xmath57 with a typical value of @xmath58 .
( exceptions are ngc 3862 and ugc 7115 .
their minimally detectable bh mass lies above the @xmath8 relation because the instrumental line broadening is larger than the dispersion due to differential gravitational rotation for such close to face - on gas disks . )
the bh masses predicted by the @xmath8 relation exceed this typical detectable bh mass by a factor @xmath39 for the lowest stellar dispersions and by almost 2 orders of magnitude for the highest stellar dispersions in the sample . in conclusion ,
the gas dispersions in non - radio galaxies support the idea that all early - type galaxies harbor supermassive black holes at their nuclei with masses according to the @xmath8 relation .
in contrast to the non - radio galaxies , all radio galaxies have a gas dispersion in excess of that expected from the thin disk models . about @xmath59 of those fall above the @xmath60 scatter expected from the intrinsic scatter in the @xmath26 relation . to explore the origin of this we plot in figure [ f : sigratiodisk ] the ratio @xmath61 of
observed dispersion and that predicted for the disk model .
there is a trend of an increasing ratio with decreasing disk inclination .
in fact , the four largest ratios are all in galaxies with gas disk inclinations @xmath62 , i.e. , close to face - on disks .
this can not be explained as a result of measurement errors in the dust disk inclinations .
the arrows in figure [ f : sigratiodisk ] show by how much @xmath61 can decrease due to the known measurements errors .
the decreases do not change the overall result that radio galaxies systematically have excess dispersions .
a more extreme ( and probably unphysical ) possibility is to assume that we might have used incorrect inclinations because the measured dust disk inclinations bear no relation to the actual inner gas disk inclinations .
however , this would still not provide a satisfactory explanation .
if all the gas disks were in reality edge - on and well described by thin disk models , then we would have expected the points in figure [ f : sigratiodisk ] to fall along the curve @xmath63 ( solid curve ) .
this still falls below many of the observed dispersion ratios ( see figure [ f : sigratiodisk ] ) .
moreover , an excess gas dispersion is not only seen in radio galaxies with relatively face - on disks . if we exlude systems with disk inclinations @xmath62 from the analysis , the median @xmath64 for radio galaxies versus @xmath65 for non - radio galaxies . a median test yields a probability of only 0.05 that this difference would occur by chance in the event that the samples were drawn from populations with the same median .
therefore , a solution to account for the excess gas dispersion in radio galaxies must probably be sought in having a significant vertical velocity dispersion component . finally , as a sanity check we analyzed the dependence of the @xmath61 on bh mass and the typical circular velocity of the system , parameterized as @xmath66 where @xmath67 is the radius inside which half of the light of the unconvolved emission - line flux distribution is contained .
there is no relation between the dispersion ratio and the two measures of the gravitational potential .
this confirms that the excess velocity dispersion in the radio galaxies is not somehow an artifact of our gravitational modeling .
many galaxies in the sample show disks of dust and gas with radii ranging from hundreds of pc up to several kpcs .
it appears reasonable , but might be wrong to assume by extrapolation that the nuclear region within the spatial resolution of the observations also has a disk - like geometry .
instead , the gas distribution might be more spheroidal in the vicinity of the black hole . in that case
the gas moves also out of the plane of the larger scale thin gas disk .
this could explain qualitatively the observed increasing ratio of observed versus predicted dispersion for decreasing dust disk inclination .
we address next whether such a spheroidal distribution can also quantitatively explain the observed @xmath21 gravitationally .
gas in a vertically extended distribution ( e.g. , a spherical one ) tends to collapse quickly to a disk due to hydrodynamical forces .
thus , to maintain a spheroidal structure , one would have to assume that the gas is located in individual clouds that move collisionlessly . as an initial assessment of the plausibility of this we estimate the collision time - scale for clouds in a simplistic model : spherical cloudlets of ionized hydrogen moving around in a spherical volume .
the typical time @xmath68 between cloudlet collisions is estimated as : @xmath69 where @xmath70 is the typical relative velocity of cloudlets , @xmath71 is the spherical volume within which the cloudlets are contained , @xmath72 is the cross - section of each cloudlet and @xmath73 is the total number of clouds .
for @xmath71 we take the spherical volume with a radius equal to the half - light radius of the gas distribution as seen projected on the sky . for @xmath70
we take the circular velocity at half the half - light radius .
the number of cloudlets @xmath74 is computed from @xmath75 where @xmath76 is the total number of electrons , @xmath77 is the electron density and @xmath78 is the radius of each cloudlet . the electron density can be obtained from the relation between the flux ratio [ sii]@xmath796716/@xmath796731 and the electron density ( osterbrock 1989 ) .
this flux ratio is only sensitive to electron densities in the range @xmath80$]@xmath81 . outside this range
the flux ratio becomes constant as a function of @xmath82 .
we obtain the flux ratio for 15 of the radio galaxies from noel - storr et al .
( 2003 ) of which six lead to upper and lower - limits on the electron density .
@xmath76 can be estimated as discussed in osterbrock ( 1989 ) assuming case b recombination : @xmath83 where @xmath84@xmath81 is the recombination coefficient ( assuming @xmath85k ) and @xmath86 and @xmath87 are the frequency and luminosity for h@xmath13 .
we obtain @xmath88 from the central h@xmath12+[nii ] luminosity ( verdoes kleijn et al .
2002 ) using the standard @xmath89 ratio of 3.1 and the [ nii]@xmath906548,6584 / h@xmath12 flux ratio determined from our gaussian line fits .
this results in the following equation for the collision time : @xmath91 this increases linearly with the radius of the cloudlets @xmath78 . if we assume that @xmath78 is as large as a tenth of the half - light radius , then the inferred @xmath68 varies between @xmath92yr and @xmath93yr with a typical value of @xmath94yr .
some of the huge scatter is most likely due to the simplicity of the estimate .
but either way , it appears unlikely that the gas could remain in a collisionless state for a significant fraction of the hubble time .
collisions lead to dissipation , loss of energy , and settling onto circular orbits .
however , this only true if there is no energy input into the gas .
that may well be incorrect for most of the galaxies in our sample , given that 16 galaxies are fr i radio galaxies , and the remainder have nuclear radio emission .
if there is energy input , then the gas can maintain an extended distribution for much longer than the timescale @xmath68 calculated above . even though collisions would occur
, one might expect the system to evolve through a sequence of states that are all approximately collisionless .
therefore , it is reasonable to study the predictions of collisionless models for the nuclear gas in our sample galaxies .
a major uncertainty in the calculation of collisionless models is the phase - space distribution of the clouds , which is needed to compute the observed velocity dispersion along the line of sight .
the phase - space distribution involves a spatial component such as a radial number density profile .
we do not know this profile .
we only know the flux distribution , which could be quite different .
the phase - space distribution also requires knowledge about the nature of the cloud kinematics , for example anisotropies in the velocity dispersion .
these are also unknown .
we therefore decided to explore the line - of - sight velocity dispersion for a plausible range of phase - space distributions .
we use axisymmetric dynamical models presented in de bruijne et al .
we explore gas clouds with an axisymmetric power - law number density distribution with axis ratio @xmath95 ( i.e. , @xmath96 $ ] ) in the kepler potential caused by the black hole .
the gas clouds are assumed to have a constant velocity dispersion anisotropy @xmath97 ( i.e. , the case ii models in de bruijne et al .
1996 ; cf .
binney 1980 , with @xmath98 the usual spherical coordinates ) .
thus we have four free parameters : the number density scaling constant @xmath42 , the axis ratio @xmath95 , the power - law slope @xmath99 of the particle density profile and the velocity dispersion anisotropy @xmath13 .
these models require @xmath100 to be physical .
de bruijne et al . ( 1995 ) evaluated the projected velocity dispersions .
the software to calculate them numerically is available from http://www.stsci.edu/@xmath101marel/. we want to know how different the observed velocity dispersion for a spheroid can be compared to the dispersion for a thin disk , for a plausible range of the free parameters .
thus we determined the ratio @xmath102 of the velocity dispersions for a spheroid and a thin disk that have the same projected surface density distribution and apparent axial ratio .
the latter criterion fixes the ratio of the @xmath42 values for the two models , and implies that @xmath102 is independent of @xmath42 .
we define @xmath102 as the ratio of the line - of - sight dispersion weighted by the number density and integrated over an aperture which is typical for the observations .
the result depends on power - law slope , axis ratio and dispersion anisotropy .
we explored the parameter ranges @xmath103 $ ] , @xmath104 $ ] and @xmath105 $ ] , which should encompass all plausible models .
figure [ f : spheroid ] shows @xmath102 as a function of apparent axis ratio for the quoted parameters ranges .
regardless of the choice of parameters , the modeled spheroidal distributions have increased velocity dispersions in comparison to thin disks only for @xmath106 ( corresponding to inclinations @xmath107 for a thin disk ) .
as also shown in figure [ f : sigratiodisk ] , the non - radio galaxies scatter around @xmath108 indicating that their dispersions are consistent with thin disks in circular rotation around bhs that follow the @xmath26 relation .
by contrast , the radio galaxies have dispersion ratios @xmath61 which are systematically larger than 1 , and also larger than the @xmath102 expected for collisionless spheroidal models .
so it seems unlikely that a spheroidal distribution of collisionless gas clouds can account for the nuclear gas velocity dispersions in radio galaxies .
we also explored the ratio of the velocity dispersions for an axisymmetric spheroid and a disk at the same inclination instead of the same apparent axis ratio .
the reason is that the axis ratio of the gas distribution at the nucleus is not constrained very well by the observations and might deviate from the axis ratio of the larger scale disk .
the resulting @xmath102 range for spheroid models at a given inclination remains very similar .
so this does not alter the conclusions inferred from figures [ f : sigratiodisk ] and [ f : spheroid ] .
the analysis of thin disks and spheroidal models indicates that non - gravitational forces might contribute significantly to the observed gas dispersion for radio galaxies . to constrain the minimal relative contribution of a non - gravitational dispersion component we plot in the top panel of figure [ f : nongrav ] the ratio of the observed dispersion and the maximum dispersion predicted by either disk or spheroid models .
the minimal ratio is typically between 1 and 2 .
there is a hint that the minimal ratio increases with axial ratio , i.e. , for extended disks that are closer to face - on .
this could indicate that the non - gravitational dispersion component does not have a random orientation with respect to the extended disk but is oriented preferentially along the spin axis of the extended disk .
we explore this idea with two simple models .
the models consist of two components : a thin circular rotating disk and a second component of non - gravitational motions .
we assume equal contribution to the emission - light by the two components . in the isotropic model
the non - gravitational dispersion is isotropic ( @xmath109 ) and the resulting total dispersion can be written as @xmath110 where @xmath111 denotes the dispersion due to circular rotation as measured in the plane of the disk and @xmath112 is the disk inclination . in the perpendicular model
the non - gravitational dispersion is assumed to be perpendicular to the plane of the disk : @xmath113 the middle and bottom panels of figure [ f : nongrav ] show the ratios @xmath114 and @xmath115 as a function of disk axis ratio .
the value of @xmath111 is calculated as before from the thin disk models and @xmath109 and @xmath116 , respectively , are chosen so that @xmath117 matches the observed value @xmath21 .
the perpendicular model shows no trend between the dispersion ratio and disk inclination .
the isotropic model shows a weak trend .
this would not be expected in an isotropic scenario . on the other hand , the trend in figure [ f : nongrav]b is dominated by a few galaxies and is not really significant .
for some galaxies the implied non - gravitational dispersion is larger than @xmath118 . in other words ,
the typical non - gravitational motions are larger than @xmath119 times the typical circular rotation in those galaxies which suggests that some of the material might be unbound .
however , the models are too simplistic to attach real significance to this result . the fact that radio galaxies show an excess gas dispersion , while it appears to be absent in non - radio galaxies , possibly points to a link between non - gravitational motion and activity .
the agn could be a source of energy to drive the non - gravitational dispersion e.g. , by coherent flows or turbulence .
we do not have a direct measure of the kinetic energy output of the agn to determine its correlation with the amount of non - gravitational dispersion present in the gas .
thus we resort to measures of the radiative energy output of the active nuclei as a proxy .
figure [ f : agn ] shows @xmath61 as a function of total radio power for both radio and non - radio galaxies .
as expected , the difference in excess dispersion for non - radio and radio galaxies is also present as a function of radio luminosity .
however , there is no clear indication that larger non - gravitational motions occur in more powerful radio galaxies .
we examined also the dependence of @xmath61 on radio core and nuclear emission - line luminosity for radio galaxies .
no clear trends were found .
similarly , no trends were found between the ratios @xmath120 and @xmath121 with these three indicators of the radiative power of the active nucleus .
* uncertainties in the flux profiles * for three galaxies in our sample , ngc 3862 , ngc 4374 and ngc 4486 , we have both the stis flux profiles and those derived from hst / wfpc2 pc - chip emission - line images which are available to us . the pc pixel size ( @xmath122 ) better samples
the hst psf fwhm ( @xmath0 ) compared to the stis apertures . from the pc emission - line image ,
a narrower flux profile is inferred for ngc 3862 and ngc 4374 , but a wider profile for ngc 4486 , as compared to the flux profiles derived from the stis data .
using the pc profiles in the thin disk modeling leads to increases in predicted gas velocity dispersion of @xmath22 , @xmath59 and @xmath123 for ngc 3862 , ngc 4374 and ngc 4486 , respectively . assuming that similar changes could be expected for the other sample galaxies , this does not change our conclusions .
another possibility to consider is that we might have systematically overestimated the width of the emission - line flux profiles .
if the flux in reality originates closer to the nucleus then in our models , it would naturally explain why the predicted dispersions are lower than observed .
figure [ f : halflight ] shows the radius inside which half of the flux from the intrinsic ( i.e. , unconvolved ) flux profile is contained as a function of distance .
the top panel shows the angular half light radius . for half light radii
much less than the half light radius of the hst psf , i.e. , @xmath124 , it could well be that the intrinsic flux profile is in fact much narrower .
thus , the profile widths for ngc 2329 , ngc 3862 and ngc 6251 should be considered upper limits .
this implies that their predicted velocity dispersions should be considered lower limits as emission - line gas might be rotating closer to the black hole than in our models . however , there are two facts that argue against a general overestimate of the width of the flux profiles for all galaxies .
first , we find no trend of narrower emission - line flux profiles with increasing dispersion ratios .
second , such a scenario can not explain the correlation between disk inclination and dispersion ratio .
a separate question is whether we are resolving the flux profile at larger distances as well as nearby .
this issue is relevant for comparing the properties of the radio galaxies and non - radio galaxies , since the former sample has a larger average distance .
the bottom panel of figure [ f : halflight ] suggests that this distance offset is not inducing the difference in excess dispersion , since there is no trend between the physical half light radius and distance .
finally , what kind of emission - line flux profile is needed to obtain the observed gas velocity dispersions gravitationally ?
to first approximation @xmath125 , where @xmath126 is the typical radius of the flux profile ( e.g. , the half light radius ) . figure [ f : sigratiodisk ] shows that for radio galaxies @xmath126 needs to decrease by a factor of about 2 up to 100 to have a model with a black hole mass according to the @xmath8 relation account for the observed gas dispersion .
figure [ f : halflight ] shows that the emission - line flux profile for radio galaxies typically has a half light radius which is 2 times the radius of the psf . in conclusion ,
the emission - line profile for radio galaxies needs to be fully unresolved in general to account for the observed gas velocity dispersion gravitationally .
this is ruled out by the observations .
* uncertainties in the location of the black hole * the spatial flux distribution in the direction of the slit length allows us to precisely locate the nucleus along this direction ( assuming it coincides with the peak of the emission - line flux , cf . section [ s : sampledata ] ) .
but the nucleus was _ assumed _ to be at the center of the slit in the direction of the slit width .
this is a reasonable assumption ( see section [ s : sampledata ] ) .
we determined that a @xmath127 displacement in this direction changes the predicted velocity dispersion by @xmath128 for the thin disk models .
this does not affect our conclusions .
* does the @xmath26 relation underestimate bh masses ? * one might account for the excess gas dispersion within the purely gravitational models , seen particularly for radio galaxies , by assuming that the bh mass from the @xmath26 relation underestimates the true bh mass .
however , this requires increases in bh masses that are too large to be credible for the following reasons .
first , comparison of figure [ f : mbhlit ] and [ f : mbhdisk ] shows that detailed estimates of bh masses from gas rotation velocity modeling do not show any indication for bh masses that are systematically larger than those inferred from the @xmath26 relation .
( this is a somewhat circular argument as some of the bh mass measurements from gas were used to determine the @xmath26 relation .
nevertheless , such large bh masses would imply velocity gradients which are larger than observed . )
second , the trend of an increasing @xmath61 with decreasing disk inclination is not expected if the excess dispersion is due to underestimated bh masses .
the dependence of bh mass on stellar dispersion is sometimes claimed to be steeper than used here ( e.g. , ferrarese 2002 ) .
however , this amounts to at most @xmath59 larger bh masses and such uncertainties will not affect our results on excess gas dispersions .
lastly , it has been argued that @xmath3 correlates equally well with host luminosity as with @xmath129 ( e.g. , marconi & hunt 2002 ) .
this might mean that the @xmath8 relation underestimates @xmath3 at the high @xmath129 end .
the reason is that brightest cluster galaxies ( bcgs ) fall in this region .
the bcgs are exceedingly luminous for their stellar dispersion which is typically above @xmath130 ( fisher , illingworth & franx 1995 ) .
if true , it is very unlikely that this will affect our results because the galaxies in our sample with the largest excess in stellar dispersion compared to gravitational models all have stellar dispersions between @xmath131 and @xmath130 ( cf . ,
figure [ f : magsigma ] ) .
* are thin disk models plausible ?
* it was found that for most of the normal galaxies in our sample the nuclear gas dispersions are reasonably well fit by thin disk models containing bhs that follow the @xmath26 relation .
but the full emission lines shapes contain additional information on the line of sight velocity distribution ( losvd ) and can further constrain the plausibility of the models .
we used the thin disk models to calculate the full losvd for the normal galaxies in the sample .
inspection of the data shows a qualitative agreement between predicted and observed losvds . a general property of the losvd of thin disk models is a tendency for being double - peaked . in some cases the predicted double peakedness seems slightly too pronounced to be consistent with the data .
but it is possible that the predicted peaks have in reality been broadened away by additional velocity dispersion contributions .
for example , the gas could be collisionless but orbit in a somewhat more three - dimensional distribution than a disk ( as in the models of section [ s : spheroid ] ) , or it could have a small additional non - gravitational contribution ( as in the models of section [ s : nongrav ] ) .
overall , the signal - to - noise ratio of the data and the blending of the h + [ nii ] lines make it difficult to draw strong conclusions from the observed line shapes .
* could an unaccounted broad h@xmath12 component in radio galaxies influence our results ? * the flux from which the velocity dispersions are measured is usually dominated by the flux from the forbidden [ nii ] and [ sii ] lines , especially the [ nii]@xmath14 line and not the h@xmath12 line .
furthermore , we ensured to correct for the presence of broad h@xmath12 if it was clearly detected .
this broad component is likely arising from a broad - line region which is located much closer to the nucleus than the narrow component that we model here .
noel - storr et al . ( 2003 ) found tentative indications that a broad line , not readily seen by eye , could be present in eight of the radio galaxies in our sample ( table [ t : sample ] ) . in these cases they performed fits to the h@xmath12+[nii],[sii ] complex assuming the permitted h@xmath12 line has both a narrow and a broad component .
we redid the thin disk modeling for these eight galaxies taking into account the tentative broad component in measuring the velocity dispersion and emission - line profile of the narrow component .
this resulted in decreases of @xmath61 by at most @xmath132 and a median decrease of @xmath133 .
thus these eight galaxies do not change the overall results obtained together with the 13 other radio galaxies .
lastly , the dispersion in the extended disk are in many radio galaxies above those expected from instrumental effects and differential rotation over the aperture ( e.g. , noel - storr et al . 2003 ) .
this strengthens the idea that the excess dispersions are not due to a nuclear broad line region . *
gaussian fit versus true second velocity moment .
* for the thin disk models we compared the predicted second moment of a gaussian fit to the full line of sight velocity distribution ( losvd ) to the observed gaussian fit .
by contrast , in the analysis of more spheroidal distributions , @xmath102 is the ratio of the true second moments .
thus we assume that the ratio of the true second moments and the ratio of the gaussian - fit moments do not differ significantly .
it is not straightforward to test this assumption quantitatively , since the line profiles are not straightforward to calculate for the spheroidal models .
however , we believe that the errors thus introduced are unlikely to exceed several tens of percent ( van der marel & franx 1993 ) .
this is insufficient to affect any of our conclusions .
also , it seems unlikely that a difference in these ratios would affect our main result that there is a systematic difference in nuclear dispersions between radio and non - radio galaxies .
* dependence of the stellar dispersion on inclination .
* in our models we have assumed that the bh mass is uniquely determined by the stellar velocity dispersion .
however , this can only be approximately true in reality .
there must be some intrincic scatter in the @xmath8 relation , if only because the observed velocity dispersion of a galaxy generally depends on the inclination under which the galaxy is viewed , whereas the bh mass does not .
this could be important in the context of our results .
the four galaxies for which thin disks models fit the observed gas dispersions most poorly have below - average inclinations , in addition to being radio galaxies ( see figure [ f : mbhdisk ] ) .
it is therefore important to study the relation between between stellar velocity dispersion and inclination in some more detail . to support its shape ,
an oblate stellar system must on average have more pressure ( @xmath134 ) parallel to the equatorial plane ( @xmath135 / 2 $ ] ) than perpendicular to it ( @xmath136 ) .
therefore , it will generally have a lower line - of - sight velocity dispersion when viewed face - on than when viewed edge - on . the tensor virial theorem ( binney & tremaine 1986 ) gives the ratio of the pressures when integrated over the entire galaxy as @xmath137 where @xmath138 is the intrinsic axial ratio .
for example , this gives @xmath139 for a spherical galaxy and @xmath140 for a galaxy with @xmath141 . if one were to observe a galaxy with an aperture of infinite size , and if the galaxy had @xmath142 at all positions in the galaxy , then the ratio of the measured dispersion when viewed edge - on and face - on , respectively , would be equal to @xmath143 .
however , these requirements are not generally met for real galaxies and real observations .
there is no reason why galaxies should have @xmath144 , and observational constraints on this ratio remain scarce ( e.g. , gebhardt et al .
also , velocity dispersion measurements are generally restricted to the central region of the galaxy and/or a specific axis ( most often the major axis ) and do not integrate over the entire galaxy .
so the true dependence of the observed stellar velocity dispersion on the inclination angle can only be addressed with detailed three - integral stellar dynamical models ( e.g. , gebhardt et al . 2003 ; cappellari et al .
2005b ) , which are outside the scope of the present work .
nonetheless , the tensor virial theorem does provide useful order - of - magnitude guidance .
almost all elliptical galaxies have axial ratios @xmath145 ( franx , illingworth & de zeeuw 1991 ; tremblay & merritt 1995 ) .
therefore , one might expect that inclination effects cause a variation of order @xmath146% ( @xmath147 dex ) in the observed velocity dispersion of galaxies of otherwise identical properties .
an alternative way to constrain the inclination dependence of the velocity dispersion is through the scatter in several well - know relations .
the velocity dispersions correlates strongly with galaxy luminosity ( the faber - jackson relation ; e.g. , dressler et al .
1987 ) , with bh mass ( the @xmath8 relation ) , and with galaxy mass - to - light ratio @xmath148 ( cappellari et al .
none of the latter quantities depends on the viewing angle of the observations , while the velocity dispersion does .
therefore , the observed scatter in these relations sets an upper limit to the rms variation that the inclination might induce in the observed velocity dispersion .
the limits are @xmath149 dex from the faber - jackson relation ( dressler et al .
1987 ) , @xmath150 dex from the @xmath8 relation ( tremaine et al .
2002 ) , and @xmath151 dex from the relation with @xmath148 ( cappellari et al .
these results are all in reasonable agreement with the expectation from the virial theorem ( despite the fact that each of these relations is likely to have other sources of intrinsic scatter as well ) .
lastly , direct estimates of the dispersion anisotropy in giant ellipticals from detailed three - integral stellar dynamical models indicate a change of @xmath152 dex in stellar dispersion for face - on and edge - on viewing angles ( cappellari et al . 2005a ) .
given that rotational velocities contribute typically less than @xmath153 to the second moment of giant ellipticals this lends further support for the estimates based on global relations for the dependence of the stellar dispersion on inclination . to obtain more specfic information for the actual galaxies in our sample we plot their faber - jackson relation in figure [
f : magsigma ] . the four galaxies with dust / gas disk inclinations @xmath62 are indicated with special symbols .
if the disks reside in the equatorial plane of the galaxies , then the disk inclinations are identical to the galaxy inclinations . one
would then expect the galaxies with @xmath62 to have relatively low stellar velocity dispersions for their luminosity .
this is exactly what the figure shows .
the galaxies with @xmath62 seem to lie @xmath154% ( @xmath155 dex ) below the average relation between galaxy magnitude and stellar dispersion .
this agrees to lowest order with the expectation from the tensor virial theorem and the scatter in well - known correlations .
so there is theoretical and observational support for the idea that the observed stellar velocity dispersion depends on the galaxy inclination angle . by using the @xmath8 relation to estimate the bh mass
one is therefore likely to underestimate the bh mass of galaxies with relatively low inclinations .
for the galaxies in our sample with @xmath62 we may have underestimated the bh mass by a factor @xmath156 .
the gas dispersions predicted by our thin disk models will then be too low by a factor of @xmath157 . applying this correction to the predicted gas dispersions for the @xmath62 galaxies in e.g. , figure [ f : mbhdisk ]
somewhat reduces the discrepancy between the observed and predicted gas dispersions for these galaxies . nonetheless , for all four of the galaxies the observed gas dispersions remain well in execss over those predicted by the thin disk models .
so our conclusion that radio galaxies have a contribution from non - gravitational motions in their central gas dynamics remains unaffected .
we have analyzed hst / stis observations of a sample of 27 galaxies .
the nuclear velocity dispersion of the gas in a stis aperture of @xmath158@xmath1 ( i.e. , scales of tens of parsecs ) generally exceeds the large - scale stellar velocity dispersion of the galaxy .
this is qualitatively consistent with the presence of central bhs , but raises the questions whether the excess gas dispersion is of gravitational or non - gravitational origin and whether the implied bh masses are consistent with our current understanding of bh demography . to address these issues
we have constructed purely gravitational axisymmetric dynamical models for the gas , both thin disk models and models with more general axis ratios and velocity anisotropies .
this has yielded the following conclusions : * for the normal galaxies in the sample ( i.e. , without large - scale radio - jets ) the nuclear gas dispersions are adequately reproduced by models that have purely gravitational motion and bhs that follow the @xmath8 relation . among the purely gravitational models we can not generally discriminate between thin disk models and vertically extended models .
the former might seem preferred theoretically because they represent a longer - term equilibrium configuration for the gas .
however , in some cases it is unclear if the observed line profiles are consistent with the double - peaked structure that is generally predicted by thin disk models .
* the nuclear gas dispersions observed for the radio galaxies generally exceed those predicted by models with only gravitational motions in either a thin disk or a more spheroidal gas distribution .
we attribute this to the presence of non - gravitational motions in the gas that are similar to or larger than the gravitational motions .
the non - gravitational dispersion is consistent with being either isotropic or perpendicular to the extended gas and dust disks .
the non - gravitational motions are presumably driven by the active galactic nucleus ( agn ) , but we do not find a relation between the radiative output of the agn and the non - gravitational dispersion . *
given the uncertainties about the dynamical state of the gas , it is not possible to uniquely determine the bh mass for each galaxy from its nuclear gas dispersion .
however , for the sample as a whole the observed dispersions do not provide evidence for significant deviations from the @xmath8 relation in either active or non - active elliptical galaxies . in no case is the observed nuclear gas dispersion so low that it puts an upper limit on the bh mass that is significantly below the @xmath8 relation .
for the normal , i.e. , non - radio galaxies , one should note that the success of purely gravitational models does not imply that there are no alternative models that can fit the data .
for example , it can not be ruled out that in reality there is a non - gravitational component to the gas motions , and that the bh masses in these galaxies are below the @xmath2 relation .
some models in the literature , such as those published for ic 1459 ( cappellari et al .
2002 ) fall in this category . nevertheless ,
if one is willing to assume _ a priori _ that the bh masses of all elliptical galaxies follow the @xmath8 relation , then one can turn the argument around and conclude that normal elliptical galaxies do not have a very significant non - gravitational component in their gas motions .
the absence of excess dispersion in non - radio galaxies is corroborated by the fact that the bh mass estimates for this sample by itself yield a relation in good agreement with the @xmath8 relation ( e.g. , figure [ f : mbhdisk ] ) .
it is interesting to note in this context that a similarly good correspondence between bh masses from gas dispersions and from the @xmath8 relation was found for non - active early - type _ spiral _ galaxies by sarzi et al .
( 2002 ) . for the low - luminosity radio galaxies in our sample
there is also other evidence , besides the dynamical evidence presented here , for non - flat gas distributions with turbulent motion .
the core emission - line luminosity of these galaxies correlates with the radio and optical core luminosity which are both most likely due to synchrotron emission from the jet ( e.g. , chiaberge et al .
1999 ; verdoes kleijn et al . 2002 ) .
this correlation could mean that the emission - line luminosity is driven by jet photo - ionization .
it then implies covering factors of @xmath54 and hence thick disks ( capetti 2005 ) .
alternatively , the correlation could indicate that the gas is excited by shocks induced by jet - gas interactions .
although it is mostly the radio galaxies that show significant excess dispersion over a purely gravitational model , there is one normal galaxy for which this is the case too : ic 989 .
it should be noted though that for this particular galaxy the error of @xmath159 on the stellar dispersion @xmath160 is larger than typical for the sample .
the predicted gas dispersion can be increased by a factor @xmath161 by simply varying @xmath6 within its 1-@xmath129 confidence range .
also , the predicted gas dispersion could go up by up to a factor @xmath39 if the unknown inclination of the purported gas disk is closer to edge - on than the default value of @xmath162 that we assumed .
so the evidence for non - gravitational gas motions in the normal galaxy ic 989 is not strong .
finally , we note that it is the sample of relatively close to face - on ( @xmath62 ) gas disks which reveal most clearly an excess dispersion compared to a purely gravitational model .
unfortunately , this face - on group only contains radio galaxies .
however , a statistically significant difference between the excess dispersion in radio and non - radio galaxies remains when one excludes galaxies with @xmath62 gas disks from the analysis .
nevertheless , a good test of the dichotomy found in this paper will be the modeling of gas disks at lower inclinations also in non - radio galaxies .
such data is not available currently .
the evidence for non - gravitational gas motions in radio galaxies makes the gas dispersion unsuited for dynamical mass estimates in this class of galaxies .
it would be interesting to establish more generally whether dynamical mass estimates from gas velocities , including rotation curves , are robust .
this would require comparisons of bh masses from independent indicators ( e.g. , stellar and gas kinematics ) within the same galaxy .
such a comparison has been performed for a radio galaxy ( ngc 4335 , verdoes kleijn et al .
2002 ) and a non - radio galaxy ( cappellari et al . 2002 ) .
unfortunately , in both cases the stellar dynamical evidence is too inconclusive to reach firm conclusions .
but it does appear in both cases that the bh mass implied by thin disk models for the gas rotation curves is lower than would otherwise have been expected , either from stellar kinematics or from the @xmath8 relation .
this suggests that the gas might be moving slower than the circular velocity , as would be expected for example if there were asymmetric drift in the gas .
this occurs for gas distributions that are dynamically hotter than a thin disk .
more independent bh mass evaluations from gas and other tracers , both for normal and active galaxies , would be quite valuable to shed more light on these issues .
support for this work was provided by a grant associated with program go-9163 from the space telescope science institute , which is operated by aura , inc .
, under nasa contract nas5 - 26555 .
this paper made use of the leda database ( see http://leda.univ-lyon1.fr ) .
we thank the anonymous referees for their constructive comments .
we thank the first referee in particular for drawing our attention to the dependence of stellar velocity dispersion on galaxy inclination ( see the discussion in section [ s : cave ] ) .
llrrrrrrrl ic 989 & e & & 101 & -20.8 & 176 & 174@xmath164 : & ... & 22.74 & r3 + ic 1459 & e & & 29 & -20.5 & 340@xmath165 & 532 & 1.3 & 23.00 & b1,r4 + ngc 315 & e & 1 & 68 & -22.4 & 328 & 528@xmath25 & ... & 24.01 & r1 + ngc 383 & e / so & 1 & 65 & -22.0 & 267 & 924@xmath25 & ... & 24.42 & r1 + ngc 541 & e / so & 1 & 73 & -21.5 & 209 & 336 & ... & 23.85 & r1 + ngc 741 & e & 1 & 70 & -22.5 & 294 & 424 & ... & 23.76 & r1 + ngc 2329 & e / so & 1 & 77 & -21.7 & 247 & 301 & ... & 23.67 & r1 + ngc 3078 & e & & 35 & -20.9 & 249 & 262@xmath164 & ... & 23.01 & r5 + ngc 3245 & so & & 21 & -20.0 & 205@xmath165 & 168@xmath166 & 2.1 & 20.49 & b2 , r6 + ngc 3862 & e & 1 & 84 & -21.5 & 265 & 302 & ... & 24.66 & r1 + ngc 3998 & so & & 14 & -19.8 & 318 & 650@xmath167 & ... & 21.47 & r7 + ngc 4278 & e & & 16 & -19.3 & 260 & 372@xmath167 & ... & 22.08 & r6 + ngc 4335 & e & 1 & 62 & -20.6 & 282@xmath168 & 248@xmath25 & @xmath169 & 23.02 & b3,r1 + ngc 4374 & e & 1 & 15 & -20.9 & 295 & 883 & 3.6 & 23.26 & b4,r1 + ngc 4459 & so & & 17 & -20.2 & 186@xmath165 & 193 & 0.7 & 20.01 & r3 + ngc 4486 & e & 1 & 15 & -22.0 & 375@xmath165 & 681:@xmath25 & 29.0 & 24.81 & b6,r1 +
ngc 4526 & so & & 36 & -19.4 & 260 & 424@xmath164 & ... & 21.36 & r3 + ngc 5077 & e & & 36 & -20.8 & 275 & 417@xmath164 & ... & 23.08 & r5 + ngc 5127 & e & 1 & 64 & -20.9 & 189 & 157 & ... & 23.99 & r1 + ngc 5490 & e & 1 & 69 & -21.4 & 292 & 280@xmath25 & ... & 23.70 & r1 + ngc 6251 & e & 1 & 99 & -21.8 & 290@xmath165 & 434@xmath166 : & 7.8 & 24.49 & b7,r7 + ngc 6861 & e / so & & 28 & -21.1 & 382@xmath170 & 814@xmath164 & ... & 20.98 & r8 + ngc 7052 & e & 1 & 55 & -21.0 & 266@xmath165 & 456 & 3.1 & 22.95 & b8,r1 + ngc 7626 & e & 1 & 47 & -21.6 & 276 & 349@xmath25 & ... & 23.28 & r1 + ugc 1841 & e & 1 & 85 & -21.7 & 348 & 592@xmath25 & ... & 24.85 & r1 + ugc 7115 & e & 1 & 91 & -20.9 & 198 & 545 : & ... & 23.85 & r1 + ugc 12064 & e /
so & 1 & 68 & -20.6 & 255 & 496@xmath25 & ... & 24.29 & r1 + lrrlllll ic 989 & 7354 & [email protected] & 3.3@xmath172 & 1.6@xmath173 & 2.2@xmath174 & 60@xmath175 & + ic 1459 & 7352 & [email protected] & 9.0@xmath176 & 2.7@xmath174 & 3.6@xmath172 & 60 & 1 + ngc 315 & 8236 & [email protected] & 1.7@xmath172 & 2.0@xmath174 & 4.7@xmath176 & 77 & 5 + ngc 383 & 8236 & [email protected] & 2.0@xmath172 & 2.1@xmath172 & 3.8@xmath172 & 40 & 5 + ngc 541 & 8236 & [email protected] & 1.4@xmath172 & 5.3@xmath174 & 1.9@xmath176 & 25 & 5 + ngc 741 & 8236 & [email protected] & 4.3@xmath172 & 4.4@xmath176 & 4.4@xmath172 & 60@xmath175 & 5 + ngc 2329 & 8236 & [email protected] & 1.2@xmath172 & 8.9@xmath176 & 3.2@xmath172 & 47 & 5 + ngc 3078 & 9163 & [email protected] & 2.0@xmath172 & 3.2@xmath174 & 4.0@xmath177 & 63 & + ngc 3245 & 7403 & [email protected] & 2.0@xmath174 & 6.4@xmath172 & 1.6@xmath174 & 63 & 2 + ngc 3862 & 8236 & [email protected] & 1.1@xmath172 & 2.0@xmath177 & 3.7@xmath172 & 08 & 5 + ngc 3998 & 7354 & [email protected] & 1.6@xmath172 & 6.2@xmath174 & 8.2@xmath176 & 60@xmath175 & + ngc 4278 & 7403 & [email protected] & 2.6@xmath172 & 3.8@xmath174 & 1.7@xmath172 & 42 & + ngc 4335 & 8236 & [email protected] & 5.7@xmath172 & 5.8@xmath172 & 4.4@xmath172 & 66 & 5 + ngc 4374 & 7124 & [email protected] & 2.1@xmath172 & 7.3@xmath174 & 8.7@xmath176 & 60@xmath175 & 5 + ngc 4459 & 7361 & [email protected] & 2.2@xmath172 & 9.2@xmath172 & 3.4@xmath172 & 47 & 3 + ngc 4486 & 8666 & [email protected] & 6.7@xmath172 & 2.2@xmath176 & 4.6@xmath172 & 52 & 5 + ngc 4526 & 9163 & [email protected] & 3.3@xmath172 & 1.1@xmath172 & 3.6@xmath172 & 80 & + ngc 5077 & 7354 & [email protected] & 3.2@xmath172 & 3.1@xmath172 & 3.9@xmath172 & 60@xmath175 & + ngc 5127 & 8236 & [email protected] & 6.4@xmath172 & 9.4@xmath172 & 5.3@xmath172 & 60@xmath175 & 5 + ngc 5490 & 8236 & [email protected] & 5.0@xmath172 & 1.4@xmath176 & 4.2@xmath172 & 60@xmath175 & 5 + ngc 6251 & 6653 & [email protected] & 1.0@xmath172 & 4.1@xmath172 & 3.1@xmath172 & 32 & 4 + ngc 6861 & 9163 & [email protected] & 2.9@xmath172 & 1.0@xmath177 & 3.8@xmath172 & 74 & + ngc 7052 & 8236 & [email protected] & 2.3@xmath172 & 1.7@xmath174 & 2.9@xmath172 & 73 & 5 + ngc 7626 & 8236 & [email protected] & 2.3@xmath172 & 7.3@xmath172 & 7.0@xmath174 & 60@xmath175 & 5 + ugc 1841 & 8236 & [email protected] & 1.9@xmath172 & 1.5@xmath172 & 4.0@xmath172 & 60@xmath175 & 5 + ugc 7115 & 8236 & [email protected] & 1.8@xmath172 & 7.0@xmath172 & 4.8@xmath172 & 18 & 5 + ugc 12064 & 8236 & [email protected] & 2.3@xmath172 & 2.6@xmath176 & 3.9@xmath172 & 57 & 5 + | the majority of nearby early - type galaxies contains detectable amounts of emission - line gas at their centers .
the nuclear gas kinematics form a valuable diagnostic of the central black hole ( bh ) mass . here
we analyze and model hst / stis observations of a sample of 27 galaxies ; 16 fanaroff & riley type i radio galaxies and 11 ( more ) normal early - type galaxies .
we focus here on what can be learned from the nuclear velocity dispersion ( line width ) of the gas as a complement to the many studies dealing with gas rotation velocities .
we find that the dispersion in a stis aperture of @xmath0@xmath1 generally exceeds the large - scale stellar velocity dispersion of the galaxy .
this is qualitatively consistent with the presence of central bhs , but raises the questions whether the excess gas dispersion is of gravitational or non - gravitational origin and whether the implied bh masses are consistent with our current understanding of bh demography ( as predicted by the @xmath2 relation between bh mass and stellar velocity dispersion ) . to address this
we construct purely gravitational axisymmetric dynamical models for the gas , both thin disk models and models with more general axis ratios and velocity anisotropies . for the normal galaxies the nuclear gas dispersions
are adequately reproduced assuming disks around bhs with masses that follow the @xmath2 relation .
in contrast , the gas dispersions observed for the radio galaxies generally exceed those predicted by any of the models . we attribute this to the presence of non - gravitational motions in the gas that are similar to or larger than the gravitational motions .
the non - gravitational motions are presumably driven by the active galactic nucleus ( agn ) , but we do not find a relation between the radiative output of the agn and the non - gravitational dispersion . given the uncertainties about the dynamical state of the gas , it is not possible to uniquely determine the bh mass for each galaxy from its nuclear gas dispersion . however , for the sample as a whole the observed dispersions do not provide evidence for significant deviations from the @xmath2 relation . |
Subsets and Splits